git-svn.perlon commit submodule: additional regression tests for relative URLs (712693e)
   1#!/usr/bin/perl
   2# Copyright (C) 2006, Eric Wong <normalperson@yhbt.net>
   3# License: GPL v2 or later
   4use 5.008;
   5use warnings;
   6use strict;
   7use vars qw/    $AUTHOR $VERSION
   8                $sha1 $sha1_short $_revision $_repository
   9                $_q $_authors $_authors_prog %users/;
  10$AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
  11$VERSION = '@@GIT_VERSION@@';
  12
  13# From which subdir have we been invoked?
  14my $cmd_dir_prefix = eval {
  15        command_oneline([qw/rev-parse --show-prefix/], STDERR => 0)
  16} || '';
  17
  18my $git_dir_user_set = 1 if defined $ENV{GIT_DIR};
  19$ENV{GIT_DIR} ||= '.git';
  20$Git::SVN::default_repo_id = 'svn';
  21$Git::SVN::default_ref_id = $ENV{GIT_SVN_ID} || 'git-svn';
  22$Git::SVN::Ra::_log_window_size = 100;
  23$Git::SVN::_minimize_url = 'unset';
  24
  25if (! exists $ENV{SVN_SSH} && exists $ENV{GIT_SSH}) {
  26        $ENV{SVN_SSH} = $ENV{GIT_SSH};
  27}
  28
  29if (exists $ENV{SVN_SSH} && $^O eq 'msys') {
  30        $ENV{SVN_SSH} =~ s/\\/\\\\/g;
  31        $ENV{SVN_SSH} =~ s/(.*)/"$1"/;
  32}
  33
  34$Git::SVN::Log::TZ = $ENV{TZ};
  35$ENV{TZ} = 'UTC';
  36$| = 1; # unbuffer STDOUT
  37
  38sub fatal (@) { print STDERR "@_\n"; exit 1 }
  39
  40# All SVN commands do it.  Otherwise we may die on SIGPIPE when the remote
  41# repository decides to close the connection which we expect to be kept alive.
  42$SIG{PIPE} = 'IGNORE';
  43
  44# Given a dot separated version number, "subtract" it from
  45# the SVN::Core::VERSION; non-negaitive return means the SVN::Core
  46# is at least at the version the caller asked for.
  47sub compare_svn_version {
  48        my (@ours) = split(/\./, $SVN::Core::VERSION);
  49        my (@theirs) = split(/\./, $_[0]);
  50        my ($i, $diff);
  51
  52        for ($i = 0; $i < @ours && $i < @theirs; $i++) {
  53                $diff = $ours[$i] - $theirs[$i];
  54                return $diff if ($diff);
  55        }
  56        return 1 if ($i < @ours);
  57        return -1 if ($i < @theirs);
  58        return 0;
  59}
  60
  61sub _req_svn {
  62        require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
  63        require SVN::Ra;
  64        require SVN::Delta;
  65        if (::compare_svn_version('1.1.0') < 0) {
  66                fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)";
  67        }
  68}
  69my $can_compress = eval { require Compress::Zlib; 1};
  70push @Git::SVN::Ra::ISA, 'SVN::Ra';
  71push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
  72push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
  73use Carp qw/croak/;
  74use Digest::MD5;
  75use IO::File qw//;
  76use File::Basename qw/dirname basename/;
  77use File::Path qw/mkpath/;
  78use File::Spec;
  79use File::Find;
  80use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev/;
  81use IPC::Open3;
  82use Git;
  83use Memoize;  # core since 5.8.0, Jul 2002
  84
  85BEGIN {
  86        # import functions from Git into our packages, en masse
  87        no strict 'refs';
  88        foreach (qw/command command_oneline command_noisy command_output_pipe
  89                    command_input_pipe command_close_pipe
  90                    command_bidi_pipe command_close_bidi_pipe/) {
  91                for my $package ( qw(SVN::Git::Editor SVN::Git::Fetcher
  92                        Git::SVN::Migration Git::SVN::Log Git::SVN),
  93                        __PACKAGE__) {
  94                        *{"${package}::$_"} = \&{"Git::$_"};
  95                }
  96        }
  97        Memoize::memoize 'Git::config';
  98        Memoize::memoize 'Git::config_bool';
  99}
 100
 101my ($SVN);
 102
 103$sha1 = qr/[a-f\d]{40}/;
 104$sha1_short = qr/[a-f\d]{4,40}/;
 105my ($_stdin, $_help, $_edit,
 106        $_message, $_file, $_branch_dest,
 107        $_template, $_shared,
 108        $_version, $_fetch_all, $_no_rebase, $_fetch_parent,
 109        $_merge, $_strategy, $_preserve_merges, $_dry_run, $_local,
 110        $_prefix, $_no_checkout, $_url, $_verbose,
 111        $_git_format, $_commit_url, $_tag, $_merge_info, $_interactive);
 112$Git::SVN::_follow_parent = 1;
 113$SVN::Git::Fetcher::_placeholder_filename = ".gitignore";
 114$_q ||= 0;
 115my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
 116                    'config-dir=s' => \$Git::SVN::Ra::config_dir,
 117                    'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache,
 118                    'ignore-paths=s' => \$SVN::Git::Fetcher::_ignore_regex,
 119                    'ignore-refs=s' => \$Git::SVN::Ra::_ignore_refs_regex );
 120my %fc_opts = ( 'follow-parent|follow!' => \$Git::SVN::_follow_parent,
 121                'authors-file|A=s' => \$_authors,
 122                'authors-prog=s' => \$_authors_prog,
 123                'repack:i' => \$Git::SVN::_repack,
 124                'noMetadata' => \$Git::SVN::_no_metadata,
 125                'useSvmProps' => \$Git::SVN::_use_svm_props,
 126                'useSvnsyncProps' => \$Git::SVN::_use_svnsync_props,
 127                'log-window-size=i' => \$Git::SVN::Ra::_log_window_size,
 128                'no-checkout' => \$_no_checkout,
 129                'quiet|q+' => \$_q,
 130                'repack-flags|repack-args|repack-opts=s' =>
 131                   \$Git::SVN::_repack_flags,
 132                'use-log-author' => \$Git::SVN::_use_log_author,
 133                'add-author-from' => \$Git::SVN::_add_author_from,
 134                'localtime' => \$Git::SVN::_localtime,
 135                %remote_opts );
 136
 137my ($_trunk, @_tags, @_branches, $_stdlayout);
 138my %icv;
 139my %init_opts = ( 'template=s' => \$_template, 'shared:s' => \$_shared,
 140                  'trunk|T=s' => \$_trunk, 'tags|t=s@' => \@_tags,
 141                  'branches|b=s@' => \@_branches, 'prefix=s' => \$_prefix,
 142                  'stdlayout|s' => \$_stdlayout,
 143                  'minimize-url|m!' => \$Git::SVN::_minimize_url,
 144                  'no-metadata' => sub { $icv{noMetadata} = 1 },
 145                  'use-svm-props' => sub { $icv{useSvmProps} = 1 },
 146                  'use-svnsync-props' => sub { $icv{useSvnsyncProps} = 1 },
 147                  'rewrite-root=s' => sub { $icv{rewriteRoot} = $_[1] },
 148                  'rewrite-uuid=s' => sub { $icv{rewriteUUID} = $_[1] },
 149                  %remote_opts );
 150my %cmt_opts = ( 'edit|e' => \$_edit,
 151                'rmdir' => \$SVN::Git::Editor::_rmdir,
 152                'find-copies-harder' => \$SVN::Git::Editor::_find_copies_harder,
 153                'l=i' => \$SVN::Git::Editor::_rename_limit,
 154                'copy-similarity|C=i'=> \$SVN::Git::Editor::_cp_similarity
 155);
 156
 157my %cmd = (
 158        fetch => [ \&cmd_fetch, "Download new revisions from SVN",
 159                        { 'revision|r=s' => \$_revision,
 160                          'fetch-all|all' => \$_fetch_all,
 161                          'parent|p' => \$_fetch_parent,
 162                           %fc_opts } ],
 163        clone => [ \&cmd_clone, "Initialize and fetch revisions",
 164                        { 'revision|r=s' => \$_revision,
 165                          'preserve-empty-dirs' =>
 166                                \$SVN::Git::Fetcher::_preserve_empty_dirs,
 167                          'placeholder-filename=s' =>
 168                                \$SVN::Git::Fetcher::_placeholder_filename,
 169                           %fc_opts, %init_opts } ],
 170        init => [ \&cmd_init, "Initialize a repo for tracking" .
 171                          " (requires URL argument)",
 172                          \%init_opts ],
 173        'multi-init' => [ \&cmd_multi_init,
 174                          "Deprecated alias for ".
 175                          "'$0 init -T<trunk> -b<branches> -t<tags>'",
 176                          \%init_opts ],
 177        dcommit => [ \&cmd_dcommit,
 178                     'Commit several diffs to merge with upstream',
 179                        { 'merge|m|M' => \$_merge,
 180                          'strategy|s=s' => \$_strategy,
 181                          'verbose|v' => \$_verbose,
 182                          'dry-run|n' => \$_dry_run,
 183                          'fetch-all|all' => \$_fetch_all,
 184                          'commit-url=s' => \$_commit_url,
 185                          'revision|r=i' => \$_revision,
 186                          'no-rebase' => \$_no_rebase,
 187                          'mergeinfo=s' => \$_merge_info,
 188                          'interactive|i' => \$_interactive,
 189                        %cmt_opts, %fc_opts } ],
 190        branch => [ \&cmd_branch,
 191                    'Create a branch in the SVN repository',
 192                    { 'message|m=s' => \$_message,
 193                      'destination|d=s' => \$_branch_dest,
 194                      'dry-run|n' => \$_dry_run,
 195                      'tag|t' => \$_tag,
 196                      'username=s' => \$Git::SVN::Prompt::_username,
 197                      'commit-url=s' => \$_commit_url } ],
 198        tag => [ sub { $_tag = 1; cmd_branch(@_) },
 199                 'Create a tag in the SVN repository',
 200                 { 'message|m=s' => \$_message,
 201                   'destination|d=s' => \$_branch_dest,
 202                   'dry-run|n' => \$_dry_run,
 203                   'username=s' => \$Git::SVN::Prompt::_username,
 204                   'commit-url=s' => \$_commit_url } ],
 205        'set-tree' => [ \&cmd_set_tree,
 206                        "Set an SVN repository to a git tree-ish",
 207                        { 'stdin' => \$_stdin, %cmt_opts, %fc_opts, } ],
 208        'create-ignore' => [ \&cmd_create_ignore,
 209                             'Create a .gitignore per svn:ignore',
 210                             { 'revision|r=i' => \$_revision
 211                             } ],
 212        'mkdirs' => [ \&cmd_mkdirs ,
 213                      "recreate empty directories after a checkout",
 214                      { 'revision|r=i' => \$_revision } ],
 215        'propget' => [ \&cmd_propget,
 216                       'Print the value of a property on a file or directory',
 217                       { 'revision|r=i' => \$_revision } ],
 218        'proplist' => [ \&cmd_proplist,
 219                       'List all properties of a file or directory',
 220                       { 'revision|r=i' => \$_revision } ],
 221        'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
 222                        { 'revision|r=i' => \$_revision
 223                        } ],
 224        'show-externals' => [ \&cmd_show_externals, "Show svn:externals listings",
 225                        { 'revision|r=i' => \$_revision
 226                        } ],
 227        'multi-fetch' => [ \&cmd_multi_fetch,
 228                           "Deprecated alias for $0 fetch --all",
 229                           { 'revision|r=s' => \$_revision, %fc_opts } ],
 230        'migrate' => [ sub { },
 231                       # no-op, we automatically run this anyways,
 232                       'Migrate configuration/metadata/layout from
 233                        previous versions of git-svn',
 234                       { 'minimize' => \$Git::SVN::Migration::_minimize,
 235                         %remote_opts } ],
 236        'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
 237                        { 'limit=i' => \$Git::SVN::Log::limit,
 238                          'revision|r=s' => \$_revision,
 239                          'verbose|v' => \$Git::SVN::Log::verbose,
 240                          'incremental' => \$Git::SVN::Log::incremental,
 241                          'oneline' => \$Git::SVN::Log::oneline,
 242                          'show-commit' => \$Git::SVN::Log::show_commit,
 243                          'non-recursive' => \$Git::SVN::Log::non_recursive,
 244                          'authors-file|A=s' => \$_authors,
 245                          'color' => \$Git::SVN::Log::color,
 246                          'pager=s' => \$Git::SVN::Log::pager
 247                        } ],
 248        'find-rev' => [ \&cmd_find_rev,
 249                        "Translate between SVN revision numbers and tree-ish",
 250                        {} ],
 251        'rebase' => [ \&cmd_rebase, "Fetch and rebase your working directory",
 252                        { 'merge|m|M' => \$_merge,
 253                          'verbose|v' => \$_verbose,
 254                          'strategy|s=s' => \$_strategy,
 255                          'local|l' => \$_local,
 256                          'fetch-all|all' => \$_fetch_all,
 257                          'dry-run|n' => \$_dry_run,
 258                          'preserve-merges|p' => \$_preserve_merges,
 259                          %fc_opts } ],
 260        'commit-diff' => [ \&cmd_commit_diff,
 261                           'Commit a diff between two trees',
 262                        { 'message|m=s' => \$_message,
 263                          'file|F=s' => \$_file,
 264                          'revision|r=s' => \$_revision,
 265                        %cmt_opts } ],
 266        'info' => [ \&cmd_info,
 267                    "Show info about the latest SVN revision
 268                     on the current branch",
 269                    { 'url' => \$_url, } ],
 270        'blame' => [ \&Git::SVN::Log::cmd_blame,
 271                    "Show what revision and author last modified each line of a file",
 272                    { 'git-format' => \$_git_format } ],
 273        'reset' => [ \&cmd_reset,
 274                     "Undo fetches back to the specified SVN revision",
 275                     { 'revision|r=s' => \$_revision,
 276                       'parent|p' => \$_fetch_parent } ],
 277        'gc' => [ \&cmd_gc,
 278                  "Compress unhandled.log files in .git/svn and remove " .
 279                  "index files in .git/svn",
 280                {} ],
 281);
 282
 283use Term::ReadLine;
 284package FakeTerm;
 285sub new {
 286        my ($class, $reason) = @_;
 287        return bless \$reason, shift;
 288}
 289sub readline {
 290        my $self = shift;
 291        die "Cannot use readline on FakeTerm: $$self";
 292}
 293package main;
 294
 295my $term = eval {
 296        $ENV{"GIT_SVN_NOTTY"}
 297                ? new Term::ReadLine 'git-svn', \*STDIN, \*STDOUT
 298                : new Term::ReadLine 'git-svn';
 299};
 300if ($@) {
 301        $term = new FakeTerm "$@: going non-interactive";
 302}
 303
 304my $cmd;
 305for (my $i = 0; $i < @ARGV; $i++) {
 306        if (defined $cmd{$ARGV[$i]}) {
 307                $cmd = $ARGV[$i];
 308                splice @ARGV, $i, 1;
 309                last;
 310        } elsif ($ARGV[$i] eq 'help') {
 311                $cmd = $ARGV[$i+1];
 312                usage(0);
 313        }
 314};
 315
 316# make sure we're always running at the top-level working directory
 317unless ($cmd && $cmd =~ /(?:clone|init|multi-init)$/) {
 318        unless (-d $ENV{GIT_DIR}) {
 319                if ($git_dir_user_set) {
 320                        die "GIT_DIR=$ENV{GIT_DIR} explicitly set, ",
 321                            "but it is not a directory\n";
 322                }
 323                my $git_dir = delete $ENV{GIT_DIR};
 324                my $cdup = undef;
 325                git_cmd_try {
 326                        $cdup = command_oneline(qw/rev-parse --show-cdup/);
 327                        $git_dir = '.' unless ($cdup);
 328                        chomp $cdup if ($cdup);
 329                        $cdup = "." unless ($cdup && length $cdup);
 330                } "Already at toplevel, but $git_dir not found\n";
 331                chdir $cdup or die "Unable to chdir up to '$cdup'\n";
 332                unless (-d $git_dir) {
 333                        die "$git_dir still not found after going to ",
 334                            "'$cdup'\n";
 335                }
 336                $ENV{GIT_DIR} = $git_dir;
 337        }
 338        $_repository = Git->repository(Repository => $ENV{GIT_DIR});
 339}
 340
 341my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
 342
 343read_git_config(\%opts);
 344if ($cmd && ($cmd eq 'log' || $cmd eq 'blame')) {
 345        Getopt::Long::Configure('pass_through');
 346}
 347my $rv = GetOptions(%opts, 'h|H' => \$_help, 'version|V' => \$_version,
 348                    'minimize-connections' => \$Git::SVN::Migration::_minimize,
 349                    'id|i=s' => \$Git::SVN::default_ref_id,
 350                    'svn-remote|remote|R=s' => sub {
 351                       $Git::SVN::no_reuse_existing = 1;
 352                       $Git::SVN::default_repo_id = $_[1] });
 353exit 1 if (!$rv && $cmd && $cmd ne 'log');
 354
 355usage(0) if $_help;
 356version() if $_version;
 357usage(1) unless defined $cmd;
 358load_authors() if $_authors;
 359if (defined $_authors_prog) {
 360        $_authors_prog = "'" . File::Spec->rel2abs($_authors_prog) . "'";
 361}
 362
 363unless ($cmd =~ /^(?:clone|init|multi-init|commit-diff)$/) {
 364        Git::SVN::Migration::migration_check();
 365}
 366Git::SVN::init_vars();
 367eval {
 368        Git::SVN::verify_remotes_sanity();
 369        $cmd{$cmd}->[0]->(@ARGV);
 370};
 371fatal $@ if $@;
 372post_fetch_checkout();
 373exit 0;
 374
 375####################### primary functions ######################
 376sub usage {
 377        my $exit = shift || 0;
 378        my $fd = $exit ? \*STDERR : \*STDOUT;
 379        print $fd <<"";
 380git-svn - bidirectional operations between a single Subversion tree and git
 381Usage: git svn <command> [options] [arguments]\n
 382
 383        print $fd "Available commands:\n" unless $cmd;
 384
 385        foreach (sort keys %cmd) {
 386                next if $cmd && $cmd ne $_;
 387                next if /^multi-/; # don't show deprecated commands
 388                print $fd '  ',pack('A17',$_),$cmd{$_}->[1],"\n";
 389                foreach (sort keys %{$cmd{$_}->[2]}) {
 390                        # mixed-case options are for .git/config only
 391                        next if /[A-Z]/ && /^[a-z]+$/i;
 392                        # prints out arguments as they should be passed:
 393                        my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
 394                        print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
 395                                                        "--$_" : "-$_" }
 396                                                split /\|/,$_)," $x\n";
 397                }
 398        }
 399        print $fd <<"";
 400\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
 401arbitrary identifier if you're tracking multiple SVN branches/repositories in
 402one git repository and want to keep them separate.  See git-svn(1) for more
 403information.
 404
 405        exit $exit;
 406}
 407
 408sub version {
 409        ::_req_svn();
 410        print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
 411        exit 0;
 412}
 413
 414sub ask {
 415        my ($prompt, %arg) = @_;
 416        my $valid_re = $arg{valid_re};
 417        my $default = $arg{default};
 418        my $resp;
 419        my $i = 0;
 420
 421        if ( !( defined($term->IN)
 422            && defined( fileno($term->IN) )
 423            && defined( $term->OUT )
 424            && defined( fileno($term->OUT) ) ) ){
 425                return defined($default) ? $default : undef;
 426        }
 427
 428        while ($i++ < 10) {
 429                $resp = $term->readline($prompt);
 430                if (!defined $resp) { # EOF
 431                        print "\n";
 432                        return defined $default ? $default : undef;
 433                }
 434                if ($resp eq '' and defined $default) {
 435                        return $default;
 436                }
 437                if (!defined $valid_re or $resp =~ /$valid_re/) {
 438                        return $resp;
 439                }
 440        }
 441        return undef;
 442}
 443
 444sub do_git_init_db {
 445        unless (-d $ENV{GIT_DIR}) {
 446                my @init_db = ('init');
 447                push @init_db, "--template=$_template" if defined $_template;
 448                if (defined $_shared) {
 449                        if ($_shared =~ /[a-z]/) {
 450                                push @init_db, "--shared=$_shared";
 451                        } else {
 452                                push @init_db, "--shared";
 453                        }
 454                }
 455                command_noisy(@init_db);
 456                $_repository = Git->repository(Repository => ".git");
 457        }
 458        my $set;
 459        my $pfx = "svn-remote.$Git::SVN::default_repo_id";
 460        foreach my $i (keys %icv) {
 461                die "'$set' and '$i' cannot both be set\n" if $set;
 462                next unless defined $icv{$i};
 463                command_noisy('config', "$pfx.$i", $icv{$i});
 464                $set = $i;
 465        }
 466        my $ignore_paths_regex = \$SVN::Git::Fetcher::_ignore_regex;
 467        command_noisy('config', "$pfx.ignore-paths", $$ignore_paths_regex)
 468                if defined $$ignore_paths_regex;
 469        my $ignore_refs_regex = \$Git::SVN::Ra::_ignore_refs_regex;
 470        command_noisy('config', "$pfx.ignore-refs", $$ignore_refs_regex)
 471                if defined $$ignore_refs_regex;
 472
 473        if (defined $SVN::Git::Fetcher::_preserve_empty_dirs) {
 474                my $fname = \$SVN::Git::Fetcher::_placeholder_filename;
 475                command_noisy('config', "$pfx.preserve-empty-dirs", 'true');
 476                command_noisy('config', "$pfx.placeholder-filename", $$fname);
 477        }
 478}
 479
 480sub init_subdir {
 481        my $repo_path = shift or return;
 482        mkpath([$repo_path]) unless -d $repo_path;
 483        chdir $repo_path or die "Couldn't chdir to $repo_path: $!\n";
 484        $ENV{GIT_DIR} = '.git';
 485        $_repository = Git->repository(Repository => $ENV{GIT_DIR});
 486}
 487
 488sub cmd_clone {
 489        my ($url, $path) = @_;
 490        if (!defined $path &&
 491            (defined $_trunk || @_branches || @_tags ||
 492             defined $_stdlayout) &&
 493            $url !~ m#^[a-z\+]+://#) {
 494                $path = $url;
 495        }
 496        $path = basename($url) if !defined $path || !length $path;
 497        my $authors_absolute = $_authors ? File::Spec->rel2abs($_authors) : "";
 498        cmd_init($url, $path);
 499        command_oneline('config', 'svn.authorsfile', $authors_absolute)
 500            if $_authors;
 501        Git::SVN::fetch_all($Git::SVN::default_repo_id);
 502}
 503
 504sub cmd_init {
 505        if (defined $_stdlayout) {
 506                $_trunk = 'trunk' if (!defined $_trunk);
 507                @_tags = 'tags' if (! @_tags);
 508                @_branches = 'branches' if (! @_branches);
 509        }
 510        if (defined $_trunk || @_branches || @_tags) {
 511                return cmd_multi_init(@_);
 512        }
 513        my $url = shift or die "SVN repository location required ",
 514                               "as a command-line argument\n";
 515        $url = canonicalize_url($url);
 516        init_subdir(@_);
 517        do_git_init_db();
 518
 519        if ($Git::SVN::_minimize_url eq 'unset') {
 520                $Git::SVN::_minimize_url = 0;
 521        }
 522
 523        Git::SVN->init($url);
 524}
 525
 526sub cmd_fetch {
 527        if (grep /^\d+=./, @_) {
 528                die "'<rev>=<commit>' fetch arguments are ",
 529                    "no longer supported.\n";
 530        }
 531        my ($remote) = @_;
 532        if (@_ > 1) {
 533                die "Usage: $0 fetch [--all] [--parent] [svn-remote]\n";
 534        }
 535        $Git::SVN::no_reuse_existing = undef;
 536        if ($_fetch_parent) {
 537                my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 538                unless ($gs) {
 539                        die "Unable to determine upstream SVN information from ",
 540                            "working tree history\n";
 541                }
 542                # just fetch, don't checkout.
 543                $_no_checkout = 'true';
 544                $_fetch_all ? $gs->fetch_all : $gs->fetch;
 545        } elsif ($_fetch_all) {
 546                cmd_multi_fetch();
 547        } else {
 548                $remote ||= $Git::SVN::default_repo_id;
 549                Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
 550        }
 551}
 552
 553sub cmd_set_tree {
 554        my (@commits) = @_;
 555        if ($_stdin || !@commits) {
 556                print "Reading from stdin...\n";
 557                @commits = ();
 558                while (<STDIN>) {
 559                        if (/\b($sha1_short)\b/o) {
 560                                unshift @commits, $1;
 561                        }
 562                }
 563        }
 564        my @revs;
 565        foreach my $c (@commits) {
 566                my @tmp = command('rev-parse',$c);
 567                if (scalar @tmp == 1) {
 568                        push @revs, $tmp[0];
 569                } elsif (scalar @tmp > 1) {
 570                        push @revs, reverse(command('rev-list',@tmp));
 571                } else {
 572                        fatal "Failed to rev-parse $c";
 573                }
 574        }
 575        my $gs = Git::SVN->new;
 576        my ($r_last, $cmt_last) = $gs->last_rev_commit;
 577        $gs->fetch;
 578        if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
 579                fatal "There are new revisions that were fetched ",
 580                      "and need to be merged (or acknowledged) ",
 581                      "before committing.\nlast rev: $r_last\n",
 582                      " current: $gs->{last_rev}";
 583        }
 584        $gs->set_tree($_) foreach @revs;
 585        print "Done committing ",scalar @revs," revisions to SVN\n";
 586        unlink $gs->{index};
 587}
 588
 589sub split_merge_info_range {
 590        my ($range) = @_;
 591        if ($range =~ /(\d+)-(\d+)/) {
 592                return (int($1), int($2));
 593        } else {
 594                return (int($range), int($range));
 595        }
 596}
 597
 598sub combine_ranges {
 599        my ($in) = @_;
 600
 601        my @fnums = ();
 602        my @arr = split(/,/, $in);
 603        for my $element (@arr) {
 604                my ($start, $end) = split_merge_info_range($element);
 605                push @fnums, $start;
 606        }
 607
 608        my @sorted = @arr [ sort {
 609                $fnums[$a] <=> $fnums[$b]
 610        } 0..$#arr ];
 611
 612        my @return = ();
 613        my $last = -1;
 614        my $first = -1;
 615        for my $element (@sorted) {
 616                my ($start, $end) = split_merge_info_range($element);
 617
 618                if ($last == -1) {
 619                        $first = $start;
 620                        $last = $end;
 621                        next;
 622                }
 623                if ($start <= $last+1) {
 624                        if ($end > $last) {
 625                                $last = $end;
 626                        }
 627                        next;
 628                }
 629                if ($first == $last) {
 630                        push @return, "$first";
 631                } else {
 632                        push @return, "$first-$last";
 633                }
 634                $first = $start;
 635                $last = $end;
 636        }
 637
 638        if ($first != -1) {
 639                if ($first == $last) {
 640                        push @return, "$first";
 641                } else {
 642                        push @return, "$first-$last";
 643                }
 644        }
 645
 646        return join(',', @return);
 647}
 648
 649sub merge_revs_into_hash {
 650        my ($hash, $minfo) = @_;
 651        my @lines = split(' ', $minfo);
 652
 653        for my $line (@lines) {
 654                my ($branchpath, $revs) = split(/:/, $line);
 655
 656                if (exists($hash->{$branchpath})) {
 657                        # Merge the two revision sets
 658                        my $combined = "$hash->{$branchpath},$revs";
 659                        $hash->{$branchpath} = combine_ranges($combined);
 660                } else {
 661                        # Just do range combining for consolidation
 662                        $hash->{$branchpath} = combine_ranges($revs);
 663                }
 664        }
 665}
 666
 667sub merge_merge_info {
 668        my ($mergeinfo_one, $mergeinfo_two) = @_;
 669        my %result_hash = ();
 670
 671        merge_revs_into_hash(\%result_hash, $mergeinfo_one);
 672        merge_revs_into_hash(\%result_hash, $mergeinfo_two);
 673
 674        my $result = '';
 675        # Sort below is for consistency's sake
 676        for my $branchname (sort keys(%result_hash)) {
 677                my $revlist = $result_hash{$branchname};
 678                $result .= "$branchname:$revlist\n"
 679        }
 680        return $result;
 681}
 682
 683sub populate_merge_info {
 684        my ($d, $gs, $uuid, $linear_refs, $rewritten_parent) = @_;
 685
 686        my %parentshash;
 687        read_commit_parents(\%parentshash, $d);
 688        my @parents = @{$parentshash{$d}};
 689        if ($#parents > 0) {
 690                # Merge commit
 691                my $all_parents_ok = 1;
 692                my $aggregate_mergeinfo = '';
 693                my $rooturl = $gs->repos_root;
 694
 695                if (defined($rewritten_parent)) {
 696                        # Replace first parent with newly-rewritten version
 697                        shift @parents;
 698                        unshift @parents, $rewritten_parent;
 699                }
 700
 701                foreach my $parent (@parents) {
 702                        my ($branchurl, $svnrev, $paruuid) =
 703                                cmt_metadata($parent);
 704
 705                        unless (defined($svnrev)) {
 706                                # Should have been caught be preflight check
 707                                fatal "merge commit $d has ancestor $parent, but that change "
 708                     ."does not have git-svn metadata!";
 709                        }
 710                        unless ($branchurl =~ /^\Q$rooturl\E(.*)/) {
 711                                fatal "commit $parent git-svn metadata changed mid-run!";
 712                        }
 713                        my $branchpath = $1;
 714
 715                        my $ra = Git::SVN::Ra->new($branchurl);
 716                        my (undef, undef, $props) =
 717                                $ra->get_dir(canonicalize_path("."), $svnrev);
 718                        my $par_mergeinfo = $props->{'svn:mergeinfo'};
 719                        unless (defined $par_mergeinfo) {
 720                                $par_mergeinfo = '';
 721                        }
 722                        # Merge previous mergeinfo values
 723                        $aggregate_mergeinfo =
 724                                merge_merge_info($aggregate_mergeinfo,
 725                                                                 $par_mergeinfo, 0);
 726
 727                        next if $parent eq $parents[0]; # Skip first parent
 728                        # Add new changes being placed in tree by merge
 729                        my @cmd = (qw/rev-list --reverse/,
 730                                           $parent, qw/--not/);
 731                        foreach my $par (@parents) {
 732                                unless ($par eq $parent) {
 733                                        push @cmd, $par;
 734                                }
 735                        }
 736                        my @revsin = ();
 737                        my ($revlist, $ctx) = command_output_pipe(@cmd);
 738                        while (<$revlist>) {
 739                                my $irev = $_;
 740                                chomp $irev;
 741                                my (undef, $csvnrev, undef) =
 742                                        cmt_metadata($irev);
 743                                unless (defined $csvnrev) {
 744                                        # A child is missing SVN annotations...
 745                                        # this might be OK, or might not be.
 746                                        warn "W:child $irev is merged into revision "
 747                                                 ."$d but does not have git-svn metadata. "
 748                                                 ."This means git-svn cannot determine the "
 749                                                 ."svn revision numbers to place into the "
 750                                                 ."svn:mergeinfo property. You must ensure "
 751                                                 ."a branch is entirely committed to "
 752                                                 ."SVN before merging it in order for "
 753                                                 ."svn:mergeinfo population to function "
 754                                                 ."properly";
 755                                }
 756                                push @revsin, $csvnrev;
 757                        }
 758                        command_close_pipe($revlist, $ctx);
 759
 760                        last unless $all_parents_ok;
 761
 762                        # We now have a list of all SVN revnos which are
 763                        # merged by this particular parent. Integrate them.
 764                        next if $#revsin == -1;
 765                        my $newmergeinfo = "$branchpath:" . join(',', @revsin);
 766                        $aggregate_mergeinfo =
 767                                merge_merge_info($aggregate_mergeinfo,
 768                                                                 $newmergeinfo, 1);
 769                }
 770                if ($all_parents_ok and $aggregate_mergeinfo) {
 771                        return $aggregate_mergeinfo;
 772                }
 773        }
 774
 775        return undef;
 776}
 777
 778sub cmd_dcommit {
 779        my $head = shift;
 780        command_noisy(qw/update-index --refresh/);
 781        git_cmd_try { command_oneline(qw/diff-index --quiet HEAD/) }
 782                'Cannot dcommit with a dirty index.  Commit your changes first, '
 783                . "or stash them with `git stash'.\n";
 784        $head ||= 'HEAD';
 785
 786        my $old_head;
 787        if ($head ne 'HEAD') {
 788                $old_head = eval {
 789                        command_oneline([qw/symbolic-ref -q HEAD/])
 790                };
 791                if ($old_head) {
 792                        $old_head =~ s{^refs/heads/}{};
 793                } else {
 794                        $old_head = eval { command_oneline(qw/rev-parse HEAD/) };
 795                }
 796                command(['checkout', $head], STDERR => 0);
 797        }
 798
 799        my @refs;
 800        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD', \@refs);
 801        unless ($gs) {
 802                die "Unable to determine upstream SVN information from ",
 803                    "$head history.\nPerhaps the repository is empty.";
 804        }
 805
 806        if (defined $_commit_url) {
 807                $url = $_commit_url;
 808        } else {
 809                $url = eval { command_oneline('config', '--get',
 810                              "svn-remote.$gs->{repo_id}.commiturl") };
 811                if (!$url) {
 812                        $url = $gs->full_pushurl
 813                }
 814        }
 815
 816        my $last_rev = $_revision if defined $_revision;
 817        if ($url) {
 818                print "Committing to $url ...\n";
 819        }
 820        my ($linear_refs, $parents) = linearize_history($gs, \@refs);
 821        if ($_no_rebase && scalar(@$linear_refs) > 1) {
 822                warn "Attempting to commit more than one change while ",
 823                     "--no-rebase is enabled.\n",
 824                     "If these changes depend on each other, re-running ",
 825                     "without --no-rebase may be required."
 826        }
 827
 828        if (defined $_interactive){
 829                my $ask_default = "y";
 830                foreach my $d (@$linear_refs){
 831                        my ($fh, $ctx) = command_output_pipe(qw(show --summary), "$d");
 832                        while (<$fh>){
 833                                print $_;
 834                        }
 835                        command_close_pipe($fh, $ctx);
 836                        $_ = ask("Commit this patch to SVN? ([y]es (default)|[n]o|[q]uit|[a]ll): ",
 837                                 valid_re => qr/^(?:yes|y|no|n|quit|q|all|a)/i,
 838                                 default => $ask_default);
 839                        die "Commit this patch reply required" unless defined $_;
 840                        if (/^[nq]/i) {
 841                                exit(0);
 842                        } elsif (/^a/i) {
 843                                last;
 844                        }
 845                }
 846        }
 847
 848        my $expect_url = $url;
 849
 850        my $push_merge_info = eval {
 851                command_oneline(qw/config --get svn.pushmergeinfo/)
 852                };
 853        if (not defined($push_merge_info)
 854                        or $push_merge_info eq "false"
 855                        or $push_merge_info eq "no"
 856                        or $push_merge_info eq "never") {
 857                $push_merge_info = 0;
 858        }
 859
 860        unless (defined($_merge_info) || ! $push_merge_info) {
 861                # Preflight check of changes to ensure no issues with mergeinfo
 862                # This includes check for uncommitted-to-SVN parents
 863                # (other than the first parent, which we will handle),
 864                # information from different SVN repos, and paths
 865                # which are not underneath this repository root.
 866                my $rooturl = $gs->repos_root;
 867                foreach my $d (@$linear_refs) {
 868                        my %parentshash;
 869                        read_commit_parents(\%parentshash, $d);
 870                        my @realparents = @{$parentshash{$d}};
 871                        if ($#realparents > 0) {
 872                                # Merge commit
 873                                shift @realparents; # Remove/ignore first parent
 874                                foreach my $parent (@realparents) {
 875                                        my ($branchurl, $svnrev, $paruuid) = cmt_metadata($parent);
 876                                        unless (defined $paruuid) {
 877                                                # A parent is missing SVN annotations...
 878                                                # abort the whole operation.
 879                                                fatal "$parent is merged into revision $d, "
 880                                                         ."but does not have git-svn metadata. "
 881                                                         ."Either dcommit the branch or use a "
 882                                                         ."local cherry-pick, FF merge, or rebase "
 883                                                         ."instead of an explicit merge commit.";
 884                                        }
 885
 886                                        unless ($paruuid eq $uuid) {
 887                                                # Parent has SVN metadata from different repository
 888                                                fatal "merge parent $parent for change $d has "
 889                                                         ."git-svn uuid $paruuid, while current change "
 890                                                         ."has uuid $uuid!";
 891                                        }
 892
 893                                        unless ($branchurl =~ /^\Q$rooturl\E(.*)/) {
 894                                                # This branch is very strange indeed.
 895                                                fatal "merge parent $parent for $d is on branch "
 896                                                         ."$branchurl, which is not under the "
 897                                                         ."git-svn root $rooturl!";
 898                                        }
 899                                }
 900                        }
 901                }
 902        }
 903
 904        my $rewritten_parent;
 905        Git::SVN::remove_username($expect_url);
 906        if (defined($_merge_info)) {
 907                $_merge_info =~ tr{ }{\n};
 908        }
 909        while (1) {
 910                my $d = shift @$linear_refs or last;
 911                unless (defined $last_rev) {
 912                        (undef, $last_rev, undef) = cmt_metadata("$d~1");
 913                        unless (defined $last_rev) {
 914                                fatal "Unable to extract revision information ",
 915                                      "from commit $d~1";
 916                        }
 917                }
 918                if ($_dry_run) {
 919                        print "diff-tree $d~1 $d\n";
 920                } else {
 921                        my $cmt_rev;
 922
 923                        unless (defined($_merge_info) || ! $push_merge_info) {
 924                                $_merge_info = populate_merge_info($d, $gs,
 925                                                             $uuid,
 926                                                             $linear_refs,
 927                                                             $rewritten_parent);
 928                        }
 929
 930                        my %ed_opts = ( r => $last_rev,
 931                                        log => get_commit_entry($d)->{log},
 932                                        ra => Git::SVN::Ra->new($url),
 933                                        config => SVN::Core::config_get_config(
 934                                                $Git::SVN::Ra::config_dir
 935                                        ),
 936                                        tree_a => "$d~1",
 937                                        tree_b => $d,
 938                                        editor_cb => sub {
 939                                               print "Committed r$_[0]\n";
 940                                               $cmt_rev = $_[0];
 941                                        },
 942                                        mergeinfo => $_merge_info,
 943                                        svn_path => '');
 944                        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
 945                                print "No changes\n$d~1 == $d\n";
 946                        } elsif ($parents->{$d} && @{$parents->{$d}}) {
 947                                $gs->{inject_parents_dcommit}->{$cmt_rev} =
 948                                                               $parents->{$d};
 949                        }
 950                        $_fetch_all ? $gs->fetch_all : $gs->fetch;
 951                        $last_rev = $cmt_rev;
 952                        next if $_no_rebase;
 953
 954                        # we always want to rebase against the current HEAD,
 955                        # not any head that was passed to us
 956                        my @diff = command('diff-tree', $d,
 957                                           $gs->refname, '--');
 958                        my @finish;
 959                        if (@diff) {
 960                                @finish = rebase_cmd();
 961                                print STDERR "W: $d and ", $gs->refname,
 962                                             " differ, using @finish:\n",
 963                                             join("\n", @diff), "\n";
 964                        } else {
 965                                print "No changes between current HEAD and ",
 966                                      $gs->refname,
 967                                      "\nResetting to the latest ",
 968                                      $gs->refname, "\n";
 969                                @finish = qw/reset --mixed/;
 970                        }
 971                        command_noisy(@finish, $gs->refname);
 972
 973                        $rewritten_parent = command_oneline(qw/rev-parse HEAD/);
 974
 975                        if (@diff) {
 976                                @refs = ();
 977                                my ($url_, $rev_, $uuid_, $gs_) =
 978                                              working_head_info('HEAD', \@refs);
 979                                my ($linear_refs_, $parents_) =
 980                                              linearize_history($gs_, \@refs);
 981                                if (scalar(@$linear_refs) !=
 982                                    scalar(@$linear_refs_)) {
 983                                        fatal "# of revisions changed ",
 984                                          "\nbefore:\n",
 985                                          join("\n", @$linear_refs),
 986                                          "\n\nafter:\n",
 987                                          join("\n", @$linear_refs_), "\n",
 988                                          'If you are attempting to commit ',
 989                                          "merges, try running:\n\t",
 990                                          'git rebase --interactive',
 991                                          '--preserve-merges ',
 992                                          $gs->refname,
 993                                          "\nBefore dcommitting";
 994                                }
 995                                if ($url_ ne $expect_url) {
 996                                        if ($url_ eq $gs->metadata_url) {
 997                                                print
 998                                                  "Accepting rewritten URL:",
 999                                                  " $url_\n";
1000                                        } else {
1001                                                fatal
1002                                                  "URL mismatch after rebase:",
1003                                                  " $url_ != $expect_url";
1004                                        }
1005                                }
1006                                if ($uuid_ ne $uuid) {
1007                                        fatal "uuid mismatch after rebase: ",
1008                                              "$uuid_ != $uuid";
1009                                }
1010                                # remap parents
1011                                my (%p, @l, $i);
1012                                for ($i = 0; $i < scalar @$linear_refs; $i++) {
1013                                        my $new = $linear_refs_->[$i] or next;
1014                                        $p{$new} =
1015                                                $parents->{$linear_refs->[$i]};
1016                                        push @l, $new;
1017                                }
1018                                $parents = \%p;
1019                                $linear_refs = \@l;
1020                        }
1021                }
1022        }
1023
1024        if ($old_head) {
1025                my $new_head = command_oneline(qw/rev-parse HEAD/);
1026                my $new_is_symbolic = eval {
1027                        command_oneline(qw/symbolic-ref -q HEAD/);
1028                };
1029                if ($new_is_symbolic) {
1030                        print "dcommitted the branch ", $head, "\n";
1031                } else {
1032                        print "dcommitted on a detached HEAD because you gave ",
1033                              "a revision argument.\n",
1034                              "The rewritten commit is: ", $new_head, "\n";
1035                }
1036                command(['checkout', $old_head], STDERR => 0);
1037        }
1038
1039        unlink $gs->{index};
1040}
1041
1042sub cmd_branch {
1043        my ($branch_name, $head) = @_;
1044
1045        unless (defined $branch_name && length $branch_name) {
1046                die(($_tag ? "tag" : "branch") . " name required\n");
1047        }
1048        $head ||= 'HEAD';
1049
1050        my (undef, $rev, undef, $gs) = working_head_info($head);
1051        my $src = $gs->full_pushurl;
1052
1053        my $remote = Git::SVN::read_all_remotes()->{$gs->{repo_id}};
1054        my $allglobs = $remote->{ $_tag ? 'tags' : 'branches' };
1055        my $glob;
1056        if ($#{$allglobs} == 0) {
1057                $glob = $allglobs->[0];
1058        } else {
1059                unless(defined $_branch_dest) {
1060                        die "Multiple ",
1061                            $_tag ? "tag" : "branch",
1062                            " paths defined for Subversion repository.\n",
1063                            "You must specify where you want to create the ",
1064                            $_tag ? "tag" : "branch",
1065                            " with the --destination argument.\n";
1066                }
1067                foreach my $g (@{$allglobs}) {
1068                        # SVN::Git::Editor could probably be moved to Git.pm..
1069                        my $re = SVN::Git::Editor::glob2pat($g->{path}->{left});
1070                        if ($_branch_dest =~ /$re/) {
1071                                $glob = $g;
1072                                last;
1073                        }
1074                }
1075                unless (defined $glob) {
1076                        my $dest_re = qr/\b\Q$_branch_dest\E\b/;
1077                        foreach my $g (@{$allglobs}) {
1078                                $g->{path}->{left} =~ /$dest_re/ or next;
1079                                if (defined $glob) {
1080                                        die "Ambiguous destination: ",
1081                                            $_branch_dest, "\nmatches both '",
1082                                            $glob->{path}->{left}, "' and '",
1083                                            $g->{path}->{left}, "'\n";
1084                                }
1085                                $glob = $g;
1086                        }
1087                        unless (defined $glob) {
1088                                die "Unknown ",
1089                                    $_tag ? "tag" : "branch",
1090                                    " destination $_branch_dest\n";
1091                        }
1092                }
1093        }
1094        my ($lft, $rgt) = @{ $glob->{path} }{qw/left right/};
1095        my $url;
1096        if (defined $_commit_url) {
1097                $url = $_commit_url;
1098        } else {
1099                $url = eval { command_oneline('config', '--get',
1100                        "svn-remote.$gs->{repo_id}.commiturl") };
1101                if (!$url) {
1102                        $url = $remote->{pushurl} || $remote->{url};
1103                }
1104        }
1105        my $dst = join '/', $url, $lft, $branch_name, ($rgt || ());
1106
1107        if ($dst =~ /^https:/ && $src =~ /^http:/) {
1108                $src=~s/^http:/https:/;
1109        }
1110
1111        ::_req_svn();
1112
1113        my $ctx = SVN::Client->new(
1114                auth    => Git::SVN::Ra::_auth_providers(),
1115                log_msg => sub {
1116                        ${ $_[0] } = defined $_message
1117                                ? $_message
1118                                : 'Create ' . ($_tag ? 'tag ' : 'branch ' )
1119                                . $branch_name;
1120                },
1121        );
1122
1123        eval {
1124                $ctx->ls($dst, 'HEAD', 0);
1125        } and die "branch ${branch_name} already exists\n";
1126
1127        print "Copying ${src} at r${rev} to ${dst}...\n";
1128        $ctx->copy($src, $rev, $dst)
1129                unless $_dry_run;
1130
1131        $gs->fetch_all;
1132}
1133
1134sub cmd_find_rev {
1135        my $revision_or_hash = shift or die "SVN or git revision required ",
1136                                            "as a command-line argument\n";
1137        my $result;
1138        if ($revision_or_hash =~ /^r\d+$/) {
1139                my $head = shift;
1140                $head ||= 'HEAD';
1141                my @refs;
1142                my (undef, undef, $uuid, $gs) = working_head_info($head, \@refs);
1143                unless ($gs) {
1144                        die "Unable to determine upstream SVN information from ",
1145                            "$head history\n";
1146                }
1147                my $desired_revision = substr($revision_or_hash, 1);
1148                $result = $gs->rev_map_get($desired_revision, $uuid);
1149        } else {
1150                my (undef, $rev, undef) = cmt_metadata($revision_or_hash);
1151                $result = $rev;
1152        }
1153        print "$result\n" if $result;
1154}
1155
1156sub auto_create_empty_directories {
1157        my ($gs) = @_;
1158        my $var = eval { command_oneline('config', '--get', '--bool',
1159                                         "svn-remote.$gs->{repo_id}.automkdirs") };
1160        # By default, create empty directories by consulting the unhandled log,
1161        # but allow setting it to 'false' to skip it.
1162        return !($var && $var eq 'false');
1163}
1164
1165sub cmd_rebase {
1166        command_noisy(qw/update-index --refresh/);
1167        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1168        unless ($gs) {
1169                die "Unable to determine upstream SVN information from ",
1170                    "working tree history\n";
1171        }
1172        if ($_dry_run) {
1173                print "Remote Branch: " . $gs->refname . "\n";
1174                print "SVN URL: " . $url . "\n";
1175                return;
1176        }
1177        if (command(qw/diff-index HEAD --/)) {
1178                print STDERR "Cannot rebase with uncommited changes:\n";
1179                command_noisy('status');
1180                exit 1;
1181        }
1182        unless ($_local) {
1183                # rebase will checkout for us, so no need to do it explicitly
1184                $_no_checkout = 'true';
1185                $_fetch_all ? $gs->fetch_all : $gs->fetch;
1186        }
1187        command_noisy(rebase_cmd(), $gs->refname);
1188        if (auto_create_empty_directories($gs)) {
1189                $gs->mkemptydirs;
1190        }
1191}
1192
1193sub cmd_show_ignore {
1194        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1195        $gs ||= Git::SVN->new;
1196        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
1197        $gs->prop_walk($gs->{path}, $r, sub {
1198                my ($gs, $path, $props) = @_;
1199                print STDOUT "\n# $path\n";
1200                my $s = $props->{'svn:ignore'} or return;
1201                $s =~ s/[\r\n]+/\n/g;
1202                $s =~ s/^\n+//;
1203                chomp $s;
1204                $s =~ s#^#$path#gm;
1205                print STDOUT "$s\n";
1206        });
1207}
1208
1209sub cmd_show_externals {
1210        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1211        $gs ||= Git::SVN->new;
1212        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
1213        $gs->prop_walk($gs->{path}, $r, sub {
1214                my ($gs, $path, $props) = @_;
1215                print STDOUT "\n# $path\n";
1216                my $s = $props->{'svn:externals'} or return;
1217                $s =~ s/[\r\n]+/\n/g;
1218                chomp $s;
1219                $s =~ s#^#$path#gm;
1220                print STDOUT "$s\n";
1221        });
1222}
1223
1224sub cmd_create_ignore {
1225        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1226        $gs ||= Git::SVN->new;
1227        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
1228        $gs->prop_walk($gs->{path}, $r, sub {
1229                my ($gs, $path, $props) = @_;
1230                # $path is of the form /path/to/dir/
1231                $path = '.' . $path;
1232                # SVN can have attributes on empty directories,
1233                # which git won't track
1234                mkpath([$path]) unless -d $path;
1235                my $ignore = $path . '.gitignore';
1236                my $s = $props->{'svn:ignore'} or return;
1237                open(GITIGNORE, '>', $ignore)
1238                  or fatal("Failed to open `$ignore' for writing: $!");
1239                $s =~ s/[\r\n]+/\n/g;
1240                $s =~ s/^\n+//;
1241                chomp $s;
1242                # Prefix all patterns so that the ignore doesn't apply
1243                # to sub-directories.
1244                $s =~ s#^#/#gm;
1245                print GITIGNORE "$s\n";
1246                close(GITIGNORE)
1247                  or fatal("Failed to close `$ignore': $!");
1248                command_noisy('add', '-f', $ignore);
1249        });
1250}
1251
1252sub cmd_mkdirs {
1253        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1254        $gs ||= Git::SVN->new;
1255        $gs->mkemptydirs($_revision);
1256}
1257
1258sub canonicalize_path {
1259        my ($path) = @_;
1260        my $dot_slash_added = 0;
1261        if (substr($path, 0, 1) ne "/") {
1262                $path = "./" . $path;
1263                $dot_slash_added = 1;
1264        }
1265        # File::Spec->canonpath doesn't collapse x/../y into y (for a
1266        # good reason), so let's do this manually.
1267        $path =~ s#/+#/#g;
1268        $path =~ s#/\.(?:/|$)#/#g;
1269        $path =~ s#/[^/]+/\.\.##g;
1270        $path =~ s#/$##g;
1271        $path =~ s#^\./## if $dot_slash_added;
1272        $path =~ s#^/##;
1273        $path =~ s#^\.$##;
1274        return $path;
1275}
1276
1277sub canonicalize_url {
1278        my ($url) = @_;
1279        $url =~ s#^([^:]+://[^/]*/)(.*)$#$1 . canonicalize_path($2)#e;
1280        return $url;
1281}
1282
1283# get_svnprops(PATH)
1284# ------------------
1285# Helper for cmd_propget and cmd_proplist below.
1286sub get_svnprops {
1287        my $path = shift;
1288        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1289        $gs ||= Git::SVN->new;
1290
1291        # prefix THE PATH by the sub-directory from which the user
1292        # invoked us.
1293        $path = $cmd_dir_prefix . $path;
1294        fatal("No such file or directory: $path") unless -e $path;
1295        my $is_dir = -d $path ? 1 : 0;
1296        $path = $gs->{path} . '/' . $path;
1297
1298        # canonicalize the path (otherwise libsvn will abort or fail to
1299        # find the file)
1300        $path = canonicalize_path($path);
1301
1302        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
1303        my $props;
1304        if ($is_dir) {
1305                (undef, undef, $props) = $gs->ra->get_dir($path, $r);
1306        }
1307        else {
1308                (undef, $props) = $gs->ra->get_file($path, $r, undef);
1309        }
1310        return $props;
1311}
1312
1313# cmd_propget (PROP, PATH)
1314# ------------------------
1315# Print the SVN property PROP for PATH.
1316sub cmd_propget {
1317        my ($prop, $path) = @_;
1318        $path = '.' if not defined $path;
1319        usage(1) if not defined $prop;
1320        my $props = get_svnprops($path);
1321        if (not defined $props->{$prop}) {
1322                fatal("`$path' does not have a `$prop' SVN property.");
1323        }
1324        print $props->{$prop} . "\n";
1325}
1326
1327# cmd_proplist (PATH)
1328# -------------------
1329# Print the list of SVN properties for PATH.
1330sub cmd_proplist {
1331        my $path = shift;
1332        $path = '.' if not defined $path;
1333        my $props = get_svnprops($path);
1334        print "Properties on '$path':\n";
1335        foreach (sort keys %{$props}) {
1336                print "  $_\n";
1337        }
1338}
1339
1340sub cmd_multi_init {
1341        my $url = shift;
1342        unless (defined $_trunk || @_branches || @_tags) {
1343                usage(1);
1344        }
1345
1346        $_prefix = '' unless defined $_prefix;
1347        if (defined $url) {
1348                $url = canonicalize_url($url);
1349                init_subdir(@_);
1350        }
1351        do_git_init_db();
1352        if (defined $_trunk) {
1353                $_trunk =~ s#^/+##;
1354                my $trunk_ref = 'refs/remotes/' . $_prefix . 'trunk';
1355                # try both old-style and new-style lookups:
1356                my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
1357                unless ($gs_trunk) {
1358                        my ($trunk_url, $trunk_path) =
1359                                              complete_svn_url($url, $_trunk);
1360                        $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
1361                                                   undef, $trunk_ref);
1362                }
1363        }
1364        return unless @_branches || @_tags;
1365        my $ra = $url ? Git::SVN::Ra->new($url) : undef;
1366        foreach my $path (@_branches) {
1367                complete_url_ls_init($ra, $path, '--branches/-b', $_prefix);
1368        }
1369        foreach my $path (@_tags) {
1370                complete_url_ls_init($ra, $path, '--tags/-t', $_prefix.'tags/');
1371        }
1372}
1373
1374sub cmd_multi_fetch {
1375        $Git::SVN::no_reuse_existing = undef;
1376        my $remotes = Git::SVN::read_all_remotes();
1377        foreach my $repo_id (sort keys %$remotes) {
1378                if ($remotes->{$repo_id}->{url}) {
1379                        Git::SVN::fetch_all($repo_id, $remotes);
1380                }
1381        }
1382}
1383
1384# this command is special because it requires no metadata
1385sub cmd_commit_diff {
1386        my ($ta, $tb, $url) = @_;
1387        my $usage = "Usage: $0 commit-diff -r<revision> ".
1388                    "<tree-ish> <tree-ish> [<URL>]";
1389        fatal($usage) if (!defined $ta || !defined $tb);
1390        my $svn_path = '';
1391        if (!defined $url) {
1392                my $gs = eval { Git::SVN->new };
1393                if (!$gs) {
1394                        fatal("Needed URL or usable git-svn --id in ",
1395                              "the command-line\n", $usage);
1396                }
1397                $url = $gs->{url};
1398                $svn_path = $gs->{path};
1399        }
1400        unless (defined $_revision) {
1401                fatal("-r|--revision is a required argument\n", $usage);
1402        }
1403        if (defined $_message && defined $_file) {
1404                fatal("Both --message/-m and --file/-F specified ",
1405                      "for the commit message.\n",
1406                      "I have no idea what you mean");
1407        }
1408        if (defined $_file) {
1409                $_message = file_to_s($_file);
1410        } else {
1411                $_message ||= get_commit_entry($tb)->{log};
1412        }
1413        my $ra ||= Git::SVN::Ra->new($url);
1414        my $r = $_revision;
1415        if ($r eq 'HEAD') {
1416                $r = $ra->get_latest_revnum;
1417        } elsif ($r !~ /^\d+$/) {
1418                die "revision argument: $r not understood by git-svn\n";
1419        }
1420        my %ed_opts = ( r => $r,
1421                        log => $_message,
1422                        ra => $ra,
1423                        tree_a => $ta,
1424                        tree_b => $tb,
1425                        editor_cb => sub { print "Committed r$_[0]\n" },
1426                        svn_path => $svn_path );
1427        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
1428                print "No changes\n$ta == $tb\n";
1429        }
1430}
1431
1432sub escape_uri_only {
1433        my ($uri) = @_;
1434        my @tmp;
1435        foreach (split m{/}, $uri) {
1436                s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
1437                push @tmp, $_;
1438        }
1439        join('/', @tmp);
1440}
1441
1442sub escape_url {
1443        my ($url) = @_;
1444        if ($url =~ m#^([^:]+)://([^/]*)(.*)$#) {
1445                my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
1446                $url = "$scheme://$domain$uri";
1447        }
1448        $url;
1449}
1450
1451sub cmd_info {
1452        my $path = canonicalize_path(defined($_[0]) ? $_[0] : ".");
1453        my $fullpath = canonicalize_path($cmd_dir_prefix . $path);
1454        if (exists $_[1]) {
1455                die "Too many arguments specified\n";
1456        }
1457
1458        my ($file_type, $diff_status) = find_file_type_and_diff_status($path);
1459
1460        if (!$file_type && !$diff_status) {
1461                print STDERR "svn: '$path' is not under version control\n";
1462                exit 1;
1463        }
1464
1465        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1466        unless ($gs) {
1467                die "Unable to determine upstream SVN information from ",
1468                    "working tree history\n";
1469        }
1470
1471        # canonicalize_path() will return "" to make libsvn 1.5.x happy,
1472        $path = "." if $path eq "";
1473
1474        my $full_url = $url . ($fullpath eq "" ? "" : "/$fullpath");
1475
1476        if ($_url) {
1477                print escape_url($full_url), "\n";
1478                return;
1479        }
1480
1481        my $result = "Path: $path\n";
1482        $result .= "Name: " . basename($path) . "\n" if $file_type ne "dir";
1483        $result .= "URL: " . escape_url($full_url) . "\n";
1484
1485        eval {
1486                my $repos_root = $gs->repos_root;
1487                Git::SVN::remove_username($repos_root);
1488                $result .= "Repository Root: " . escape_url($repos_root) . "\n";
1489        };
1490        if ($@) {
1491                $result .= "Repository Root: (offline)\n";
1492        }
1493        ::_req_svn();
1494        $result .= "Repository UUID: $uuid\n" unless $diff_status eq "A" &&
1495                (::compare_svn_version('1.5.4') <= 0 || $file_type ne "dir");
1496        $result .= "Revision: " . ($diff_status eq "A" ? 0 : $rev) . "\n";
1497
1498        $result .= "Node Kind: " .
1499                   ($file_type eq "dir" ? "directory" : "file") . "\n";
1500
1501        my $schedule = $diff_status eq "A"
1502                       ? "add"
1503                       : ($diff_status eq "D" ? "delete" : "normal");
1504        $result .= "Schedule: $schedule\n";
1505
1506        if ($diff_status eq "A") {
1507                print $result, "\n";
1508                return;
1509        }
1510
1511        my ($lc_author, $lc_rev, $lc_date_utc);
1512        my @args = Git::SVN::Log::git_svn_log_cmd($rev, $rev, "--", $fullpath);
1513        my $log = command_output_pipe(@args);
1514        my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
1515        while (<$log>) {
1516                if (/^${esc_color}author (.+) <[^>]+> (\d+) ([\-\+]?\d+)$/o) {
1517                        $lc_author = $1;
1518                        $lc_date_utc = Git::SVN::Log::parse_git_date($2, $3);
1519                } elsif (/^${esc_color}    (git-svn-id:.+)$/o) {
1520                        (undef, $lc_rev, undef) = ::extract_metadata($1);
1521                }
1522        }
1523        close $log;
1524
1525        Git::SVN::Log::set_local_timezone();
1526
1527        $result .= "Last Changed Author: $lc_author\n";
1528        $result .= "Last Changed Rev: $lc_rev\n";
1529        $result .= "Last Changed Date: " .
1530                   Git::SVN::Log::format_svn_date($lc_date_utc) . "\n";
1531
1532        if ($file_type ne "dir") {
1533                my $text_last_updated_date =
1534                    ($diff_status eq "D" ? $lc_date_utc : (stat $path)[9]);
1535                $result .=
1536                    "Text Last Updated: " .
1537                    Git::SVN::Log::format_svn_date($text_last_updated_date) .
1538                    "\n";
1539                my $checksum;
1540                if ($diff_status eq "D") {
1541                        my ($fh, $ctx) =
1542                            command_output_pipe(qw(cat-file blob), "HEAD:$path");
1543                        if ($file_type eq "link") {
1544                                my $file_name = <$fh>;
1545                                $checksum = md5sum("link $file_name");
1546                        } else {
1547                                $checksum = md5sum($fh);
1548                        }
1549                        command_close_pipe($fh, $ctx);
1550                } elsif ($file_type eq "link") {
1551                        my $file_name =
1552                            command(qw(cat-file blob), "HEAD:$path");
1553                        $checksum =
1554                            md5sum("link " . $file_name);
1555                } else {
1556                        open FILE, "<", $path or die $!;
1557                        $checksum = md5sum(\*FILE);
1558                        close FILE or die $!;
1559                }
1560                $result .= "Checksum: " . $checksum . "\n";
1561        }
1562
1563        print $result, "\n";
1564}
1565
1566sub cmd_reset {
1567        my $target = shift || $_revision or die "SVN revision required\n";
1568        $target = $1 if $target =~ /^r(\d+)$/;
1569        $target =~ /^\d+$/ or die "Numeric SVN revision expected\n";
1570        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1571        unless ($gs) {
1572                die "Unable to determine upstream SVN information from ".
1573                    "history\n";
1574        }
1575        my ($r, $c) = $gs->find_rev_before($target, not $_fetch_parent);
1576        die "Cannot find SVN revision $target\n" unless defined($c);
1577        $gs->rev_map_set($r, $c, 'reset', $uuid);
1578        print "r$r = $c ($gs->{ref_id})\n";
1579}
1580
1581sub cmd_gc {
1582        if (!$can_compress) {
1583                warn "Compress::Zlib could not be found; unhandled.log " .
1584                     "files will not be compressed.\n";
1585        }
1586        find({ wanted => \&gc_directory, no_chdir => 1}, "$ENV{GIT_DIR}/svn");
1587}
1588
1589########################### utility functions #########################
1590
1591sub rebase_cmd {
1592        my @cmd = qw/rebase/;
1593        push @cmd, '-v' if $_verbose;
1594        push @cmd, qw/--merge/ if $_merge;
1595        push @cmd, "--strategy=$_strategy" if $_strategy;
1596        push @cmd, "--preserve-merges" if $_preserve_merges;
1597        @cmd;
1598}
1599
1600sub post_fetch_checkout {
1601        return if $_no_checkout;
1602        my $gs = $Git::SVN::_head or return;
1603        return if verify_ref('refs/heads/master^0');
1604
1605        # look for "trunk" ref if it exists
1606        my $remote = Git::SVN::read_all_remotes()->{$gs->{repo_id}};
1607        my $fetch = $remote->{fetch};
1608        if ($fetch) {
1609                foreach my $p (keys %$fetch) {
1610                        basename($fetch->{$p}) eq 'trunk' or next;
1611                        $gs = Git::SVN->new($fetch->{$p}, $gs->{repo_id}, $p);
1612                        last;
1613                }
1614        }
1615
1616        my $valid_head = verify_ref('HEAD^0');
1617        command_noisy(qw(update-ref refs/heads/master), $gs->refname);
1618        return if ($valid_head || !verify_ref('HEAD^0'));
1619
1620        return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
1621        my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
1622        return if -f $index;
1623
1624        return if command_oneline(qw/rev-parse --is-inside-work-tree/) eq 'false';
1625        return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
1626        command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
1627        print STDERR "Checked out HEAD:\n  ",
1628                     $gs->full_url, " r", $gs->last_rev, "\n";
1629        if (auto_create_empty_directories($gs)) {
1630                $gs->mkemptydirs($gs->last_rev);
1631        }
1632}
1633
1634sub complete_svn_url {
1635        my ($url, $path) = @_;
1636        $path =~ s#/+$##;
1637        if ($path !~ m#^[a-z\+]+://#) {
1638                if (!defined $url || $url !~ m#^[a-z\+]+://#) {
1639                        fatal("E: '$path' is not a complete URL ",
1640                              "and a separate URL is not specified");
1641                }
1642                return ($url, $path);
1643        }
1644        return ($path, '');
1645}
1646
1647sub complete_url_ls_init {
1648        my ($ra, $repo_path, $switch, $pfx) = @_;
1649        unless ($repo_path) {
1650                print STDERR "W: $switch not specified\n";
1651                return;
1652        }
1653        $repo_path =~ s#/+$##;
1654        if ($repo_path =~ m#^[a-z\+]+://#) {
1655                $ra = Git::SVN::Ra->new($repo_path);
1656                $repo_path = '';
1657        } else {
1658                $repo_path =~ s#^/+##;
1659                unless ($ra) {
1660                        fatal("E: '$repo_path' is not a complete URL ",
1661                              "and a separate URL is not specified");
1662                }
1663        }
1664        my $url = $ra->{url};
1665        my $gs = Git::SVN->init($url, undef, undef, undef, 1);
1666        my $k = "svn-remote.$gs->{repo_id}.url";
1667        my $orig_url = eval { command_oneline(qw/config --get/, $k) };
1668        if ($orig_url && ($orig_url ne $gs->{url})) {
1669                die "$k already set: $orig_url\n",
1670                    "wanted to set to: $gs->{url}\n";
1671        }
1672        command_oneline('config', $k, $gs->{url}) unless $orig_url;
1673        my $remote_path = "$gs->{path}/$repo_path";
1674        $remote_path =~ s{%([0-9A-F]{2})}{chr hex($1)}ieg;
1675        $remote_path =~ s#/+#/#g;
1676        $remote_path =~ s#^/##g;
1677        $remote_path .= "/*" if $remote_path !~ /\*/;
1678        my ($n) = ($switch =~ /^--(\w+)/);
1679        if (length $pfx && $pfx !~ m#/$#) {
1680                die "--prefix='$pfx' must have a trailing slash '/'\n";
1681        }
1682        command_noisy('config',
1683                      '--add',
1684                      "svn-remote.$gs->{repo_id}.$n",
1685                      "$remote_path:refs/remotes/$pfx*" .
1686                        ('/*' x (($remote_path =~ tr/*/*/) - 1)) );
1687}
1688
1689sub verify_ref {
1690        my ($ref) = @_;
1691        eval { command_oneline([ 'rev-parse', '--verify', $ref ],
1692                               { STDERR => 0 }); };
1693}
1694
1695sub get_tree_from_treeish {
1696        my ($treeish) = @_;
1697        # $treeish can be a symbolic ref, too:
1698        my $type = command_oneline(qw/cat-file -t/, $treeish);
1699        my $expected;
1700        while ($type eq 'tag') {
1701                ($treeish, $type) = command(qw/cat-file tag/, $treeish);
1702        }
1703        if ($type eq 'commit') {
1704                $expected = (grep /^tree /, command(qw/cat-file commit/,
1705                                                    $treeish))[0];
1706                ($expected) = ($expected =~ /^tree ($sha1)$/o);
1707                die "Unable to get tree from $treeish\n" unless $expected;
1708        } elsif ($type eq 'tree') {
1709                $expected = $treeish;
1710        } else {
1711                die "$treeish is a $type, expected tree, tag or commit\n";
1712        }
1713        return $expected;
1714}
1715
1716sub get_commit_entry {
1717        my ($treeish) = shift;
1718        my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
1719        my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
1720        my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
1721        open my $log_fh, '>', $commit_editmsg or croak $!;
1722
1723        my $type = command_oneline(qw/cat-file -t/, $treeish);
1724        if ($type eq 'commit' || $type eq 'tag') {
1725                my ($msg_fh, $ctx) = command_output_pipe('cat-file',
1726                                                         $type, $treeish);
1727                my $in_msg = 0;
1728                my $author;
1729                my $saw_from = 0;
1730                my $msgbuf = "";
1731                while (<$msg_fh>) {
1732                        if (!$in_msg) {
1733                                $in_msg = 1 if (/^\s*$/);
1734                                $author = $1 if (/^author (.*>)/);
1735                        } elsif (/^git-svn-id: /) {
1736                                # skip this for now, we regenerate the
1737                                # correct one on re-fetch anyways
1738                                # TODO: set *:merge properties or like...
1739                        } else {
1740                                if (/^From:/ || /^Signed-off-by:/) {
1741                                        $saw_from = 1;
1742                                }
1743                                $msgbuf .= $_;
1744                        }
1745                }
1746                $msgbuf =~ s/\s+$//s;
1747                if ($Git::SVN::_add_author_from && defined($author)
1748                    && !$saw_from) {
1749                        $msgbuf .= "\n\nFrom: $author";
1750                }
1751                print $log_fh $msgbuf or croak $!;
1752                command_close_pipe($msg_fh, $ctx);
1753        }
1754        close $log_fh or croak $!;
1755
1756        if ($_edit || ($type eq 'tree')) {
1757                chomp(my $editor = command_oneline(qw(var GIT_EDITOR)));
1758                system('sh', '-c', $editor.' "$@"', $editor, $commit_editmsg);
1759        }
1760        rename $commit_editmsg, $commit_msg or croak $!;
1761        {
1762                require Encode;
1763                # SVN requires messages to be UTF-8 when entering the repo
1764                local $/;
1765                open $log_fh, '<', $commit_msg or croak $!;
1766                binmode $log_fh;
1767                chomp($log_entry{log} = <$log_fh>);
1768
1769                my $enc = Git::config('i18n.commitencoding') || 'UTF-8';
1770                my $msg = $log_entry{log};
1771
1772                eval { $msg = Encode::decode($enc, $msg, 1) };
1773                if ($@) {
1774                        die "Could not decode as $enc:\n", $msg,
1775                            "\nPerhaps you need to set i18n.commitencoding\n";
1776                }
1777
1778                eval { $msg = Encode::encode('UTF-8', $msg, 1) };
1779                die "Could not encode as UTF-8:\n$msg\n" if $@;
1780
1781                $log_entry{log} = $msg;
1782
1783                close $log_fh or croak $!;
1784        }
1785        unlink $commit_msg;
1786        \%log_entry;
1787}
1788
1789sub s_to_file {
1790        my ($str, $file, $mode) = @_;
1791        open my $fd,'>',$file or croak $!;
1792        print $fd $str,"\n" or croak $!;
1793        close $fd or croak $!;
1794        chmod ($mode &~ umask, $file) if (defined $mode);
1795}
1796
1797sub file_to_s {
1798        my $file = shift;
1799        open my $fd,'<',$file or croak "$!: file: $file\n";
1800        local $/;
1801        my $ret = <$fd>;
1802        close $fd or croak $!;
1803        $ret =~ s/\s*$//s;
1804        return $ret;
1805}
1806
1807# '<svn username> = real-name <email address>' mapping based on git-svnimport:
1808sub load_authors {
1809        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1810        my $log = $cmd eq 'log';
1811        while (<$authors>) {
1812                chomp;
1813                next unless /^(.+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
1814                my ($user, $name, $email) = ($1, $2, $3);
1815                if ($log) {
1816                        $Git::SVN::Log::rusers{"$name <$email>"} = $user;
1817                } else {
1818                        $users{$user} = [$name, $email];
1819                }
1820        }
1821        close $authors or croak $!;
1822}
1823
1824# convert GetOpt::Long specs for use by git-config
1825sub read_git_config {
1826        my $opts = shift;
1827        my @config_only;
1828        foreach my $o (keys %$opts) {
1829                # if we have mixedCase and a long option-only, then
1830                # it's a config-only variable that we don't need for
1831                # the command-line.
1832                push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
1833                my $v = $opts->{$o};
1834                my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
1835                $key =~ s/-//g;
1836                my $arg = 'git config';
1837                $arg .= ' --int' if ($o =~ /[:=]i$/);
1838                $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
1839                if (ref $v eq 'ARRAY') {
1840                        chomp(my @tmp = `$arg --get-all svn.$key`);
1841                        @$v = @tmp if @tmp;
1842                } else {
1843                        chomp(my $tmp = `$arg --get svn.$key`);
1844                        if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
1845                                $$v = $tmp;
1846                        }
1847                }
1848        }
1849        delete @$opts{@config_only} if @config_only;
1850}
1851
1852sub extract_metadata {
1853        my $id = shift or return (undef, undef, undef);
1854        my ($url, $rev, $uuid) = ($id =~ /^\s*git-svn-id:\s+(.*)\@(\d+)
1855                                                        \s([a-f\d\-]+)$/ix);
1856        if (!defined $rev || !$uuid || !$url) {
1857                # some of the original repositories I made had
1858                # identifiers like this:
1859                ($rev, $uuid) = ($id =~/^\s*git-svn-id:\s(\d+)\@([a-f\d\-]+)/i);
1860        }
1861        return ($url, $rev, $uuid);
1862}
1863
1864sub cmt_metadata {
1865        return extract_metadata((grep(/^git-svn-id: /,
1866                command(qw/cat-file commit/, shift)))[-1]);
1867}
1868
1869sub cmt_sha2rev_batch {
1870        my %s2r;
1871        my ($pid, $in, $out, $ctx) = command_bidi_pipe(qw/cat-file --batch/);
1872        my $list = shift;
1873
1874        foreach my $sha (@{$list}) {
1875                my $first = 1;
1876                my $size = 0;
1877                print $out $sha, "\n";
1878
1879                while (my $line = <$in>) {
1880                        if ($first && $line =~ /^[[:xdigit:]]{40}\smissing$/) {
1881                                last;
1882                        } elsif ($first &&
1883                               $line =~ /^[[:xdigit:]]{40}\scommit\s(\d+)$/) {
1884                                $first = 0;
1885                                $size = $1;
1886                                next;
1887                        } elsif ($line =~ /^(git-svn-id: )/) {
1888                                my (undef, $rev, undef) =
1889                                                      extract_metadata($line);
1890                                $s2r{$sha} = $rev;
1891                        }
1892
1893                        $size -= length($line);
1894                        last if ($size == 0);
1895                }
1896        }
1897
1898        command_close_bidi_pipe($pid, $in, $out, $ctx);
1899
1900        return \%s2r;
1901}
1902
1903sub working_head_info {
1904        my ($head, $refs) = @_;
1905        my @args = qw/rev-list --first-parent --pretty=medium/;
1906        my ($fh, $ctx) = command_output_pipe(@args, $head);
1907        my $hash;
1908        my %max;
1909        while (<$fh>) {
1910                if ( m{^commit ($::sha1)$} ) {
1911                        unshift @$refs, $hash if $hash and $refs;
1912                        $hash = $1;
1913                        next;
1914                }
1915                next unless s{^\s*(git-svn-id:)}{$1};
1916                my ($url, $rev, $uuid) = extract_metadata($_);
1917                if (defined $url && defined $rev) {
1918                        next if $max{$url} and $max{$url} < $rev;
1919                        if (my $gs = Git::SVN->find_by_url($url)) {
1920                                my $c = $gs->rev_map_get($rev, $uuid);
1921                                if ($c && $c eq $hash) {
1922                                        close $fh; # break the pipe
1923                                        return ($url, $rev, $uuid, $gs);
1924                                } else {
1925                                        $max{$url} ||= $gs->rev_map_max;
1926                                }
1927                        }
1928                }
1929        }
1930        command_close_pipe($fh, $ctx);
1931        (undef, undef, undef, undef);
1932}
1933
1934sub read_commit_parents {
1935        my ($parents, $c) = @_;
1936        chomp(my $p = command_oneline(qw/rev-list --parents -1/, $c));
1937        $p =~ s/^($c)\s*// or die "rev-list --parents -1 $c failed!\n";
1938        @{$parents->{$c}} = split(/ /, $p);
1939}
1940
1941sub linearize_history {
1942        my ($gs, $refs) = @_;
1943        my %parents;
1944        foreach my $c (@$refs) {
1945                read_commit_parents(\%parents, $c);
1946        }
1947
1948        my @linear_refs;
1949        my %skip = ();
1950        my $last_svn_commit = $gs->last_commit;
1951        foreach my $c (reverse @$refs) {
1952                next if $c eq $last_svn_commit;
1953                last if $skip{$c};
1954
1955                unshift @linear_refs, $c;
1956                $skip{$c} = 1;
1957
1958                # we only want the first parent to diff against for linear
1959                # history, we save the rest to inject when we finalize the
1960                # svn commit
1961                my $fp_a = verify_ref("$c~1");
1962                my $fp_b = shift @{$parents{$c}} if $parents{$c};
1963                if (!$fp_a || !$fp_b) {
1964                        die "Commit $c\n",
1965                            "has no parent commit, and therefore ",
1966                            "nothing to diff against.\n",
1967                            "You should be working from a repository ",
1968                            "originally created by git-svn\n";
1969                }
1970                if ($fp_a ne $fp_b) {
1971                        die "$c~1 = $fp_a, however parsing commit $c ",
1972                            "revealed that:\n$c~1 = $fp_b\nBUG!\n";
1973                }
1974
1975                foreach my $p (@{$parents{$c}}) {
1976                        $skip{$p} = 1;
1977                }
1978        }
1979        (\@linear_refs, \%parents);
1980}
1981
1982sub find_file_type_and_diff_status {
1983        my ($path) = @_;
1984        return ('dir', '') if $path eq '';
1985
1986        my $diff_output =
1987            command_oneline(qw(diff --cached --name-status --), $path) || "";
1988        my $diff_status = (split(' ', $diff_output))[0] || "";
1989
1990        my $ls_tree = command_oneline(qw(ls-tree HEAD), $path) || "";
1991
1992        return (undef, undef) if !$diff_status && !$ls_tree;
1993
1994        if ($diff_status eq "A") {
1995                return ("link", $diff_status) if -l $path;
1996                return ("dir", $diff_status) if -d $path;
1997                return ("file", $diff_status);
1998        }
1999
2000        my $mode = (split(' ', $ls_tree))[0] || "";
2001
2002        return ("link", $diff_status) if $mode eq "120000";
2003        return ("dir", $diff_status) if $mode eq "040000";
2004        return ("file", $diff_status);
2005}
2006
2007sub md5sum {
2008        my $arg = shift;
2009        my $ref = ref $arg;
2010        my $md5 = Digest::MD5->new();
2011        if ($ref eq 'GLOB' || $ref eq 'IO::File' || $ref eq 'File::Temp') {
2012                $md5->addfile($arg) or croak $!;
2013        } elsif ($ref eq 'SCALAR') {
2014                $md5->add($$arg) or croak $!;
2015        } elsif (!$ref) {
2016                $md5->add($arg) or croak $!;
2017        } else {
2018                ::fatal "Can't provide MD5 hash for unknown ref type: '", $ref, "'";
2019        }
2020        return $md5->hexdigest();
2021}
2022
2023sub gc_directory {
2024        if ($can_compress && -f $_ && basename($_) eq "unhandled.log") {
2025                my $out_filename = $_ . ".gz";
2026                open my $in_fh, "<", $_ or die "Unable to open $_: $!\n";
2027                binmode $in_fh;
2028                my $gz = Compress::Zlib::gzopen($out_filename, "ab") or
2029                                die "Unable to open $out_filename: $!\n";
2030
2031                my $res;
2032                while ($res = sysread($in_fh, my $str, 1024)) {
2033                        $gz->gzwrite($str) or
2034                                die "Unable to write: ".$gz->gzerror()."!\n";
2035                }
2036                unlink $_ or die "unlink $File::Find::name: $!\n";
2037        } elsif (-f $_ && basename($_) eq "index") {
2038                unlink $_ or die "unlink $_: $!\n";
2039        }
2040}
2041
2042package Git::SVN;
2043use strict;
2044use warnings;
2045use Fcntl qw/:DEFAULT :seek/;
2046use constant rev_map_fmt => 'NH40';
2047use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
2048            $_repack $_repack_flags $_use_svm_props $_head
2049            $_use_svnsync_props $no_reuse_existing $_minimize_url
2050            $_use_log_author $_add_author_from $_localtime/;
2051use Carp qw/croak/;
2052use File::Path qw/mkpath/;
2053use File::Copy qw/copy/;
2054use IPC::Open3;
2055use Time::Local;
2056use Memoize;  # core since 5.8.0, Jul 2002
2057use Memoize::Storable;
2058use POSIX qw(:signal_h);
2059
2060my ($_gc_nr, $_gc_period);
2061
2062# properties that we do not log:
2063my %SKIP_PROP;
2064BEGIN {
2065        %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
2066                                        svn:special svn:executable
2067                                        svn:entry:committed-rev
2068                                        svn:entry:last-author
2069                                        svn:entry:uuid
2070                                        svn:entry:committed-date/;
2071
2072        # some options are read globally, but can be overridden locally
2073        # per [svn-remote "..."] section.  Command-line options will *NOT*
2074        # override options set in an [svn-remote "..."] section
2075        no strict 'refs';
2076        for my $option (qw/follow_parent no_metadata use_svm_props
2077                           use_svnsync_props/) {
2078                my $key = $option;
2079                $key =~ tr/_//d;
2080                my $prop = "-$option";
2081                *$option = sub {
2082                        my ($self) = @_;
2083                        return $self->{$prop} if exists $self->{$prop};
2084                        my $k = "svn-remote.$self->{repo_id}.$key";
2085                        eval { command_oneline(qw/config --get/, $k) };
2086                        if ($@) {
2087                                $self->{$prop} = ${"Git::SVN::_$option"};
2088                        } else {
2089                                my $v = command_oneline(qw/config --bool/,$k);
2090                                $self->{$prop} = $v eq 'false' ? 0 : 1;
2091                        }
2092                        return $self->{$prop};
2093                }
2094        }
2095}
2096
2097
2098my (%LOCKFILES, %INDEX_FILES);
2099END {
2100        unlink keys %LOCKFILES if %LOCKFILES;
2101        unlink keys %INDEX_FILES if %INDEX_FILES;
2102}
2103
2104sub resolve_local_globs {
2105        my ($url, $fetch, $glob_spec) = @_;
2106        return unless defined $glob_spec;
2107        my $ref = $glob_spec->{ref};
2108        my $path = $glob_spec->{path};
2109        foreach (command(qw#for-each-ref --format=%(refname) refs/#)) {
2110                next unless m#^$ref->{regex}$#;
2111                my $p = $1;
2112                my $pathname = desanitize_refname($path->full_path($p));
2113                my $refname = desanitize_refname($ref->full_path($p));
2114                if (my $existing = $fetch->{$pathname}) {
2115                        if ($existing ne $refname) {
2116                                die "Refspec conflict:\n",
2117                                    "existing: $existing\n",
2118                                    " globbed: $refname\n";
2119                        }
2120                        my $u = (::cmt_metadata("$refname"))[0];
2121                        $u =~ s!^\Q$url\E(/|$)!! or die
2122                          "$refname: '$url' not found in '$u'\n";
2123                        if ($pathname ne $u) {
2124                                warn "W: Refspec glob conflict ",
2125                                     "(ref: $refname):\n",
2126                                     "expected path: $pathname\n",
2127                                     "    real path: $u\n",
2128                                     "Continuing ahead with $u\n";
2129                                next;
2130                        }
2131                } else {
2132                        $fetch->{$pathname} = $refname;
2133                }
2134        }
2135}
2136
2137sub parse_revision_argument {
2138        my ($base, $head) = @_;
2139        if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
2140                return ($base, $head);
2141        }
2142        return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
2143        return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
2144        return ($head, $head) if ($::_revision eq 'HEAD');
2145        return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
2146        return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
2147        die "revision argument: $::_revision not understood by git-svn\n";
2148}
2149
2150sub fetch_all {
2151        my ($repo_id, $remotes) = @_;
2152        if (ref $repo_id) {
2153                my $gs = $repo_id;
2154                $repo_id = undef;
2155                $repo_id = $gs->{repo_id};
2156        }
2157        $remotes ||= read_all_remotes();
2158        my $remote = $remotes->{$repo_id} or
2159                     die "[svn-remote \"$repo_id\"] unknown\n";
2160        my $fetch = $remote->{fetch};
2161        my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
2162        my (@gs, @globs);
2163        my $ra = Git::SVN::Ra->new($url);
2164        my $uuid = $ra->get_uuid;
2165        my $head = $ra->get_latest_revnum;
2166
2167        # ignore errors, $head revision may not even exist anymore
2168        eval { $ra->get_log("", $head, 0, 1, 0, 1, sub { $head = $_[1] }) };
2169        warn "W: $@\n" if $@;
2170
2171        my $base = defined $fetch ? $head : 0;
2172
2173        # read the max revs for wildcard expansion (branches/*, tags/*)
2174        foreach my $t (qw/branches tags/) {
2175                defined $remote->{$t} or next;
2176                push @globs, @{$remote->{$t}};
2177
2178                my $max_rev = eval { tmp_config(qw/--int --get/,
2179                                         "svn-remote.$repo_id.${t}-maxRev") };
2180                if (defined $max_rev && ($max_rev < $base)) {
2181                        $base = $max_rev;
2182                } elsif (!defined $max_rev) {
2183                        $base = 0;
2184                }
2185        }
2186
2187        if ($fetch) {
2188                foreach my $p (sort keys %$fetch) {
2189                        my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
2190                        my $lr = $gs->rev_map_max;
2191                        if (defined $lr) {
2192                                $base = $lr if ($lr < $base);
2193                        }
2194                        push @gs, $gs;
2195                }
2196        }
2197
2198        ($base, $head) = parse_revision_argument($base, $head);
2199        $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
2200}
2201
2202sub read_all_remotes {
2203        my $r = {};
2204        my $use_svm_props = eval { command_oneline(qw/config --bool
2205            svn.useSvmProps/) };
2206        $use_svm_props = $use_svm_props eq 'true' if $use_svm_props;
2207        my $svn_refspec = qr{\s*(.*?)\s*:\s*(.+?)\s*};
2208        foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
2209                if (m!^(.+)\.fetch=$svn_refspec$!) {
2210                        my ($remote, $local_ref, $remote_ref) = ($1, $2, $3);
2211                        die("svn-remote.$remote: remote ref '$remote_ref' "
2212                            . "must start with 'refs/'\n")
2213                                unless $remote_ref =~ m{^refs/};
2214                        $local_ref = uri_decode($local_ref);
2215                        $r->{$remote}->{fetch}->{$local_ref} = $remote_ref;
2216                        $r->{$remote}->{svm} = {} if $use_svm_props;
2217                } elsif (m!^(.+)\.usesvmprops=\s*(.*)\s*$!) {
2218                        $r->{$1}->{svm} = {};
2219                } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
2220                        $r->{$1}->{url} = $2;
2221                } elsif (m!^(.+)\.pushurl=\s*(.*)\s*$!) {
2222                        $r->{$1}->{pushurl} = $2;
2223                } elsif (m!^(.+)\.ignore-refs=\s*(.*)\s*$!) {
2224                        $r->{$1}->{ignore_refs_regex} = $2;
2225                } elsif (m!^(.+)\.(branches|tags)=$svn_refspec$!) {
2226                        my ($remote, $t, $local_ref, $remote_ref) =
2227                                                             ($1, $2, $3, $4);
2228                        die("svn-remote.$remote: remote ref '$remote_ref' ($t) "
2229                            . "must start with 'refs/'\n")
2230                                unless $remote_ref =~ m{^refs/};
2231                        $local_ref = uri_decode($local_ref);
2232                        my $rs = {
2233                            t => $t,
2234                            remote => $remote,
2235                            path => Git::SVN::GlobSpec->new($local_ref, 1),
2236                            ref => Git::SVN::GlobSpec->new($remote_ref, 0) };
2237                        if (length($rs->{ref}->{right}) != 0) {
2238                                die "The '*' glob character must be the last ",
2239                                    "character of '$remote_ref'\n";
2240                        }
2241                        push @{ $r->{$remote}->{$t} }, $rs;
2242                }
2243        }
2244
2245        map {
2246                if (defined $r->{$_}->{svm}) {
2247                        my $svm;
2248                        eval {
2249                                my $section = "svn-remote.$_";
2250                                $svm = {
2251                                        source => tmp_config('--get',
2252                                            "$section.svm-source"),
2253                                        replace => tmp_config('--get',
2254                                            "$section.svm-replace"),
2255                                }
2256                        };
2257                        $r->{$_}->{svm} = $svm;
2258                }
2259        } keys %$r;
2260
2261        foreach my $remote (keys %$r) {
2262                foreach ( grep { defined $_ }
2263                          map { $r->{$remote}->{$_} } qw(branches tags) ) {
2264                        foreach my $rs ( @$_ ) {
2265                                $rs->{ignore_refs_regex} =
2266                                    $r->{$remote}->{ignore_refs_regex};
2267                        }
2268                }
2269        }
2270
2271        $r;
2272}
2273
2274sub init_vars {
2275        $_gc_nr = $_gc_period = 1000;
2276        if (defined $_repack || defined $_repack_flags) {
2277               warn "Repack options are obsolete; they have no effect.\n";
2278        }
2279}
2280
2281sub verify_remotes_sanity {
2282        return unless -d $ENV{GIT_DIR};
2283        my %seen;
2284        foreach (command(qw/config -l/)) {
2285                if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
2286                        if ($seen{$1}) {
2287                                die "Remote ref refs/remote/$1 is tracked by",
2288                                    "\n  \"$_\"\nand\n  \"$seen{$1}\"\n",
2289                                    "Please resolve this ambiguity in ",
2290                                    "your git configuration file before ",
2291                                    "continuing\n";
2292                        }
2293                        $seen{$1} = $_;
2294                }
2295        }
2296}
2297
2298sub find_existing_remote {
2299        my ($url, $remotes) = @_;
2300        return undef if $no_reuse_existing;
2301        my $existing;
2302        foreach my $repo_id (keys %$remotes) {
2303                my $u = $remotes->{$repo_id}->{url} or next;
2304                next if $u ne $url;
2305                $existing = $repo_id;
2306                last;
2307        }
2308        $existing;
2309}
2310
2311sub init_remote_config {
2312        my ($self, $url, $no_write) = @_;
2313        $url =~ s!/+$!!; # strip trailing slash
2314        my $r = read_all_remotes();
2315        my $existing = find_existing_remote($url, $r);
2316        if ($existing) {
2317                unless ($no_write) {
2318                        print STDERR "Using existing ",
2319                                     "[svn-remote \"$existing\"]\n";
2320                }
2321                $self->{repo_id} = $existing;
2322        } elsif ($_minimize_url) {
2323                my $min_url = Git::SVN::Ra->new($url)->minimize_url;
2324                $existing = find_existing_remote($min_url, $r);
2325                if ($existing) {
2326                        unless ($no_write) {
2327                                print STDERR "Using existing ",
2328                                             "[svn-remote \"$existing\"]\n";
2329                        }
2330                        $self->{repo_id} = $existing;
2331                }
2332                if ($min_url ne $url) {
2333                        unless ($no_write) {
2334                                print STDERR "Using higher level of URL: ",
2335                                             "$url => $min_url\n";
2336                        }
2337                        my $old_path = $self->{path};
2338                        $self->{path} = $url;
2339                        $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
2340                        if (length $old_path) {
2341                                $self->{path} .= "/$old_path";
2342                        }
2343                        $url = $min_url;
2344                }
2345        }
2346        my $orig_url;
2347        if (!$existing) {
2348                # verify that we aren't overwriting anything:
2349                $orig_url = eval {
2350                        command_oneline('config', '--get',
2351                                        "svn-remote.$self->{repo_id}.url")
2352                };
2353                if ($orig_url && ($orig_url ne $url)) {
2354                        die "svn-remote.$self->{repo_id}.url already set: ",
2355                            "$orig_url\nwanted to set to: $url\n";
2356                }
2357        }
2358        my ($xrepo_id, $xpath) = find_ref($self->refname);
2359        if (!$no_write && defined $xpath) {
2360                die "svn-remote.$xrepo_id.fetch already set to track ",
2361                    "$xpath:", $self->refname, "\n";
2362        }
2363        unless ($no_write) {
2364                command_noisy('config',
2365                              "svn-remote.$self->{repo_id}.url", $url);
2366                $self->{path} =~ s{^/}{};
2367                $self->{path} =~ s{%([0-9A-F]{2})}{chr hex($1)}ieg;
2368                command_noisy('config', '--add',
2369                              "svn-remote.$self->{repo_id}.fetch",
2370                              "$self->{path}:".$self->refname);
2371        }
2372        $self->{url} = $url;
2373}
2374
2375sub find_by_url { # repos_root and, path are optional
2376        my ($class, $full_url, $repos_root, $path) = @_;
2377
2378        return undef unless defined $full_url;
2379        remove_username($full_url);
2380        remove_username($repos_root) if defined $repos_root;
2381        my $remotes = read_all_remotes();
2382        if (defined $full_url && defined $repos_root && !defined $path) {
2383                $path = $full_url;
2384                $path =~ s#^\Q$repos_root\E(?:/|$)##;
2385        }
2386        foreach my $repo_id (keys %$remotes) {
2387                my $u = $remotes->{$repo_id}->{url} or next;
2388                remove_username($u);
2389                next if defined $repos_root && $repos_root ne $u;
2390
2391                my $fetch = $remotes->{$repo_id}->{fetch} || {};
2392                foreach my $t (qw/branches tags/) {
2393                        foreach my $globspec (@{$remotes->{$repo_id}->{$t}}) {
2394                                resolve_local_globs($u, $fetch, $globspec);
2395                        }
2396                }
2397                my $p = $path;
2398                my $rwr = rewrite_root({repo_id => $repo_id});
2399                my $svm = $remotes->{$repo_id}->{svm}
2400                        if defined $remotes->{$repo_id}->{svm};
2401                unless (defined $p) {
2402                        $p = $full_url;
2403                        my $z = $u;
2404                        my $prefix = '';
2405                        if ($rwr) {
2406                                $z = $rwr;
2407                                remove_username($z);
2408                        } elsif (defined $svm) {
2409                                $z = $svm->{source};
2410                                $prefix = $svm->{replace};
2411                                $prefix =~ s#^\Q$u\E(?:/|$)##;
2412                                $prefix =~ s#/$##;
2413                        }
2414                        $p =~ s#^\Q$z\E(?:/|$)#$prefix# or next;
2415                }
2416                foreach my $f (keys %$fetch) {
2417                        next if $f ne $p;
2418                        return Git::SVN->new($fetch->{$f}, $repo_id, $f);
2419                }
2420        }
2421        undef;
2422}
2423
2424sub init {
2425        my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
2426        my $self = _new($class, $repo_id, $ref_id, $path);
2427        if (defined $url) {
2428                $self->init_remote_config($url, $no_write);
2429        }
2430        $self;
2431}
2432
2433sub find_ref {
2434        my ($ref_id) = @_;
2435        foreach (command(qw/config -l/)) {
2436                next unless m!^svn-remote\.(.+)\.fetch=
2437                              \s*(.*?)\s*:\s*(.+?)\s*$!x;
2438                my ($repo_id, $path, $ref) = ($1, $2, $3);
2439                if ($ref eq $ref_id) {
2440                        $path = '' if ($path =~ m#^\./?#);
2441                        return ($repo_id, $path);
2442                }
2443        }
2444        (undef, undef, undef);
2445}
2446
2447sub new {
2448        my ($class, $ref_id, $repo_id, $path) = @_;
2449        if (defined $ref_id && !defined $repo_id && !defined $path) {
2450                ($repo_id, $path) = find_ref($ref_id);
2451                if (!defined $repo_id) {
2452                        die "Could not find a \"svn-remote.*.fetch\" key ",
2453                            "in the repository configuration matching: ",
2454                            "$ref_id\n";
2455                }
2456        }
2457        my $self = _new($class, $repo_id, $ref_id, $path);
2458        if (!defined $self->{path} || !length $self->{path}) {
2459                my $fetch = command_oneline('config', '--get',
2460                                            "svn-remote.$repo_id.fetch",
2461                                            ":$ref_id\$") or
2462                     die "Failed to read \"svn-remote.$repo_id.fetch\" ",
2463                         "\":$ref_id\$\" in config\n";
2464                ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
2465        }
2466        $self->{path} =~ s{/+}{/}g;
2467        $self->{path} =~ s{\A/}{};
2468        $self->{path} =~ s{/\z}{};
2469        $self->{url} = command_oneline('config', '--get',
2470                                       "svn-remote.$repo_id.url") or
2471                  die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
2472        $self->{pushurl} = eval { command_oneline('config', '--get',
2473                                  "svn-remote.$repo_id.pushurl") };
2474        $self->rebuild;
2475        $self;
2476}
2477
2478sub refname {
2479        my ($refname) = $_[0]->{ref_id} ;
2480
2481        # It cannot end with a slash /, we'll throw up on this because
2482        # SVN can't have directories with a slash in their name, either:
2483        if ($refname =~ m{/$}) {
2484                die "ref: '$refname' ends with a trailing slash, this is ",
2485                    "not permitted by git nor Subversion\n";
2486        }
2487
2488        # It cannot have ASCII control character space, tilde ~, caret ^,
2489        # colon :, question-mark ?, asterisk *, space, or open bracket [
2490        # anywhere.
2491        #
2492        # Additionally, % must be escaped because it is used for escaping
2493        # and we want our escaped refname to be reversible
2494        $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
2495
2496        # no slash-separated component can begin with a dot .
2497        # /.* becomes /%2E*
2498        $refname =~ s{/\.}{/%2E}g;
2499
2500        # It cannot have two consecutive dots .. anywhere
2501        # .. becomes %2E%2E
2502        $refname =~ s{\.\.}{%2E%2E}g;
2503
2504        # trailing dots and .lock are not allowed
2505        # .$ becomes %2E and .lock becomes %2Elock
2506        $refname =~ s{\.(?=$|lock$)}{%2E};
2507
2508        # the sequence @{ is used to access the reflog
2509        # @{ becomes %40{
2510        $refname =~ s{\@\{}{%40\{}g;
2511
2512        return $refname;
2513}
2514
2515sub desanitize_refname {
2516        my ($refname) = @_;
2517        $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
2518        return $refname;
2519}
2520
2521sub svm_uuid {
2522        my ($self) = @_;
2523        return $self->{svm}->{uuid} if $self->svm;
2524        $self->ra;
2525        unless ($self->{svm}) {
2526                die "SVM UUID not cached, and reading remotely failed\n";
2527        }
2528        $self->{svm}->{uuid};
2529}
2530
2531sub svm {
2532        my ($self) = @_;
2533        return $self->{svm} if $self->{svm};
2534        my $svm;
2535        # see if we have it in our config, first:
2536        eval {
2537                my $section = "svn-remote.$self->{repo_id}";
2538                $svm = {
2539                  source => tmp_config('--get', "$section.svm-source"),
2540                  uuid => tmp_config('--get', "$section.svm-uuid"),
2541                  replace => tmp_config('--get', "$section.svm-replace"),
2542                }
2543        };
2544        if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
2545                $self->{svm} = $svm;
2546        }
2547        $self->{svm};
2548}
2549
2550sub _set_svm_vars {
2551        my ($self, $ra) = @_;
2552        return $ra if $self->svm;
2553
2554        my @err = ( "useSvmProps set, but failed to read SVM properties\n",
2555                    "(svm:source, svm:uuid) ",
2556                    "from the following URLs:\n" );
2557        sub read_svm_props {
2558                my ($self, $ra, $path, $r) = @_;
2559                my $props = ($ra->get_dir($path, $r))[2];
2560                my $src = $props->{'svm:source'};
2561                my $uuid = $props->{'svm:uuid'};
2562                return undef if (!$src || !$uuid);
2563
2564                chomp($src, $uuid);
2565
2566                $uuid =~ m{^[0-9a-f\-]{30,}$}i
2567                    or die "doesn't look right - svm:uuid is '$uuid'\n";
2568
2569                # the '!' is used to mark the repos_root!/relative/path
2570                $src =~ s{/?!/?}{/};
2571                $src =~ s{/+$}{}; # no trailing slashes please
2572                # username is of no interest
2573                $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
2574
2575                my $replace = $ra->{url};
2576                $replace .= "/$path" if length $path;
2577
2578                my $section = "svn-remote.$self->{repo_id}";
2579                tmp_config("$section.svm-source", $src);
2580                tmp_config("$section.svm-replace", $replace);
2581                tmp_config("$section.svm-uuid", $uuid);
2582                $self->{svm} = {
2583                        source => $src,
2584                        uuid => $uuid,
2585                        replace => $replace
2586                };
2587        }
2588
2589        my $r = $ra->get_latest_revnum;
2590        my $path = $self->{path};
2591        my %tried;
2592        while (length $path) {
2593                unless ($tried{"$self->{url}/$path"}) {
2594                        return $ra if $self->read_svm_props($ra, $path, $r);
2595                        $tried{"$self->{url}/$path"} = 1;
2596                }
2597                $path =~ s#/?[^/]+$##;
2598        }
2599        die "Path: '$path' should be ''\n" if $path ne '';
2600        return $ra if $self->read_svm_props($ra, $path, $r);
2601        $tried{"$self->{url}/$path"} = 1;
2602
2603        if ($ra->{repos_root} eq $self->{url}) {
2604                die @err, (map { "  $_\n" } keys %tried), "\n";
2605        }
2606
2607        # nope, make sure we're connected to the repository root:
2608        my $ok;
2609        my @tried_b;
2610        $path = $ra->{svn_path};
2611        $ra = Git::SVN::Ra->new($ra->{repos_root});
2612        while (length $path) {
2613                unless ($tried{"$ra->{url}/$path"}) {
2614                        $ok = $self->read_svm_props($ra, $path, $r);
2615                        last if $ok;
2616                        $tried{"$ra->{url}/$path"} = 1;
2617                }
2618                $path =~ s#/?[^/]+$##;
2619        }
2620        die "Path: '$path' should be ''\n" if $path ne '';
2621        $ok ||= $self->read_svm_props($ra, $path, $r);
2622        $tried{"$ra->{url}/$path"} = 1;
2623        if (!$ok) {
2624                die @err, (map { "  $_\n" } keys %tried), "\n";
2625        }
2626        Git::SVN::Ra->new($self->{url});
2627}
2628
2629sub svnsync {
2630        my ($self) = @_;
2631        return $self->{svnsync} if $self->{svnsync};
2632
2633        if ($self->no_metadata) {
2634                die "Can't have both 'noMetadata' and ",
2635                    "'useSvnsyncProps' options set!\n";
2636        }
2637        if ($self->rewrite_root) {
2638                die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
2639                    "options set!\n";
2640        }
2641        if ($self->rewrite_uuid) {
2642                die "Can't have both 'useSvnsyncProps' and 'rewriteUUID' ",
2643                    "options set!\n";
2644        }
2645
2646        my $svnsync;
2647        # see if we have it in our config, first:
2648        eval {
2649                my $section = "svn-remote.$self->{repo_id}";
2650
2651                my $url = tmp_config('--get', "$section.svnsync-url");
2652                ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2653                   die "doesn't look right - svn:sync-from-url is '$url'\n";
2654
2655                my $uuid = tmp_config('--get', "$section.svnsync-uuid");
2656                ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2657                   die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2658
2659                $svnsync = { url => $url, uuid => $uuid }
2660        };
2661        if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
2662                return $self->{svnsync} = $svnsync;
2663        }
2664
2665        my $err = "useSvnsyncProps set, but failed to read " .
2666                  "svnsync property: svn:sync-from-";
2667        my $rp = $self->ra->rev_proplist(0);
2668
2669        my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
2670        ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2671                   die "doesn't look right - svn:sync-from-url is '$url'\n";
2672
2673        my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
2674        ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2675                   die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2676
2677        my $section = "svn-remote.$self->{repo_id}";
2678        tmp_config('--add', "$section.svnsync-uuid", $uuid);
2679        tmp_config('--add', "$section.svnsync-url", $url);
2680        return $self->{svnsync} = { url => $url, uuid => $uuid };
2681}
2682
2683# this allows us to memoize our SVN::Ra UUID locally and avoid a
2684# remote lookup (useful for 'git svn log').
2685sub ra_uuid {
2686        my ($self) = @_;
2687        unless ($self->{ra_uuid}) {
2688                my $key = "svn-remote.$self->{repo_id}.uuid";
2689                my $uuid = eval { tmp_config('--get', $key) };
2690                if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/i) {
2691                        $self->{ra_uuid} = $uuid;
2692                } else {
2693                        die "ra_uuid called without URL\n" unless $self->{url};
2694                        $self->{ra_uuid} = $self->ra->get_uuid;
2695                        tmp_config('--add', $key, $self->{ra_uuid});
2696                }
2697        }
2698        $self->{ra_uuid};
2699}
2700
2701sub _set_repos_root {
2702        my ($self, $repos_root) = @_;
2703        my $k = "svn-remote.$self->{repo_id}.reposRoot";
2704        $repos_root ||= $self->ra->{repos_root};
2705        tmp_config($k, $repos_root);
2706        $repos_root;
2707}
2708
2709sub repos_root {
2710        my ($self) = @_;
2711        my $k = "svn-remote.$self->{repo_id}.reposRoot";
2712        eval { tmp_config('--get', $k) } || $self->_set_repos_root;
2713}
2714
2715sub ra {
2716        my ($self) = shift;
2717        my $ra = Git::SVN::Ra->new($self->{url});
2718        $self->_set_repos_root($ra->{repos_root});
2719        if ($self->use_svm_props && !$self->{svm}) {
2720                if ($self->no_metadata) {
2721                        die "Can't have both 'noMetadata' and ",
2722                            "'useSvmProps' options set!\n";
2723                } elsif ($self->use_svnsync_props) {
2724                        die "Can't have both 'useSvnsyncProps' and ",
2725                            "'useSvmProps' options set!\n";
2726                }
2727                $ra = $self->_set_svm_vars($ra);
2728                $self->{-want_revprops} = 1;
2729        }
2730        $ra;
2731}
2732
2733# prop_walk(PATH, REV, SUB)
2734# -------------------------
2735# Recursively traverse PATH at revision REV and invoke SUB for each
2736# directory that contains a SVN property.  SUB will be invoked as
2737# follows:  &SUB(gs, path, props);  where `gs' is this instance of
2738# Git::SVN, `path' the path to the directory where the properties
2739# `props' were found.  The `path' will be relative to point of checkout,
2740# that is, if url://repo/trunk is the current Git branch, and that
2741# directory contains a sub-directory `d', SUB will be invoked with `/d/'
2742# as `path' (note the trailing `/').
2743sub prop_walk {
2744        my ($self, $path, $rev, $sub) = @_;
2745
2746        $path =~ s#^/##;
2747        my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev);
2748        $path =~ s#^/*#/#g;
2749        my $p = $path;
2750        # Strip the irrelevant part of the path.
2751        $p =~ s#^/+\Q$self->{path}\E(/|$)#/#;
2752        # Ensure the path is terminated by a `/'.
2753        $p =~ s#/*$#/#;
2754
2755        # The properties contain all the internal SVN stuff nobody
2756        # (usually) cares about.
2757        my $interesting_props = 0;
2758        foreach (keys %{$props}) {
2759                # If it doesn't start with `svn:', it must be a
2760                # user-defined property.
2761                ++$interesting_props and next if $_ !~ /^svn:/;
2762                # FIXME: Fragile, if SVN adds new public properties,
2763                # this needs to be updated.
2764                ++$interesting_props if /^svn:(?:ignore|keywords|executable
2765                                                 |eol-style|mime-type
2766                                                 |externals|needs-lock)$/x;
2767        }
2768        &$sub($self, $p, $props) if $interesting_props;
2769
2770        foreach (sort keys %$dirent) {
2771                next if $dirent->{$_}->{kind} != $SVN::Node::dir;
2772                $self->prop_walk($self->{path} . $p . $_, $rev, $sub);
2773        }
2774}
2775
2776sub last_rev { ($_[0]->last_rev_commit)[0] }
2777sub last_commit { ($_[0]->last_rev_commit)[1] }
2778
2779# returns the newest SVN revision number and newest commit SHA1
2780sub last_rev_commit {
2781        my ($self) = @_;
2782        if (defined $self->{last_rev} && defined $self->{last_commit}) {
2783                return ($self->{last_rev}, $self->{last_commit});
2784        }
2785        my $c = ::verify_ref($self->refname.'^0');
2786        if ($c && !$self->use_svm_props && !$self->no_metadata) {
2787                my $rev = (::cmt_metadata($c))[1];
2788                if (defined $rev) {
2789                        ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
2790                        return ($rev, $c);
2791                }
2792        }
2793        my $map_path = $self->map_path;
2794        unless (-e $map_path) {
2795                ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
2796                return (undef, undef);
2797        }
2798        my ($rev, $commit) = $self->rev_map_max(1);
2799        ($self->{last_rev}, $self->{last_commit}) = ($rev, $commit);
2800        return ($rev, $commit);
2801}
2802
2803sub get_fetch_range {
2804        my ($self, $min, $max) = @_;
2805        $max ||= $self->ra->get_latest_revnum;
2806        $min ||= $self->rev_map_max;
2807        (++$min, $max);
2808}
2809
2810sub tmp_config {
2811        my (@args) = @_;
2812        my $old_def_config = "$ENV{GIT_DIR}/svn/config";
2813        my $config = "$ENV{GIT_DIR}/svn/.metadata";
2814        if (! -f $config && -f $old_def_config) {
2815                rename $old_def_config, $config or
2816                       die "Failed rename $old_def_config => $config: $!\n";
2817        }
2818        my $old_config = $ENV{GIT_CONFIG};
2819        $ENV{GIT_CONFIG} = $config;
2820        $@ = undef;
2821        my @ret = eval {
2822                unless (-f $config) {
2823                        mkfile($config);
2824                        open my $fh, '>', $config or
2825                            die "Can't open $config: $!\n";
2826                        print $fh "; This file is used internally by ",
2827                                  "git-svn\n" or die
2828                                  "Couldn't write to $config: $!\n";
2829                        print $fh "; You should not have to edit it\n" or
2830                              die "Couldn't write to $config: $!\n";
2831                        close $fh or die "Couldn't close $config: $!\n";
2832                }
2833                command('config', @args);
2834        };
2835        my $err = $@;
2836        if (defined $old_config) {
2837                $ENV{GIT_CONFIG} = $old_config;
2838        } else {
2839                delete $ENV{GIT_CONFIG};
2840        }
2841        die $err if $err;
2842        wantarray ? @ret : $ret[0];
2843}
2844
2845sub tmp_index_do {
2846        my ($self, $sub) = @_;
2847        my $old_index = $ENV{GIT_INDEX_FILE};
2848        $ENV{GIT_INDEX_FILE} = $self->{index};
2849        $@ = undef;
2850        my @ret = eval {
2851                my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
2852                mkpath([$dir]) unless -d $dir;
2853                &$sub;
2854        };
2855        my $err = $@;
2856        if (defined $old_index) {
2857                $ENV{GIT_INDEX_FILE} = $old_index;
2858        } else {
2859                delete $ENV{GIT_INDEX_FILE};
2860        }
2861        die $err if $err;
2862        wantarray ? @ret : $ret[0];
2863}
2864
2865sub assert_index_clean {
2866        my ($self, $treeish) = @_;
2867
2868        $self->tmp_index_do(sub {
2869                command_noisy('read-tree', $treeish) unless -e $self->{index};
2870                my $x = command_oneline('write-tree');
2871                my ($y) = (command(qw/cat-file commit/, $treeish) =~
2872                           /^tree ($::sha1)/mo);
2873                return if $y eq $x;
2874
2875                warn "Index mismatch: $y != $x\nrereading $treeish\n";
2876                unlink $self->{index} or die "unlink $self->{index}: $!\n";
2877                command_noisy('read-tree', $treeish);
2878                $x = command_oneline('write-tree');
2879                if ($y ne $x) {
2880                        ::fatal "trees ($treeish) $y != $x\n",
2881                                "Something is seriously wrong...";
2882                }
2883        });
2884}
2885
2886sub get_commit_parents {
2887        my ($self, $log_entry) = @_;
2888        my (%seen, @ret, @tmp);
2889        # legacy support for 'set-tree'; this is only used by set_tree_cb:
2890        if (my $ip = $self->{inject_parents}) {
2891                if (my $commit = delete $ip->{$log_entry->{revision}}) {
2892                        push @tmp, $commit;
2893                }
2894        }
2895        if (my $cur = ::verify_ref($self->refname.'^0')) {
2896                push @tmp, $cur;
2897        }
2898        if (my $ipd = $self->{inject_parents_dcommit}) {
2899                if (my $commit = delete $ipd->{$log_entry->{revision}}) {
2900                        push @tmp, @$commit;
2901                }
2902        }
2903        push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
2904        while (my $p = shift @tmp) {
2905                next if $seen{$p};
2906                $seen{$p} = 1;
2907                push @ret, $p;
2908        }
2909        @ret;
2910}
2911
2912sub rewrite_root {
2913        my ($self) = @_;
2914        return $self->{-rewrite_root} if exists $self->{-rewrite_root};
2915        my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
2916        my $rwr = eval { command_oneline(qw/config --get/, $k) };
2917        if ($rwr) {
2918                $rwr =~ s#/+$##;
2919                if ($rwr !~ m#^[a-z\+]+://#) {
2920                        die "$rwr is not a valid URL (key: $k)\n";
2921                }
2922        }
2923        $self->{-rewrite_root} = $rwr;
2924}
2925
2926sub rewrite_uuid {
2927        my ($self) = @_;
2928        return $self->{-rewrite_uuid} if exists $self->{-rewrite_uuid};
2929        my $k = "svn-remote.$self->{repo_id}.rewriteUUID";
2930        my $rwid = eval { command_oneline(qw/config --get/, $k) };
2931        if ($rwid) {
2932                $rwid =~ s#/+$##;
2933                if ($rwid !~ m#^[a-f0-9]{8}-(?:[a-f0-9]{4}-){3}[a-f0-9]{12}$#) {
2934                        die "$rwid is not a valid UUID (key: $k)\n";
2935                }
2936        }
2937        $self->{-rewrite_uuid} = $rwid;
2938}
2939
2940sub metadata_url {
2941        my ($self) = @_;
2942        ($self->rewrite_root || $self->{url}) .
2943           (length $self->{path} ? '/' . $self->{path} : '');
2944}
2945
2946sub full_url {
2947        my ($self) = @_;
2948        $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
2949}
2950
2951sub full_pushurl {
2952        my ($self) = @_;
2953        if ($self->{pushurl}) {
2954                return $self->{pushurl} . (length $self->{path} ? '/' .
2955                       $self->{path} : '');
2956        } else {
2957                return $self->full_url;
2958        }
2959}
2960
2961sub set_commit_header_env {
2962        my ($log_entry) = @_;
2963        my %env;
2964        foreach my $ned (qw/NAME EMAIL DATE/) {
2965                foreach my $ac (qw/AUTHOR COMMITTER/) {
2966                        $env{"GIT_${ac}_${ned}"} = $ENV{"GIT_${ac}_${ned}"};
2967                }
2968        }
2969
2970        $ENV{GIT_AUTHOR_NAME} = $log_entry->{name};
2971        $ENV{GIT_AUTHOR_EMAIL} = $log_entry->{email};
2972        $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
2973
2974        $ENV{GIT_COMMITTER_NAME} = (defined $log_entry->{commit_name})
2975                                                ? $log_entry->{commit_name}
2976                                                : $log_entry->{name};
2977        $ENV{GIT_COMMITTER_EMAIL} = (defined $log_entry->{commit_email})
2978                                                ? $log_entry->{commit_email}
2979                                                : $log_entry->{email};
2980        \%env;
2981}
2982
2983sub restore_commit_header_env {
2984        my ($env) = @_;
2985        foreach my $ned (qw/NAME EMAIL DATE/) {
2986                foreach my $ac (qw/AUTHOR COMMITTER/) {
2987                        my $k = "GIT_${ac}_${ned}";
2988                        if (defined $env->{$k}) {
2989                                $ENV{$k} = $env->{$k};
2990                        } else {
2991                                delete $ENV{$k};
2992                        }
2993                }
2994        }
2995}
2996
2997sub gc {
2998        command_noisy('gc', '--auto');
2999};
3000
3001sub do_git_commit {
3002        my ($self, $log_entry) = @_;
3003        my $lr = $self->last_rev;
3004        if (defined $lr && $lr >= $log_entry->{revision}) {
3005                die "Last fetched revision of ", $self->refname,
3006                    " was r$lr, but we are about to fetch: ",
3007                    "r$log_entry->{revision}!\n";
3008        }
3009        if (my $c = $self->rev_map_get($log_entry->{revision})) {
3010                croak "$log_entry->{revision} = $c already exists! ",
3011                      "Why are we refetching it?\n";
3012        }
3013        my $old_env = set_commit_header_env($log_entry);
3014        my $tree = $log_entry->{tree};
3015        if (!defined $tree) {
3016                $tree = $self->tmp_index_do(sub {
3017                                            command_oneline('write-tree') });
3018        }
3019        die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
3020
3021        my @exec = ('git', 'commit-tree', $tree);
3022        foreach ($self->get_commit_parents($log_entry)) {
3023                push @exec, '-p', $_;
3024        }
3025        defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
3026                                                                   or croak $!;
3027        binmode $msg_fh;
3028
3029        # we always get UTF-8 from SVN, but we may want our commits in
3030        # a different encoding.
3031        if (my $enc = Git::config('i18n.commitencoding')) {
3032                require Encode;
3033                Encode::from_to($log_entry->{log}, 'UTF-8', $enc);
3034        }
3035        print $msg_fh $log_entry->{log} or croak $!;
3036        restore_commit_header_env($old_env);
3037        unless ($self->no_metadata) {
3038                print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
3039                              or croak $!;
3040        }
3041        $msg_fh->flush == 0 or croak $!;
3042        close $msg_fh or croak $!;
3043        chomp(my $commit = do { local $/; <$out_fh> });
3044        close $out_fh or croak $!;
3045        waitpid $pid, 0;
3046        croak $? if $?;
3047        if ($commit !~ /^$::sha1$/o) {
3048                die "Failed to commit, invalid sha1: $commit\n";
3049        }
3050
3051        $self->rev_map_set($log_entry->{revision}, $commit, 1);
3052
3053        $self->{last_rev} = $log_entry->{revision};
3054        $self->{last_commit} = $commit;
3055        print "r$log_entry->{revision}" unless $::_q > 1;
3056        if (defined $log_entry->{svm_revision}) {
3057                 print " (\@$log_entry->{svm_revision})" unless $::_q > 1;
3058                 $self->rev_map_set($log_entry->{svm_revision}, $commit,
3059                                   0, $self->svm_uuid);
3060        }
3061        print " = $commit ($self->{ref_id})\n" unless $::_q > 1;
3062        if (--$_gc_nr == 0) {
3063                $_gc_nr = $_gc_period;
3064                gc();
3065        }
3066        return $commit;
3067}
3068
3069sub match_paths {
3070        my ($self, $paths, $r) = @_;
3071        return 1 if $self->{path} eq '';
3072        if (my $path = $paths->{"/$self->{path}"}) {
3073                return ($path->{action} eq 'D') ? 0 : 1;
3074        }
3075        $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
3076        if (grep /$self->{path_regex}/, keys %$paths) {
3077                return 1;
3078        }
3079        my $c = '';
3080        foreach (split m#/#, $self->{path}) {
3081                $c .= "/$_";
3082                next unless ($paths->{$c} &&
3083                             ($paths->{$c}->{action} =~ /^[AR]$/));
3084                if ($self->ra->check_path($self->{path}, $r) ==
3085                    $SVN::Node::dir) {
3086                        return 1;
3087                }
3088        }
3089        return 0;
3090}
3091
3092sub find_parent_branch {
3093        my ($self, $paths, $rev) = @_;
3094        return undef unless $self->follow_parent;
3095        unless (defined $paths) {
3096                my $err_handler = $SVN::Error::handler;
3097                $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
3098                $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1,
3099                                   sub { $paths = $_[0] });
3100                $SVN::Error::handler = $err_handler;
3101        }
3102        return undef unless defined $paths;
3103
3104        # look for a parent from another branch:
3105        my @b_path_components = split m#/#, $self->{path};
3106        my @a_path_components;
3107        my $i;
3108        while (@b_path_components) {
3109                $i = $paths->{'/'.join('/', @b_path_components)};
3110                last if $i && defined $i->{copyfrom_path};
3111                unshift(@a_path_components, pop(@b_path_components));
3112        }
3113        return undef unless defined $i && defined $i->{copyfrom_path};
3114        my $branch_from = $i->{copyfrom_path};
3115        if (@a_path_components) {
3116                print STDERR "branch_from: $branch_from => ";
3117                $branch_from .= '/'.join('/', @a_path_components);
3118                print STDERR $branch_from, "\n";
3119        }
3120        my $r = $i->{copyfrom_rev};
3121        my $repos_root = $self->ra->{repos_root};
3122        my $url = $self->ra->{url};
3123        my $new_url = $url . $branch_from;
3124        print STDERR  "Found possible branch point: ",
3125                      "$new_url => ", $self->full_url, ", $r\n"
3126                      unless $::_q > 1;
3127        $branch_from =~ s#^/##;
3128        my $gs = $self->other_gs($new_url, $url,
3129                                 $branch_from, $r, $self->{ref_id});
3130        my ($r0, $parent) = $gs->find_rev_before($r, 1);
3131        {
3132                my ($base, $head);
3133                if (!defined $r0 || !defined $parent) {
3134                        ($base, $head) = parse_revision_argument(0, $r);
3135                } else {
3136                        if ($r0 < $r) {
3137                                $gs->ra->get_log([$gs->{path}], $r0 + 1, $r, 1,
3138                                        0, 1, sub { $base = $_[1] - 1 });
3139                        }
3140                }
3141                if (defined $base && $base <= $r) {
3142                        $gs->fetch($base, $r);
3143                }
3144                ($r0, $parent) = $gs->find_rev_before($r, 1);
3145        }
3146        if (defined $r0 && defined $parent) {
3147                print STDERR "Found branch parent: ($self->{ref_id}) $parent\n"
3148                             unless $::_q > 1;
3149                my $ed;
3150                if ($self->ra->can_do_switch) {
3151                        $self->assert_index_clean($parent);
3152                        print STDERR "Following parent with do_switch\n"
3153                                     unless $::_q > 1;
3154                        # do_switch works with svn/trunk >= r22312, but that
3155                        # is not included with SVN 1.4.3 (the latest version
3156                        # at the moment), so we can't rely on it
3157                        $self->{last_rev} = $r0;
3158                        $self->{last_commit} = $parent;
3159                        $ed = SVN::Git::Fetcher->new($self, $gs->{path});
3160                        $gs->ra->gs_do_switch($r0, $rev, $gs,
3161                                              $self->full_url, $ed)
3162                          or die "SVN connection failed somewhere...\n";
3163                } elsif ($self->ra->trees_match($new_url, $r0,
3164                                                $self->full_url, $rev)) {
3165                        print STDERR "Trees match:\n",
3166                                     "  $new_url\@$r0\n",
3167                                     "  ${\$self->full_url}\@$rev\n",
3168                                     "Following parent with no changes\n"
3169                                     unless $::_q > 1;
3170                        $self->tmp_index_do(sub {
3171                            command_noisy('read-tree', $parent);
3172                        });
3173                        $self->{last_commit} = $parent;
3174                } else {
3175                        print STDERR "Following parent with do_update\n"
3176                                     unless $::_q > 1;
3177                        $ed = SVN::Git::Fetcher->new($self);
3178                        $self->ra->gs_do_update($rev, $rev, $self, $ed)
3179                          or die "SVN connection failed somewhere...\n";
3180                }
3181                print STDERR "Successfully followed parent\n" unless $::_q > 1;
3182                return $self->make_log_entry($rev, [$parent], $ed);
3183        }
3184        return undef;
3185}
3186
3187sub do_fetch {
3188        my ($self, $paths, $rev) = @_;
3189        my $ed;
3190        my ($last_rev, @parents);
3191        if (my $lc = $self->last_commit) {
3192                # we can have a branch that was deleted, then re-added
3193                # under the same name but copied from another path, in
3194                # which case we'll have multiple parents (we don't
3195                # want to break the original ref, nor lose copypath info):
3196                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
3197                        push @{$log_entry->{parents}}, $lc;
3198                        return $log_entry;
3199                }
3200                $ed = SVN::Git::Fetcher->new($self);
3201                $last_rev = $self->{last_rev};
3202                $ed->{c} = $lc;
3203                @parents = ($lc);
3204        } else {
3205                $last_rev = $rev;
3206                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
3207                        return $log_entry;
3208                }
3209                $ed = SVN::Git::Fetcher->new($self);
3210        }
3211        unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
3212                die "SVN connection failed somewhere...\n";
3213        }
3214        $self->make_log_entry($rev, \@parents, $ed);
3215}
3216
3217sub mkemptydirs {
3218        my ($self, $r) = @_;
3219
3220        sub scan {
3221                my ($r, $empty_dirs, $line) = @_;
3222                if (defined $r && $line =~ /^r(\d+)$/) {
3223                        return 0 if $1 > $r;
3224                } elsif ($line =~ /^  \+empty_dir: (.+)$/) {
3225                        $empty_dirs->{$1} = 1;
3226                } elsif ($line =~ /^  \-empty_dir: (.+)$/) {
3227                        my @d = grep {m[^\Q$1\E(/|$)]} (keys %$empty_dirs);
3228                        delete @$empty_dirs{@d};
3229                }
3230                1; # continue
3231        };
3232
3233        my %empty_dirs = ();
3234        my $gz_file = "$self->{dir}/unhandled.log.gz";
3235        if (-f $gz_file) {
3236                if (!$can_compress) {
3237                        warn "Compress::Zlib could not be found; ",
3238                             "empty directories in $gz_file will not be read\n";
3239                } else {
3240                        my $gz = Compress::Zlib::gzopen($gz_file, "rb") or
3241                                die "Unable to open $gz_file: $!\n";
3242                        my $line;
3243                        while ($gz->gzreadline($line) > 0) {
3244                                scan($r, \%empty_dirs, $line) or last;
3245                        }
3246                        $gz->gzclose;
3247                }
3248        }
3249
3250        if (open my $fh, '<', "$self->{dir}/unhandled.log") {
3251                binmode $fh or croak "binmode: $!";
3252                while (<$fh>) {
3253                        scan($r, \%empty_dirs, $_) or last;
3254                }
3255                close $fh;
3256        }
3257
3258        my $strip = qr/\A\Q$self->{path}\E(?:\/|$)/;
3259        foreach my $d (sort keys %empty_dirs) {
3260                $d = uri_decode($d);
3261                $d =~ s/$strip//;
3262                next unless length($d);
3263                next if -d $d;
3264                if (-e $d) {
3265                        warn "$d exists but is not a directory\n";
3266                } else {
3267                        print "creating empty directory: $d\n";
3268                        mkpath([$d]);
3269                }
3270        }
3271}
3272
3273sub get_untracked {
3274        my ($self, $ed) = @_;
3275        my @out;
3276        my $h = $ed->{empty};
3277        foreach (sort keys %$h) {
3278                my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
3279                push @out, "  $act: " . uri_encode($_);
3280                warn "W: $act: $_\n";
3281        }
3282        foreach my $t (qw/dir_prop file_prop/) {
3283                $h = $ed->{$t} or next;
3284                foreach my $path (sort keys %$h) {
3285                        my $ppath = $path eq '' ? '.' : $path;
3286                        foreach my $prop (sort keys %{$h->{$path}}) {
3287                                next if $SKIP_PROP{$prop};
3288                                my $v = $h->{$path}->{$prop};
3289                                my $t_ppath_prop = "$t: " .
3290                                                    uri_encode($ppath) . ' ' .
3291                                                    uri_encode($prop);
3292                                if (defined $v) {
3293                                        push @out, "  +$t_ppath_prop " .
3294                                                   uri_encode($v);
3295                                } else {
3296                                        push @out, "  -$t_ppath_prop";
3297                                }
3298                        }
3299                }
3300        }
3301        foreach my $t (qw/absent_file absent_directory/) {
3302                $h = $ed->{$t} or next;
3303                foreach my $parent (sort keys %$h) {
3304                        foreach my $path (sort @{$h->{$parent}}) {
3305                                push @out, "  $t: " .
3306                                           uri_encode("$parent/$path");
3307                                warn "W: $t: $parent/$path ",
3308                                     "Insufficient permissions?\n";
3309                        }
3310                }
3311        }
3312        \@out;
3313}
3314
3315sub get_tz {
3316        # some systmes don't handle or mishandle %z, so be creative.
3317        my $t = shift || time;
3318        my $gm = timelocal(gmtime($t));
3319        my $sign = qw( + + - )[ $t <=> $gm ];
3320        return sprintf("%s%02d%02d", $sign, (gmtime(abs($t - $gm)))[2,1]);
3321}
3322
3323# parse_svn_date(DATE)
3324# --------------------
3325# Given a date (in UTC) from Subversion, return a string in the format
3326# "<TZ Offset> <local date/time>" that Git will use.
3327#
3328# By default the parsed date will be in UTC; if $Git::SVN::_localtime
3329# is true we'll convert it to the local timezone instead.
3330sub parse_svn_date {
3331        my $date = shift || return '+0000 1970-01-01 00:00:00';
3332        my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
3333                                            (\d\d)\:(\d\d)\:(\d\d)\.\d*Z$/x) or
3334                                         croak "Unable to parse date: $date\n";
3335        my $parsed_date;    # Set next.
3336
3337        if ($Git::SVN::_localtime) {
3338                # Translate the Subversion datetime to an epoch time.
3339                # Begin by switching ourselves to $date's timezone, UTC.
3340                my $old_env_TZ = $ENV{TZ};
3341                $ENV{TZ} = 'UTC';
3342
3343                my $epoch_in_UTC =
3344                    POSIX::strftime('%s', $S, $M, $H, $d, $m - 1, $Y - 1900);
3345
3346                # Determine our local timezone (including DST) at the
3347                # time of $epoch_in_UTC.  $Git::SVN::Log::TZ stored the
3348                # value of TZ, if any, at the time we were run.
3349                if (defined $Git::SVN::Log::TZ) {
3350                        $ENV{TZ} = $Git::SVN::Log::TZ;
3351                } else {
3352                        delete $ENV{TZ};
3353                }
3354
3355                my $our_TZ = get_tz();
3356
3357                # This converts $epoch_in_UTC into our local timezone.
3358                my ($sec, $min, $hour, $mday, $mon, $year,
3359                    $wday, $yday, $isdst) = localtime($epoch_in_UTC);
3360
3361                $parsed_date = sprintf('%s %04d-%02d-%02d %02d:%02d:%02d',
3362                                       $our_TZ, $year + 1900, $mon + 1,
3363                                       $mday, $hour, $min, $sec);
3364
3365                # Reset us to the timezone in effect when we entered
3366                # this routine.
3367                if (defined $old_env_TZ) {
3368                        $ENV{TZ} = $old_env_TZ;
3369                } else {
3370                        delete $ENV{TZ};
3371                }
3372        } else {
3373                $parsed_date = "+0000 $Y-$m-$d $H:$M:$S";
3374        }
3375
3376        return $parsed_date;
3377}
3378
3379sub other_gs {
3380        my ($self, $new_url, $url,
3381            $branch_from, $r, $old_ref_id) = @_;
3382        my $gs = Git::SVN->find_by_url($new_url, $url, $branch_from);
3383        unless ($gs) {
3384                my $ref_id = $old_ref_id;
3385                $ref_id =~ s/\@\d+-*$//;
3386                $ref_id .= "\@$r";
3387                # just grow a tail if we're not unique enough :x
3388                $ref_id .= '-' while find_ref($ref_id);
3389                my ($u, $p, $repo_id) = ($new_url, '', $ref_id);
3390                if ($u =~ s#^\Q$url\E(/|$)##) {
3391                        $p = $u;
3392                        $u = $url;
3393                        $repo_id = $self->{repo_id};
3394                }
3395                while (1) {
3396                        # It is possible to tag two different subdirectories at
3397                        # the same revision.  If the url for an existing ref
3398                        # does not match, we must either find a ref with a
3399                        # matching url or create a new ref by growing a tail.
3400                        $gs = Git::SVN->init($u, $p, $repo_id, $ref_id, 1);
3401                        my (undef, $max_commit) = $gs->rev_map_max(1);
3402                        last if (!$max_commit);
3403                        my ($url) = ::cmt_metadata($max_commit);
3404                        last if ($url eq $gs->metadata_url);
3405                        $ref_id .= '-';
3406                }
3407                print STDERR "Initializing parent: $ref_id\n" unless $::_q > 1;
3408        }
3409        $gs
3410}
3411
3412sub call_authors_prog {
3413        my ($orig_author) = @_;
3414        $orig_author = command_oneline('rev-parse', '--sq-quote', $orig_author);
3415        my $author = `$::_authors_prog $orig_author`;
3416        if ($? != 0) {
3417                die "$::_authors_prog failed with exit code $?\n"
3418        }
3419        if ($author =~ /^\s*(.+?)\s*<(.*)>\s*$/) {
3420                my ($name, $email) = ($1, $2);
3421                $email = undef if length $2 == 0;
3422                return [$name, $email];
3423        } else {
3424                die "Author: $orig_author: $::_authors_prog returned "
3425                        . "invalid author format: $author\n";
3426        }
3427}
3428
3429sub check_author {
3430        my ($author) = @_;
3431        if (!defined $author || length $author == 0) {
3432                $author = '(no author)';
3433        }
3434        if (!defined $::users{$author}) {
3435                if (defined $::_authors_prog) {
3436                        $::users{$author} = call_authors_prog($author);
3437                } elsif (defined $::_authors) {
3438                        die "Author: $author not defined in $::_authors file\n";
3439                }
3440        }
3441        $author;
3442}
3443
3444sub find_extra_svk_parents {
3445        my ($self, $ed, $tickets, $parents) = @_;
3446        # aha!  svk:merge property changed...
3447        my @tickets = split "\n", $tickets;
3448        my @known_parents;
3449        for my $ticket ( @tickets ) {
3450                my ($uuid, $path, $rev) = split /:/, $ticket;
3451                if ( $uuid eq $self->ra_uuid ) {
3452                        my $url = $self->{url};
3453                        my $repos_root = $url;
3454                        my $branch_from = $path;
3455                        $branch_from =~ s{^/}{};
3456                        my $gs = $self->other_gs($repos_root."/".$branch_from,
3457                                                 $url,
3458                                                 $branch_from,
3459                                                 $rev,
3460                                                 $self->{ref_id});
3461                        if ( my $commit = $gs->rev_map_get($rev, $uuid) ) {
3462                                # wahey!  we found it, but it might be
3463                                # an old one (!)
3464                                push @known_parents, [ $rev, $commit ];
3465                        }
3466                }
3467        }
3468        # Ordering matters; highest-numbered commit merge tickets
3469        # first, as they may account for later merge ticket additions
3470        # or changes.
3471        @known_parents = map {$_->[1]} sort {$b->[0] <=> $a->[0]} @known_parents;
3472        for my $parent ( @known_parents ) {
3473                my @cmd = ('rev-list', $parent, map { "^$_" } @$parents );
3474                my ($msg_fh, $ctx) = command_output_pipe(@cmd);
3475                my $new;
3476                while ( <$msg_fh> ) {
3477                        $new=1;last;
3478                }
3479                command_close_pipe($msg_fh, $ctx);
3480                if ( $new ) {
3481                        print STDERR
3482                            "Found merge parent (svk:merge ticket): $parent\n";
3483                        push @$parents, $parent;
3484                }
3485        }
3486}
3487
3488sub lookup_svn_merge {
3489        my $uuid = shift;
3490        my $url = shift;
3491        my $merge = shift;
3492
3493        my ($source, $revs) = split ":", $merge;
3494        my $path = $source;
3495        $path =~ s{^/}{};
3496        my $gs = Git::SVN->find_by_url($url.$source, $url, $path);
3497        if ( !$gs ) {
3498                warn "Couldn't find revmap for $url$source\n";
3499                return;
3500        }
3501        my @ranges = split ",", $revs;
3502        my ($tip, $tip_commit);
3503        my @merged_commit_ranges;
3504        # find the tip
3505        for my $range ( @ranges ) {
3506                my ($bottom, $top) = split "-", $range;
3507                $top ||= $bottom;
3508                my $bottom_commit = $gs->find_rev_after( $bottom, 1, $top );
3509                my $top_commit = $gs->find_rev_before( $top, 1, $bottom );
3510
3511                unless ($top_commit and $bottom_commit) {
3512                        warn "W:unknown path/rev in svn:mergeinfo "
3513                                ."dirprop: $source:$range\n";
3514                        next;
3515                }
3516
3517                if (scalar(command('rev-parse', "$bottom_commit^@"))) {
3518                        push @merged_commit_ranges,
3519                             "$bottom_commit^..$top_commit";
3520                } else {
3521                        push @merged_commit_ranges, "$top_commit";
3522                }
3523
3524                if ( !defined $tip or $top > $tip ) {
3525                        $tip = $top;
3526                        $tip_commit = $top_commit;
3527                }
3528        }
3529        return ($tip_commit, @merged_commit_ranges);
3530}
3531
3532sub _rev_list {
3533        my ($msg_fh, $ctx) = command_output_pipe(
3534                "rev-list", @_,
3535               );
3536        my @rv;
3537        while ( <$msg_fh> ) {
3538                chomp;
3539                push @rv, $_;
3540        }
3541        command_close_pipe($msg_fh, $ctx);
3542        @rv;
3543}
3544
3545sub check_cherry_pick {
3546        my $base = shift;
3547        my $tip = shift;
3548        my $parents = shift;
3549        my @ranges = @_;
3550        my %commits = map { $_ => 1 }
3551                _rev_list("--no-merges", $tip, "--not", $base, @$parents, "--");
3552        for my $range ( @ranges ) {
3553                delete @commits{_rev_list($range, "--")};
3554        }
3555        for my $commit (keys %commits) {
3556                if (has_no_changes($commit)) {
3557                        delete $commits{$commit};
3558                }
3559        }
3560        return (keys %commits);
3561}
3562
3563sub has_no_changes {
3564        my $commit = shift;
3565
3566        my @revs = split / /, command_oneline(
3567                qw(rev-list --parents -1 -m), $commit);
3568
3569        # Commits with no parents, e.g. the start of a partial branch,
3570        # have changes by definition.
3571        return 1 if (@revs < 2);
3572
3573        # Commits with multiple parents, e.g a merge, have no changes
3574        # by definition.
3575        return 0 if (@revs > 2);
3576
3577        return (command_oneline("rev-parse", "$commit^{tree}") eq
3578                command_oneline("rev-parse", "$commit~1^{tree}"));
3579}
3580
3581# The GIT_DIR environment variable is not always set until after the command
3582# line arguments are processed, so we can't memoize in a BEGIN block.
3583{
3584        my $memoized = 0;
3585
3586        sub memoize_svn_mergeinfo_functions {
3587                return if $memoized;
3588                $memoized = 1;
3589
3590                my $cache_path = "$ENV{GIT_DIR}/svn/.caches/";
3591                mkpath([$cache_path]) unless -d $cache_path;
3592
3593                tie my %lookup_svn_merge_cache => 'Memoize::Storable',
3594                    "$cache_path/lookup_svn_merge.db", 'nstore';
3595                memoize 'lookup_svn_merge',
3596                        SCALAR_CACHE => 'FAULT',
3597                        LIST_CACHE => ['HASH' => \%lookup_svn_merge_cache],
3598                ;
3599
3600                tie my %check_cherry_pick_cache => 'Memoize::Storable',
3601                    "$cache_path/check_cherry_pick.db", 'nstore';
3602                memoize 'check_cherry_pick',
3603                        SCALAR_CACHE => 'FAULT',
3604                        LIST_CACHE => ['HASH' => \%check_cherry_pick_cache],
3605                ;
3606
3607                tie my %has_no_changes_cache => 'Memoize::Storable',
3608                    "$cache_path/has_no_changes.db", 'nstore';
3609                memoize 'has_no_changes',
3610                        SCALAR_CACHE => ['HASH' => \%has_no_changes_cache],
3611                        LIST_CACHE => 'FAULT',
3612                ;
3613        }
3614
3615        sub unmemoize_svn_mergeinfo_functions {
3616                return if not $memoized;
3617                $memoized = 0;
3618
3619                Memoize::unmemoize 'lookup_svn_merge';
3620                Memoize::unmemoize 'check_cherry_pick';
3621                Memoize::unmemoize 'has_no_changes';
3622        }
3623
3624        Memoize::memoize 'Git::SVN::repos_root';
3625}
3626
3627END {
3628        # Force cache writeout explicitly instead of waiting for
3629        # global destruction to avoid segfault in Storable:
3630        # http://rt.cpan.org/Public/Bug/Display.html?id=36087
3631        unmemoize_svn_mergeinfo_functions();
3632}
3633
3634sub parents_exclude {
3635        my $parents = shift;
3636        my @commits = @_;
3637        return unless @commits;
3638
3639        my @excluded;
3640        my $excluded;
3641        do {
3642                my @cmd = ('rev-list', "-1", @commits, "--not", @$parents );
3643                $excluded = command_oneline(@cmd);
3644                if ( $excluded ) {
3645                        my @new;
3646                        my $found;
3647                        for my $commit ( @commits ) {
3648                                if ( $commit eq $excluded ) {
3649                                        push @excluded, $commit;
3650                                        $found++;
3651                                        last;
3652                                }
3653                                else {
3654                                        push @new, $commit;
3655                                }
3656                        }
3657                        die "saw commit '$excluded' in rev-list output, "
3658                                ."but we didn't ask for that commit (wanted: @commits --not @$parents)"
3659                                        unless $found;
3660                        @commits = @new;
3661                }
3662        }
3663                while ($excluded and @commits);
3664
3665        return @excluded;
3666}
3667
3668
3669# note: this function should only be called if the various dirprops
3670# have actually changed
3671sub find_extra_svn_parents {
3672        my ($self, $ed, $mergeinfo, $parents) = @_;
3673        # aha!  svk:merge property changed...
3674
3675        memoize_svn_mergeinfo_functions();
3676
3677        # We first search for merged tips which are not in our
3678        # history.  Then, we figure out which git revisions are in
3679        # that tip, but not this revision.  If all of those revisions
3680        # are now marked as merge, we can add the tip as a parent.
3681        my @merges = split "\n", $mergeinfo;
3682        my @merge_tips;
3683        my $url = $self->{url};
3684        my $uuid = $self->ra_uuid;
3685        my %ranges;
3686        for my $merge ( @merges ) {
3687                my ($tip_commit, @ranges) =
3688                        lookup_svn_merge( $uuid, $url, $merge );
3689                unless (!$tip_commit or
3690                                grep { $_ eq $tip_commit } @$parents ) {
3691                        push @merge_tips, $tip_commit;
3692                        $ranges{$tip_commit} = \@ranges;
3693                } else {
3694                        push @merge_tips, undef;
3695                }
3696        }
3697
3698        my %excluded = map { $_ => 1 }
3699                parents_exclude($parents, grep { defined } @merge_tips);
3700
3701        # check merge tips for new parents
3702        my @new_parents;
3703        for my $merge_tip ( @merge_tips ) {
3704                my $spec = shift @merges;
3705                next unless $merge_tip and $excluded{$merge_tip};
3706
3707                my $ranges = $ranges{$merge_tip};
3708
3709                # check out 'new' tips
3710                my $merge_base;
3711                eval {
3712                        $merge_base = command_oneline(
3713                                "merge-base",
3714                                @$parents, $merge_tip,
3715                        );
3716                };
3717                if ($@) {
3718                        die "An error occurred during merge-base"
3719                                unless $@->isa("Git::Error::Command");
3720
3721                        warn "W: Cannot find common ancestor between ".
3722                             "@$parents and $merge_tip. Ignoring merge info.\n";
3723                        next;
3724                }
3725
3726                # double check that there are no missing non-merge commits
3727                my (@incomplete) = check_cherry_pick(
3728                        $merge_base, $merge_tip,
3729                        $parents,
3730                        @$ranges,
3731                       );
3732
3733                if ( @incomplete ) {
3734                        warn "W:svn cherry-pick ignored ($spec) - missing "
3735                                .@incomplete." commit(s) (eg $incomplete[0])\n";
3736                } else {
3737                        warn
3738                                "Found merge parent (svn:mergeinfo prop): ",
3739                                        $merge_tip, "\n";
3740                        push @new_parents, $merge_tip;
3741                }
3742        }
3743
3744        # cater for merges which merge commits from multiple branches
3745        if ( @new_parents > 1 ) {
3746                for ( my $i = 0; $i <= $#new_parents; $i++ ) {
3747                        for ( my $j = 0; $j <= $#new_parents; $j++ ) {
3748                                next if $i == $j;
3749                                next unless $new_parents[$i];
3750                                next unless $new_parents[$j];
3751                                my $revs = command_oneline(
3752                                        "rev-list", "-1",
3753                                        "$new_parents[$i]..$new_parents[$j]",
3754                                       );
3755                                if ( !$revs ) {
3756                                        undef($new_parents[$j]);
3757                                }
3758                        }
3759                }
3760        }
3761        push @$parents, grep { defined } @new_parents;
3762}
3763
3764sub make_log_entry {
3765        my ($self, $rev, $parents, $ed) = @_;
3766        my $untracked = $self->get_untracked($ed);
3767
3768        my @parents = @$parents;
3769        my $ps = $ed->{path_strip} || "";
3770        for my $path ( grep { m/$ps/ } %{$ed->{dir_prop}} ) {
3771                my $props = $ed->{dir_prop}{$path};
3772                if ( $props->{"svk:merge"} ) {
3773                        $self->find_extra_svk_parents
3774                                ($ed, $props->{"svk:merge"}, \@parents);
3775                }
3776                if ( $props->{"svn:mergeinfo"} ) {
3777                        $self->find_extra_svn_parents
3778                                ($ed,
3779                                 $props->{"svn:mergeinfo"},
3780                                 \@parents);
3781                }
3782        }
3783
3784        open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
3785        print $un "r$rev\n" or croak $!;
3786        print $un $_, "\n" foreach @$untracked;
3787        my %log_entry = ( parents => \@parents, revision => $rev,
3788                          log => '');
3789
3790        my $headrev;
3791        my $logged = delete $self->{logged_rev_props};
3792        if (!$logged || $self->{-want_revprops}) {
3793                my $rp = $self->ra->rev_proplist($rev);
3794                foreach (sort keys %$rp) {
3795                        my $v = $rp->{$_};
3796                        if (/^svn:(author|date|log)$/) {
3797                                $log_entry{$1} = $v;
3798                        } elsif ($_ eq 'svm:headrev') {
3799                                $headrev = $v;
3800                        } else {
3801                                print $un "  rev_prop: ", uri_encode($_), ' ',
3802                                          uri_encode($v), "\n";
3803                        }
3804                }
3805        } else {
3806                map { $log_entry{$_} = $logged->{$_} } keys %$logged;
3807        }
3808        close $un or croak $!;
3809
3810        $log_entry{date} = parse_svn_date($log_entry{date});
3811        $log_entry{log} .= "\n";
3812        my $author = $log_entry{author} = check_author($log_entry{author});
3813        my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
3814                                                       : ($author, undef);
3815
3816        my ($commit_name, $commit_email) = ($name, $email);
3817        if ($_use_log_author) {
3818                my $name_field;
3819                if ($log_entry{log} =~ /From:\s+(.*\S)\s*\n/i) {
3820                        $name_field = $1;
3821                } elsif ($log_entry{log} =~ /Signed-off-by:\s+(.*\S)\s*\n/i) {
3822                        $name_field = $1;
3823                }
3824                if (!defined $name_field) {
3825                        if (!defined $email) {
3826                                $email = $name;
3827                        }
3828                } elsif ($name_field =~ /(.*?)\s+<(.*)>/) {
3829                        ($name, $email) = ($1, $2);
3830                } elsif ($name_field =~ /(.*)@/) {
3831                        ($name, $email) = ($1, $name_field);
3832                } else {
3833                        ($name, $email) = ($name_field, $name_field);
3834                }
3835        }
3836        if (defined $headrev && $self->use_svm_props) {
3837                if ($self->rewrite_root) {
3838                        die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
3839                            "options set!\n";
3840                }
3841                if ($self->rewrite_uuid) {
3842                        die "Can't have both 'useSvmProps' and 'rewriteUUID' ",
3843                            "options set!\n";
3844                }
3845                my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$}i;
3846                # we don't want "SVM: initializing mirror for junk" ...
3847                return undef if $r == 0;
3848                my $svm = $self->svm;
3849                if ($uuid ne $svm->{uuid}) {
3850                        die "UUID mismatch on SVM path:\n",
3851                            "expected: $svm->{uuid}\n",
3852                            "     got: $uuid\n";
3853                }
3854                my $full_url = $self->full_url;
3855                $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
3856                             die "Failed to replace '$svm->{replace}' with ",
3857                                 "'$svm->{source}' in $full_url\n";
3858                # throw away username for storing in records
3859                remove_username($full_url);
3860                $log_entry{metadata} = "$full_url\@$r $uuid";
3861                $log_entry{svm_revision} = $r;
3862                $email ||= "$author\@$uuid";
3863                $commit_email ||= "$author\@$uuid";
3864        } elsif ($self->use_svnsync_props) {
3865                my $full_url = $self->svnsync->{url};
3866                $full_url .= "/$self->{path}" if length $self->{path};
3867                remove_username($full_url);
3868                my $uuid = $self->svnsync->{uuid};
3869                $log_entry{metadata} = "$full_url\@$rev $uuid";
3870                $email ||= "$author\@$uuid";
3871                $commit_email ||= "$author\@$uuid";
3872        } else {
3873                my $url = $self->metadata_url;
3874                remove_username($url);
3875                my $uuid = $self->rewrite_uuid || $self->ra->get_uuid;
3876                $log_entry{metadata} = "$url\@$rev " . $uuid;
3877                $email ||= "$author\@" . $uuid;
3878                $commit_email ||= "$author\@" . $uuid;
3879        }
3880        $log_entry{name} = $name;
3881        $log_entry{email} = $email;
3882        $log_entry{commit_name} = $commit_name;
3883        $log_entry{commit_email} = $commit_email;
3884        \%log_entry;
3885}
3886
3887sub fetch {
3888        my ($self, $min_rev, $max_rev, @parents) = @_;
3889        my ($last_rev, $last_commit) = $self->last_rev_commit;
3890        my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
3891        $self->ra->gs_fetch_loop_common($base, $head, [$self]);
3892}
3893
3894sub set_tree_cb {
3895        my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
3896        $self->{inject_parents} = { $rev => $tree };
3897        $self->fetch(undef, undef);
3898}
3899
3900sub set_tree {
3901        my ($self, $tree) = (shift, shift);
3902        my $log_entry = ::get_commit_entry($tree);
3903        unless ($self->{last_rev}) {
3904                ::fatal("Must have an existing revision to commit");
3905        }
3906        my %ed_opts = ( r => $self->{last_rev},
3907                        log => $log_entry->{log},
3908                        ra => $self->ra,
3909                        tree_a => $self->{last_commit},
3910                        tree_b => $tree,
3911                        editor_cb => sub {
3912                               $self->set_tree_cb($log_entry, $tree, @_) },
3913                        svn_path => $self->{path} );
3914        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
3915                print "No changes\nr$self->{last_rev} = $tree\n";
3916        }
3917}
3918
3919sub rebuild_from_rev_db {
3920        my ($self, $path) = @_;
3921        my $r = -1;
3922        open my $fh, '<', $path or croak "open: $!";
3923        binmode $fh or croak "binmode: $!";
3924        while (<$fh>) {
3925                length($_) == 41 or croak "inconsistent size in ($_) != 41";
3926                chomp($_);
3927                ++$r;
3928                next if $_ eq ('0' x 40);
3929                $self->rev_map_set($r, $_);
3930                print "r$r = $_\n";
3931        }
3932        close $fh or croak "close: $!";
3933        unlink $path or croak "unlink: $!";
3934}
3935
3936sub rebuild {
3937        my ($self) = @_;
3938        my $map_path = $self->map_path;
3939        my $partial = (-e $map_path && ! -z $map_path);
3940        return unless ::verify_ref($self->refname.'^0');
3941        if (!$partial && ($self->use_svm_props || $self->no_metadata)) {
3942                my $rev_db = $self->rev_db_path;
3943                $self->rebuild_from_rev_db($rev_db);
3944                if ($self->use_svm_props) {
3945                        my $svm_rev_db = $self->rev_db_path($self->svm_uuid);
3946                        $self->rebuild_from_rev_db($svm_rev_db);
3947                }
3948                $self->unlink_rev_db_symlink;
3949                return;
3950        }
3951        print "Rebuilding $map_path ...\n" if (!$partial);
3952        my ($base_rev, $head) = ($partial ? $self->rev_map_max_norebuild(1) :
3953                (undef, undef));
3954        my ($log, $ctx) =
3955            command_output_pipe(qw/rev-list --pretty=raw --reverse/,
3956                                ($head ? "$head.." : "") . $self->refname,
3957                                '--');
3958        my $metadata_url = $self->metadata_url;
3959        remove_username($metadata_url);
3960        my $svn_uuid = $self->rewrite_uuid || $self->ra_uuid;
3961        my $c;
3962        while (<$log>) {
3963                if ( m{^commit ($::sha1)$} ) {
3964                        $c = $1;
3965                        next;
3966                }
3967                next unless s{^\s*(git-svn-id:)}{$1};
3968                my ($url, $rev, $uuid) = ::extract_metadata($_);
3969                remove_username($url);
3970
3971                # ignore merges (from set-tree)
3972                next if (!defined $rev || !$uuid);
3973
3974                # if we merged or otherwise started elsewhere, this is
3975                # how we break out of it
3976                if (($uuid ne $svn_uuid) ||
3977                    ($metadata_url && $url && ($url ne $metadata_url))) {
3978                        next;
3979                }
3980                if ($partial && $head) {
3981                        print "Partial-rebuilding $map_path ...\n";
3982                        print "Currently at $base_rev = $head\n";
3983                        $head = undef;
3984                }
3985
3986                $self->rev_map_set($rev, $c);
3987                print "r$rev = $c\n";
3988        }
3989        command_close_pipe($log, $ctx);
3990        print "Done rebuilding $map_path\n" if (!$partial || !$head);
3991        my $rev_db_path = $self->rev_db_path;
3992        if (-f $self->rev_db_path) {
3993                unlink $self->rev_db_path or croak "unlink: $!";
3994        }
3995        $self->unlink_rev_db_symlink;
3996}
3997
3998# rev_map:
3999# Tie::File seems to be prone to offset errors if revisions get sparse,
4000# it's not that fast, either.  Tie::File is also not in Perl 5.6.  So
4001# one of my favorite modules is out :<  Next up would be one of the DBM
4002# modules, but I'm not sure which is most portable...
4003#
4004# This is the replacement for the rev_db format, which was too big
4005# and inefficient for large repositories with a lot of sparse history
4006# (mainly tags)
4007#
4008# The format is this:
4009#   - 24 bytes for every record,
4010#     * 4 bytes for the integer representing an SVN revision number
4011#     * 20 bytes representing the sha1 of a git commit
4012#   - No empty padding records like the old format
4013#     (except the last record, which can be overwritten)
4014#   - new records are written append-only since SVN revision numbers
4015#     increase monotonically
4016#   - lookups on SVN revision number are done via a binary search
4017#   - Piping the file to xxd -c24 is a good way of dumping it for
4018#     viewing or editing (piped back through xxd -r), should the need
4019#     ever arise.
4020#   - The last record can be padding revision with an all-zero sha1
4021#     This is used to optimize fetch performance when using multiple
4022#     "fetch" directives in .git/config
4023#
4024# These files are disposable unless noMetadata or useSvmProps is set
4025
4026sub _rev_map_set {
4027        my ($fh, $rev, $commit) = @_;
4028
4029        binmode $fh or croak "binmode: $!";
4030        my $size = (stat($fh))[7];
4031        ($size % 24) == 0 or croak "inconsistent size: $size";
4032
4033        my $wr_offset = 0;
4034        if ($size > 0) {
4035                sysseek($fh, -24, SEEK_END) or croak "seek: $!";
4036                my $read = sysread($fh, my $buf, 24) or croak "read: $!";
4037                $read == 24 or croak "read only $read bytes (!= 24)";
4038                my ($last_rev, $last_commit) = unpack(rev_map_fmt, $buf);
4039                if ($last_commit eq ('0' x40)) {
4040                        if ($size >= 48) {
4041                                sysseek($fh, -48, SEEK_END) or croak "seek: $!";
4042                                $read = sysread($fh, $buf, 24) or
4043                                    croak "read: $!";
4044                                $read == 24 or
4045                                    croak "read only $read bytes (!= 24)";
4046                                ($last_rev, $last_commit) =
4047                                    unpack(rev_map_fmt, $buf);
4048                                if ($last_commit eq ('0' x40)) {
4049                                        croak "inconsistent .rev_map\n";
4050                                }
4051                        }
4052                        if ($last_rev >= $rev) {
4053                                croak "last_rev is higher!: $last_rev >= $rev";
4054                        }
4055                        $wr_offset = -24;
4056                }
4057        }
4058        sysseek($fh, $wr_offset, SEEK_END) or croak "seek: $!";
4059        syswrite($fh, pack(rev_map_fmt, $rev, $commit), 24) == 24 or
4060          croak "write: $!";
4061}
4062
4063sub _rev_map_reset {
4064        my ($fh, $rev, $commit) = @_;
4065        my $c = _rev_map_get($fh, $rev);
4066        $c eq $commit or die "_rev_map_reset(@_) commit $c does not match!\n";
4067        my $offset = sysseek($fh, 0, SEEK_CUR) or croak "seek: $!";
4068        truncate $fh, $offset or croak "truncate: $!";
4069}
4070
4071sub mkfile {
4072        my ($path) = @_;
4073        unless (-e $path) {
4074                my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
4075                mkpath([$dir]) unless -d $dir;
4076                open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
4077                close $fh or die "Couldn't close (create) $path: $!\n";
4078        }
4079}
4080
4081sub rev_map_set {
4082        my ($self, $rev, $commit, $update_ref, $uuid) = @_;
4083        defined $commit or die "missing arg3\n";
4084        length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
4085        my $db = $self->map_path($uuid);
4086        my $db_lock = "$db.lock";
4087        my $sigmask;
4088        $update_ref ||= 0;
4089        if ($update_ref) {
4090                $sigmask = POSIX::SigSet->new();
4091                my $signew = POSIX::SigSet->new(SIGINT, SIGHUP, SIGTERM,
4092                        SIGALRM, SIGUSR1, SIGUSR2);
4093                sigprocmask(SIG_BLOCK, $signew, $sigmask) or
4094                        croak "Can't block signals: $!";
4095        }
4096        mkfile($db);
4097
4098        $LOCKFILES{$db_lock} = 1;
4099        my $sync;
4100        # both of these options make our .rev_db file very, very important
4101        # and we can't afford to lose it because rebuild() won't work
4102        if ($self->use_svm_props || $self->no_metadata) {
4103                $sync = 1;
4104                copy($db, $db_lock) or die "rev_map_set(@_): ",
4105                                           "Failed to copy: ",
4106                                           "$db => $db_lock ($!)\n";
4107        } else {
4108                rename $db, $db_lock or die "rev_map_set(@_): ",
4109                                            "Failed to rename: ",
4110                                            "$db => $db_lock ($!)\n";
4111        }
4112
4113        sysopen(my $fh, $db_lock, O_RDWR | O_CREAT)
4114             or croak "Couldn't open $db_lock: $!\n";
4115        $update_ref eq 'reset' ? _rev_map_reset($fh, $rev, $commit) :
4116                                 _rev_map_set($fh, $rev, $commit);
4117        if ($sync) {
4118                $fh->flush or die "Couldn't flush $db_lock: $!\n";
4119                $fh->sync or die "Couldn't sync $db_lock: $!\n";
4120        }
4121        close $fh or croak $!;
4122        if ($update_ref) {
4123                $_head = $self;
4124                my $note = "";
4125                $note = " ($update_ref)" if ($update_ref !~ /^\d*$/);
4126                command_noisy('update-ref', '-m', "r$rev$note",
4127                              $self->refname, $commit);
4128        }
4129        rename $db_lock, $db or die "rev_map_set(@_): ", "Failed to rename: ",
4130                                    "$db_lock => $db ($!)\n";
4131        delete $LOCKFILES{$db_lock};
4132        if ($update_ref) {
4133                sigprocmask(SIG_SETMASK, $sigmask) or
4134                        croak "Can't restore signal mask: $!";
4135        }
4136}
4137
4138# If want_commit, this will return an array of (rev, commit) where
4139# commit _must_ be a valid commit in the archive.
4140# Otherwise, it'll return the max revision (whether or not the
4141# commit is valid or just a 0x40 placeholder).
4142sub rev_map_max {
4143        my ($self, $want_commit) = @_;
4144        $self->rebuild;
4145        my ($r, $c) = $self->rev_map_max_norebuild($want_commit);
4146        $want_commit ? ($r, $c) : $r;
4147}
4148
4149sub rev_map_max_norebuild {
4150        my ($self, $want_commit) = @_;
4151        my $map_path = $self->map_path;
4152        stat $map_path or return $want_commit ? (0, undef) : 0;
4153        sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
4154        binmode $fh or croak "binmode: $!";
4155        my $size = (stat($fh))[7];
4156        ($size % 24) == 0 or croak "inconsistent size: $size";
4157
4158        if ($size == 0) {
4159                close $fh or croak "close: $!";
4160                return $want_commit ? (0, undef) : 0;
4161        }
4162
4163        sysseek($fh, -24, SEEK_END) or croak "seek: $!";
4164        sysread($fh, my $buf, 24) == 24 or croak "read: $!";
4165        my ($r, $c) = unpack(rev_map_fmt, $buf);
4166        if ($want_commit && $c eq ('0' x40)) {
4167                if ($size < 48) {
4168                        return $want_commit ? (0, undef) : 0;
4169                }
4170                sysseek($fh, -48, SEEK_END) or croak "seek: $!";
4171                sysread($fh, $buf, 24) == 24 or croak "read: $!";
4172                ($r, $c) = unpack(rev_map_fmt, $buf);
4173                if ($c eq ('0'x40)) {
4174                        croak "Penultimate record is all-zeroes in $map_path";
4175                }
4176        }
4177        close $fh or croak "close: $!";
4178        $want_commit ? ($r, $c) : $r;
4179}
4180
4181sub rev_map_get {
4182        my ($self, $rev, $uuid) = @_;
4183        my $map_path = $self->map_path($uuid);
4184        return undef unless -e $map_path;
4185
4186        sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
4187        my $c = _rev_map_get($fh, $rev);
4188        close($fh) or croak "close: $!";
4189        $c
4190}
4191
4192sub _rev_map_get {
4193        my ($fh, $rev) = @_;
4194
4195        binmode $fh or croak "binmode: $!";
4196        my $size = (stat($fh))[7];
4197        ($size % 24) == 0 or croak "inconsistent size: $size";
4198
4199        if ($size == 0) {
4200                return undef;
4201        }
4202
4203        my ($l, $u) = (0, $size - 24);
4204        my ($r, $c, $buf);
4205
4206        while ($l <= $u) {
4207                my $i = int(($l/24 + $u/24) / 2) * 24;
4208                sysseek($fh, $i, SEEK_SET) or croak "seek: $!";
4209                sysread($fh, my $buf, 24) == 24 or croak "read: $!";
4210                my ($r, $c) = unpack(rev_map_fmt, $buf);
4211
4212                if ($r < $rev) {
4213                        $l = $i + 24;
4214                } elsif ($r > $rev) {
4215                        $u = $i - 24;
4216                } else { # $r == $rev
4217                        return $c eq ('0' x 40) ? undef : $c;
4218                }
4219        }
4220        undef;
4221}
4222
4223# Finds the first svn revision that exists on (if $eq_ok is true) or
4224# before $rev for the current branch.  It will not search any lower
4225# than $min_rev.  Returns the git commit hash and svn revision number
4226# if found, else (undef, undef).
4227sub find_rev_before {
4228        my ($self, $rev, $eq_ok, $min_rev) = @_;
4229        --$rev unless $eq_ok;
4230        $min_rev ||= 1;
4231        my $max_rev = $self->rev_map_max;
4232        $rev = $max_rev if ($rev > $max_rev);
4233        while ($rev >= $min_rev) {
4234                if (my $c = $self->rev_map_get($rev)) {
4235                        return ($rev, $c);
4236                }
4237                --$rev;
4238        }
4239        return (undef, undef);
4240}
4241
4242# Finds the first svn revision that exists on (if $eq_ok is true) or
4243# after $rev for the current branch.  It will not search any higher
4244# than $max_rev.  Returns the git commit hash and svn revision number
4245# if found, else (undef, undef).
4246sub find_rev_after {
4247        my ($self, $rev, $eq_ok, $max_rev) = @_;
4248        ++$rev unless $eq_ok;
4249        $max_rev ||= $self->rev_map_max;
4250        while ($rev <= $max_rev) {
4251                if (my $c = $self->rev_map_get($rev)) {
4252                        return ($rev, $c);
4253                }
4254                ++$rev;
4255        }
4256        return (undef, undef);
4257}
4258
4259sub _new {
4260        my ($class, $repo_id, $ref_id, $path) = @_;
4261        unless (defined $repo_id && length $repo_id) {
4262                $repo_id = $Git::SVN::default_repo_id;
4263        }
4264        unless (defined $ref_id && length $ref_id) {
4265                $_prefix = '' unless defined($_prefix);
4266                $_[2] = $ref_id =
4267                             "refs/remotes/$_prefix$Git::SVN::default_ref_id";
4268        }
4269        $_[1] = $repo_id;
4270        my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
4271
4272        # Older repos imported by us used $GIT_DIR/svn/foo instead of
4273        # $GIT_DIR/svn/refs/remotes/foo when tracking refs/remotes/foo
4274        if ($ref_id =~ m{^refs/remotes/(.*)}) {
4275                my $old_dir = "$ENV{GIT_DIR}/svn/$1";
4276                if (-d $old_dir && ! -d $dir) {
4277                        $dir = $old_dir;
4278                }
4279        }
4280
4281        $_[3] = $path = '' unless (defined $path);
4282        mkpath([$dir]);
4283        bless {
4284                ref_id => $ref_id, dir => $dir, index => "$dir/index",
4285                path => $path, config => "$ENV{GIT_DIR}/svn/config",
4286                map_root => "$dir/.rev_map", repo_id => $repo_id }, $class;
4287}
4288
4289# for read-only access of old .rev_db formats
4290sub unlink_rev_db_symlink {
4291        my ($self) = @_;
4292        my $link = $self->rev_db_path;
4293        $link =~ s/\.[\w-]+$// or croak "missing UUID at the end of $link";
4294        if (-l $link) {
4295                unlink $link or croak "unlink: $link failed!";
4296        }
4297}
4298
4299sub rev_db_path {
4300        my ($self, $uuid) = @_;
4301        my $db_path = $self->map_path($uuid);
4302        $db_path =~ s{/\.rev_map\.}{/\.rev_db\.}
4303            or croak "map_path: $db_path does not contain '/.rev_map.' !";
4304        $db_path;
4305}
4306
4307# the new replacement for .rev_db
4308sub map_path {
4309        my ($self, $uuid) = @_;
4310        $uuid ||= $self->ra_uuid;
4311        "$self->{map_root}.$uuid";
4312}
4313
4314sub uri_encode {
4315        my ($f) = @_;
4316        $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
4317        $f
4318}
4319
4320sub uri_decode {
4321        my ($f) = @_;
4322        $f =~ s#%([0-9a-fA-F]{2})#chr(hex($1))#eg;
4323        $f
4324}
4325
4326sub remove_username {
4327        $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
4328}
4329
4330package Git::SVN::Prompt;
4331use strict;
4332use warnings;
4333require SVN::Core;
4334use vars qw/$_no_auth_cache $_username/;
4335
4336sub simple {
4337        my ($cred, $realm, $default_username, $may_save, $pool) = @_;
4338        $may_save = undef if $_no_auth_cache;
4339        $default_username = $_username if defined $_username;
4340        if (defined $default_username && length $default_username) {
4341                if (defined $realm && length $realm) {
4342                        print STDERR "Authentication realm: $realm\n";
4343                        STDERR->flush;
4344                }
4345                $cred->username($default_username);
4346        } else {
4347                username($cred, $realm, $may_save, $pool);
4348        }
4349        $cred->password(_read_password("Password for '" .
4350                                       $cred->username . "': ", $realm));
4351        $cred->may_save($may_save);
4352        $SVN::_Core::SVN_NO_ERROR;
4353}
4354
4355sub ssl_server_trust {
4356        my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
4357        $may_save = undef if $_no_auth_cache;
4358        print STDERR "Error validating server certificate for '$realm':\n";
4359        {
4360                no warnings 'once';
4361                # All variables SVN::Auth::SSL::* are used only once,
4362                # so we're shutting up Perl warnings about this.
4363                if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
4364                        print STDERR " - The certificate is not issued ",
4365                            "by a trusted authority. Use the\n",
4366                            "   fingerprint to validate ",
4367                            "the certificate manually!\n";
4368                }
4369                if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
4370                        print STDERR " - The certificate hostname ",
4371                            "does not match.\n";
4372                }
4373                if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
4374                        print STDERR " - The certificate is not yet valid.\n";
4375                }
4376                if ($failures & $SVN::Auth::SSL::EXPIRED) {
4377                        print STDERR " - The certificate has expired.\n";
4378                }
4379                if ($failures & $SVN::Auth::SSL::OTHER) {
4380                        print STDERR " - The certificate has ",
4381                            "an unknown error.\n";
4382                }
4383        } # no warnings 'once'
4384        printf STDERR
4385                "Certificate information:\n".
4386                " - Hostname: %s\n".
4387                " - Valid: from %s until %s\n".
4388                " - Issuer: %s\n".
4389                " - Fingerprint: %s\n",
4390                map $cert_info->$_, qw(hostname valid_from valid_until
4391                                       issuer_dname fingerprint);
4392        my $choice;
4393prompt:
4394        print STDERR $may_save ?
4395              "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
4396              "(R)eject or accept (t)emporarily? ";
4397        STDERR->flush;
4398        $choice = lc(substr(<STDIN> || 'R', 0, 1));
4399        if ($choice =~ /^t$/i) {
4400                $cred->may_save(undef);
4401        } elsif ($choice =~ /^r$/i) {
4402                return -1;
4403        } elsif ($may_save && $choice =~ /^p$/i) {
4404                $cred->may_save($may_save);
4405        } else {
4406                goto prompt;
4407        }
4408        $cred->accepted_failures($failures);
4409        $SVN::_Core::SVN_NO_ERROR;
4410}
4411
4412sub ssl_client_cert {
4413        my ($cred, $realm, $may_save, $pool) = @_;
4414        $may_save = undef if $_no_auth_cache;
4415        print STDERR "Client certificate filename: ";
4416        STDERR->flush;
4417        chomp(my $filename = <STDIN>);
4418        $cred->cert_file($filename);
4419        $cred->may_save($may_save);
4420        $SVN::_Core::SVN_NO_ERROR;
4421}
4422
4423sub ssl_client_cert_pw {
4424        my ($cred, $realm, $may_save, $pool) = @_;
4425        $may_save = undef if $_no_auth_cache;
4426        $cred->password(_read_password("Password: ", $realm));
4427        $cred->may_save($may_save);
4428        $SVN::_Core::SVN_NO_ERROR;
4429}
4430
4431sub username {
4432        my ($cred, $realm, $may_save, $pool) = @_;
4433        $may_save = undef if $_no_auth_cache;
4434        if (defined $realm && length $realm) {
4435                print STDERR "Authentication realm: $realm\n";
4436        }
4437        my $username;
4438        if (defined $_username) {
4439                $username = $_username;
4440        } else {
4441                print STDERR "Username: ";
4442                STDERR->flush;
4443                chomp($username = <STDIN>);
4444        }
4445        $cred->username($username);
4446        $cred->may_save($may_save);
4447        $SVN::_Core::SVN_NO_ERROR;
4448}
4449
4450sub _read_password {
4451        my ($prompt, $realm) = @_;
4452        my $password = '';
4453        if (exists $ENV{GIT_ASKPASS}) {
4454                open(PH, "-|", $ENV{GIT_ASKPASS}, $prompt);
4455                $password = <PH>;
4456                $password =~ s/[\012\015]//; # \n\r
4457                close(PH);
4458        } else {
4459                print STDERR $prompt;
4460                STDERR->flush;
4461                require Term::ReadKey;
4462                Term::ReadKey::ReadMode('noecho');
4463                while (defined(my $key = Term::ReadKey::ReadKey(0))) {
4464                        last if $key =~ /[\012\015]/; # \n\r
4465                        $password .= $key;
4466                }
4467                Term::ReadKey::ReadMode('restore');
4468                print STDERR "\n";
4469                STDERR->flush;
4470        }
4471        $password;
4472}
4473
4474package SVN::Git::Fetcher;
4475use vars qw/@ISA $_ignore_regex $_preserve_empty_dirs $_placeholder_filename
4476            @deleted_gpath %added_placeholder $repo_id/;
4477use strict;
4478use warnings;
4479use Carp qw/croak/;
4480use File::Basename qw/dirname/;
4481use IO::File qw//;
4482
4483# file baton members: path, mode_a, mode_b, pool, fh, blob, base
4484sub new {
4485        my ($class, $git_svn, $switch_path) = @_;
4486        my $self = SVN::Delta::Editor->new;
4487        bless $self, $class;
4488        if (exists $git_svn->{last_commit}) {
4489                $self->{c} = $git_svn->{last_commit};
4490                $self->{empty_symlinks} =
4491                                  _mark_empty_symlinks($git_svn, $switch_path);
4492        }
4493
4494        # some options are read globally, but can be overridden locally
4495        # per [svn-remote "..."] section.  Command-line options will *NOT*
4496        # override options set in an [svn-remote "..."] section
4497        $repo_id = $git_svn->{repo_id};
4498        my $k = "svn-remote.$repo_id.ignore-paths";
4499        my $v = eval { command_oneline('config', '--get', $k) };
4500        $self->{ignore_regex} = $v;
4501
4502        $k = "svn-remote.$repo_id.preserve-empty-dirs";
4503        $v = eval { command_oneline('config', '--get', '--bool', $k) };
4504        if ($v && $v eq 'true') {
4505                $_preserve_empty_dirs = 1;
4506                $k = "svn-remote.$repo_id.placeholder-filename";
4507                $v = eval { command_oneline('config', '--get', $k) };
4508                $_placeholder_filename = $v;
4509        }
4510
4511        # Load the list of placeholder files added during previous invocations.
4512        $k = "svn-remote.$repo_id.added-placeholder";
4513        $v = eval { command_oneline('config', '--get-all', $k) };
4514        if ($_preserve_empty_dirs && $v) {
4515                # command() prints errors to stderr, so we only call it if
4516                # command_oneline() succeeded.
4517                my @v = command('config', '--get-all', $k);
4518                $added_placeholder{ dirname($_) } = $_ foreach @v;
4519        }
4520
4521        $self->{empty} = {};
4522        $self->{dir_prop} = {};
4523        $self->{file_prop} = {};
4524        $self->{absent_dir} = {};
4525        $self->{absent_file} = {};
4526        $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
4527        $self->{pathnameencoding} = Git::config('svn.pathnameencoding');
4528        $self;
4529}
4530
4531# this uses the Ra object, so it must be called before do_{switch,update},
4532# not inside them (when the Git::SVN::Fetcher object is passed) to
4533# do_{switch,update}
4534sub _mark_empty_symlinks {
4535        my ($git_svn, $switch_path) = @_;
4536        my $bool = Git::config_bool('svn.brokenSymlinkWorkaround');
4537        return {} if (!defined($bool)) || (defined($bool) && ! $bool);
4538
4539        my %ret;
4540        my ($rev, $cmt) = $git_svn->last_rev_commit;
4541        return {} unless ($rev && $cmt);
4542
4543        # allow the warning to be printed for each revision we fetch to
4544        # ensure the user sees it.  The user can also disable the workaround
4545        # on the repository even while git svn is running and the next
4546        # revision fetched will skip this expensive function.
4547        my $printed_warning;
4548        chomp(my $empty_blob = `git hash-object -t blob --stdin < /dev/null`);
4549        my ($ls, $ctx) = command_output_pipe(qw/ls-tree -r -z/, $cmt);
4550        local $/ = "\0";
4551        my $pfx = defined($switch_path) ? $switch_path : $git_svn->{path};
4552        $pfx .= '/' if length($pfx);
4553        while (<$ls>) {
4554                chomp;
4555                s/\A100644 blob $empty_blob\t//o or next;
4556                unless ($printed_warning) {
4557                        print STDERR "Scanning for empty symlinks, ",
4558                                     "this may take a while if you have ",
4559                                     "many empty files\n",
4560                                     "You may disable this with `",
4561                                     "git config svn.brokenSymlinkWorkaround ",
4562                                     "false'.\n",
4563                                     "This may be done in a different ",
4564                                     "terminal without restarting ",
4565                                     "git svn\n";
4566                        $printed_warning = 1;
4567                }
4568                my $path = $_;
4569                my (undef, $props) =
4570                               $git_svn->ra->get_file($pfx.$path, $rev, undef);
4571                if ($props->{'svn:special'}) {
4572                        $ret{$path} = 1;
4573                }
4574        }
4575        command_close_pipe($ls, $ctx);
4576        \%ret;
4577}
4578
4579# returns true if a given path is inside a ".git" directory
4580sub in_dot_git {
4581        $_[0] =~ m{(?:^|/)\.git(?:/|$)};
4582}
4583
4584# return value: 0 -- don't ignore, 1 -- ignore
4585sub is_path_ignored {
4586        my ($self, $path) = @_;
4587        return 1 if in_dot_git($path);
4588        return 1 if defined($self->{ignore_regex}) &&
4589                    $path =~ m!$self->{ignore_regex}!;
4590        return 0 unless defined($_ignore_regex);
4591        return 1 if $path =~ m!$_ignore_regex!o;
4592        return 0;
4593}
4594
4595sub set_path_strip {
4596        my ($self, $path) = @_;
4597        $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
4598}
4599
4600sub open_root {
4601        { path => '' };
4602}
4603
4604sub open_directory {
4605        my ($self, $path, $pb, $rev) = @_;
4606        { path => $path };
4607}
4608
4609sub git_path {
4610        my ($self, $path) = @_;
4611        if (my $enc = $self->{pathnameencoding}) {
4612                require Encode;
4613                Encode::from_to($path, 'UTF-8', $enc);
4614        }
4615        if ($self->{path_strip}) {
4616                $path =~ s!$self->{path_strip}!! or
4617                  die "Failed to strip path '$path' ($self->{path_strip})\n";
4618        }
4619        $path;
4620}
4621
4622sub delete_entry {
4623        my ($self, $path, $rev, $pb) = @_;
4624        return undef if $self->is_path_ignored($path);
4625
4626        my $gpath = $self->git_path($path);
4627        return undef if ($gpath eq '');
4628
4629        # remove entire directories.
4630        my ($tree) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
4631                         =~ /\A040000 tree ([a-f\d]{40})\t\Q$gpath\E\0/);
4632        if ($tree) {
4633                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
4634                                                     -r --name-only -z/,
4635                                                     $tree);
4636                local $/ = "\0";
4637                while (<$ls>) {
4638                        chomp;
4639                        my $rmpath = "$gpath/$_";
4640                        $self->{gii}->remove($rmpath);
4641                        print "\tD\t$rmpath\n" unless $::_q;
4642                }
4643                print "\tD\t$gpath/\n" unless $::_q;
4644                command_close_pipe($ls, $ctx);
4645        } else {
4646                $self->{gii}->remove($gpath);
4647                print "\tD\t$gpath\n" unless $::_q;
4648        }
4649        # Don't add to @deleted_gpath if we're deleting a placeholder file.
4650        push @deleted_gpath, $gpath unless $added_placeholder{dirname($path)};
4651        $self->{empty}->{$path} = 0;
4652        undef;
4653}
4654
4655sub open_file {
4656        my ($self, $path, $pb, $rev) = @_;
4657        my ($mode, $blob);
4658
4659        goto out if $self->is_path_ignored($path);
4660
4661        my $gpath = $self->git_path($path);
4662        ($mode, $blob) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
4663                             =~ /\A(\d{6}) blob ([a-f\d]{40})\t\Q$gpath\E\0/);
4664        unless (defined $mode && defined $blob) {
4665                die "$path was not found in commit $self->{c} (r$rev)\n";
4666        }
4667        if ($mode eq '100644' && $self->{empty_symlinks}->{$path}) {
4668                $mode = '120000';
4669        }
4670out:
4671        { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
4672          pool => SVN::Pool->new, action => 'M' };
4673}
4674
4675sub add_file {
4676        my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
4677        my $mode;
4678
4679        if (!$self->is_path_ignored($path)) {
4680                my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
4681                delete $self->{empty}->{$dir};
4682                $mode = '100644';
4683
4684                if ($added_placeholder{$dir}) {
4685                        # Remove our placeholder file, if we created one.
4686                        delete_entry($self, $added_placeholder{$dir})
4687                                unless $path eq $added_placeholder{$dir};
4688                        delete $added_placeholder{$dir}
4689                }
4690        }
4691
4692        { path => $path, mode_a => $mode, mode_b => $mode,
4693          pool => SVN::Pool->new, action => 'A' };
4694}
4695
4696sub add_directory {
4697        my ($self, $path, $cp_path, $cp_rev) = @_;
4698        goto out if $self->is_path_ignored($path);
4699        my $gpath = $self->git_path($path);
4700        if ($gpath eq '') {
4701                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
4702                                                     -r --name-only -z/,
4703                                                     $self->{c});
4704                local $/ = "\0";
4705                while (<$ls>) {
4706                        chomp;
4707                        $self->{gii}->remove($_);
4708                        print "\tD\t$_\n" unless $::_q;
4709                        push @deleted_gpath, $gpath;
4710                }
4711                command_close_pipe($ls, $ctx);
4712                $self->{empty}->{$path} = 0;
4713        }
4714        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
4715        delete $self->{empty}->{$dir};
4716        $self->{empty}->{$path} = 1;
4717
4718        if ($added_placeholder{$dir}) {
4719                # Remove our placeholder file, if we created one.
4720                delete_entry($self, $added_placeholder{$dir});
4721                delete $added_placeholder{$dir}
4722        }
4723
4724out:
4725        { path => $path };
4726}
4727
4728sub change_dir_prop {
4729        my ($self, $db, $prop, $value) = @_;
4730        return undef if $self->is_path_ignored($db->{path});
4731        $self->{dir_prop}->{$db->{path}} ||= {};
4732        $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
4733        undef;
4734}
4735
4736sub absent_directory {
4737        my ($self, $path, $pb) = @_;
4738        return undef if $self->is_path_ignored($path);
4739        $self->{absent_dir}->{$pb->{path}} ||= [];
4740        push @{$self->{absent_dir}->{$pb->{path}}}, $path;
4741        undef;
4742}
4743
4744sub absent_file {
4745        my ($self, $path, $pb) = @_;
4746        return undef if $self->is_path_ignored($path);
4747        $self->{absent_file}->{$pb->{path}} ||= [];
4748        push @{$self->{absent_file}->{$pb->{path}}}, $path;
4749        undef;
4750}
4751
4752sub change_file_prop {
4753        my ($self, $fb, $prop, $value) = @_;
4754        return undef if $self->is_path_ignored($fb->{path});
4755        if ($prop eq 'svn:executable') {
4756                if ($fb->{mode_b} != 120000) {
4757                        $fb->{mode_b} = defined $value ? 100755 : 100644;
4758                }
4759        } elsif ($prop eq 'svn:special') {
4760                $fb->{mode_b} = defined $value ? 120000 : 100644;
4761        } else {
4762                $self->{file_prop}->{$fb->{path}} ||= {};
4763                $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
4764        }
4765        undef;
4766}
4767
4768sub apply_textdelta {
4769        my ($self, $fb, $exp) = @_;
4770        return undef if $self->is_path_ignored($fb->{path});
4771        my $fh = $::_repository->temp_acquire('svn_delta');
4772        # $fh gets auto-closed() by SVN::TxDelta::apply(),
4773        # (but $base does not,) so dup() it for reading in close_file
4774        open my $dup, '<&', $fh or croak $!;
4775        my $base = $::_repository->temp_acquire('git_blob');
4776
4777        if ($fb->{blob}) {
4778                my ($base_is_link, $size);
4779
4780                if ($fb->{mode_a} eq '120000' &&
4781                    ! $self->{empty_symlinks}->{$fb->{path}}) {
4782                        print $base 'link ' or die "print $!\n";
4783                        $base_is_link = 1;
4784                }
4785        retry:
4786                $size = $::_repository->cat_blob($fb->{blob}, $base);
4787                die "Failed to read object $fb->{blob}" if ($size < 0);
4788
4789                if (defined $exp) {
4790                        seek $base, 0, 0 or croak $!;
4791                        my $got = ::md5sum($base);
4792                        if ($got ne $exp) {
4793                                my $err = "Checksum mismatch: ".
4794                                       "$fb->{path} $fb->{blob}\n" .
4795                                       "expected: $exp\n" .
4796                                       "     got: $got\n";
4797                                if ($base_is_link) {
4798                                        warn $err,
4799                                             "Retrying... (possibly ",
4800                                             "a bad symlink from SVN)\n";
4801                                        $::_repository->temp_reset($base);
4802                                        $base_is_link = 0;
4803                                        goto retry;
4804                                }
4805                                die $err;
4806                        }
4807                }
4808        }
4809        seek $base, 0, 0 or croak $!;
4810        $fb->{fh} = $fh;
4811        $fb->{base} = $base;
4812        [ SVN::TxDelta::apply($base, $dup, undef, $fb->{path}, $fb->{pool}) ];
4813}
4814
4815sub close_file {
4816        my ($self, $fb, $exp) = @_;
4817        return undef if $self->is_path_ignored($fb->{path});
4818
4819        my $hash;
4820        my $path = $self->git_path($fb->{path});
4821        if (my $fh = $fb->{fh}) {
4822                if (defined $exp) {
4823                        seek($fh, 0, 0) or croak $!;
4824                        my $got = ::md5sum($fh);
4825                        if ($got ne $exp) {
4826                                die "Checksum mismatch: $path\n",
4827                                    "expected: $exp\n    got: $got\n";
4828                        }
4829                }
4830                if ($fb->{mode_b} == 120000) {
4831                        sysseek($fh, 0, 0) or croak $!;
4832                        my $rd = sysread($fh, my $buf, 5);
4833
4834                        if (!defined $rd) {
4835                                croak "sysread: $!\n";
4836                        } elsif ($rd == 0) {
4837                                warn "$path has mode 120000",
4838                                     " but it points to nothing\n",
4839                                     "converting to an empty file with mode",
4840                                     " 100644\n";
4841                                $fb->{mode_b} = '100644';
4842                        } elsif ($buf ne 'link ') {
4843                                warn "$path has mode 120000",
4844                                     " but is not a link\n";
4845                        } else {
4846                                my $tmp_fh = $::_repository->temp_acquire(
4847                                        'svn_hash');
4848                                my $res;
4849                                while ($res = sysread($fh, my $str, 1024)) {
4850                                        my $out = syswrite($tmp_fh, $str, $res);
4851                                        defined($out) && $out == $res
4852                                                or croak("write ",
4853                                                        Git::temp_path($tmp_fh),
4854                                                        ": $!\n");
4855                                }
4856                                defined $res or croak $!;
4857
4858                                ($fh, $tmp_fh) = ($tmp_fh, $fh);
4859                                Git::temp_release($tmp_fh, 1);
4860                        }
4861                }
4862
4863                $hash = $::_repository->hash_and_insert_object(
4864                                Git::temp_path($fh));
4865                $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
4866
4867                Git::temp_release($fb->{base}, 1);
4868                Git::temp_release($fh, 1);
4869        } else {
4870                $hash = $fb->{blob} or die "no blob information\n";
4871        }
4872        $fb->{pool}->clear;
4873        $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
4874        print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
4875        undef;
4876}
4877
4878sub abort_edit {
4879        my $self = shift;
4880        $self->{nr} = $self->{gii}->{nr};
4881        delete $self->{gii};
4882        $self->SUPER::abort_edit(@_);
4883}
4884
4885sub close_edit {
4886        my $self = shift;
4887
4888        if ($_preserve_empty_dirs) {
4889                my @empty_dirs;
4890
4891                # Any entry flagged as empty that also has an associated
4892                # dir_prop represents a newly created empty directory.
4893                foreach my $i (keys %{$self->{empty}}) {
4894                        push @empty_dirs, $i if exists $self->{dir_prop}->{$i};
4895                }
4896
4897                # Search for directories that have become empty due subsequent
4898                # file deletes.
4899                push @empty_dirs, $self->find_empty_directories();
4900
4901                # Finally, add a placeholder file to each empty directory.
4902                $self->add_placeholder_file($_) foreach (@empty_dirs);
4903
4904                $self->stash_placeholder_list();
4905        }
4906
4907        $self->{git_commit_ok} = 1;
4908        $self->{nr} = $self->{gii}->{nr};
4909        delete $self->{gii};
4910        $self->SUPER::close_edit(@_);
4911}
4912
4913sub find_empty_directories {
4914        my ($self) = @_;
4915        my @empty_dirs;
4916        my %dirs = map { dirname($_) => 1 } @deleted_gpath;
4917
4918        foreach my $dir (sort keys %dirs) {
4919                next if $dir eq ".";
4920
4921                # If there have been any additions to this directory, there is
4922                # no reason to check if it is empty.
4923                my $skip_added = 0;
4924                foreach my $t (qw/dir_prop file_prop/) {
4925                        foreach my $path (keys %{ $self->{$t} }) {
4926                                if (exists $self->{$t}->{dirname($path)}) {
4927                                        $skip_added = 1;
4928                                        last;
4929                                }
4930                        }
4931                        last if $skip_added;
4932                }
4933                next if $skip_added;
4934
4935                # Use `git ls-tree` to get the filenames of this directory
4936                # that existed prior to this particular commit.
4937                my $ls = command('ls-tree', '-z', '--name-only',
4938                                 $self->{c}, "$dir/");
4939                my %files = map { $_ => 1 } split(/\0/, $ls);
4940
4941                # Remove the filenames that were deleted during this commit.
4942                delete $files{$_} foreach (@deleted_gpath);
4943
4944                # Report the directory if there are no filenames left.
4945                push @empty_dirs, $dir unless (scalar %files);
4946        }
4947        @empty_dirs;
4948}
4949
4950sub add_placeholder_file {
4951        my ($self, $dir) = @_;
4952        my $path = "$dir/$_placeholder_filename";
4953        my $gpath = $self->git_path($path);
4954
4955        my $fh = $::_repository->temp_acquire($gpath);
4956        my $hash = $::_repository->hash_and_insert_object(Git::temp_path($fh));
4957        Git::temp_release($fh, 1);
4958        $self->{gii}->update('100644', $hash, $gpath) or croak $!;
4959
4960        # The directory should no longer be considered empty.
4961        delete $self->{empty}->{$dir} if exists $self->{empty}->{$dir};
4962
4963        # Keep track of any placeholder files we create.
4964        $added_placeholder{$dir} = $path;
4965}
4966
4967sub stash_placeholder_list {
4968        my ($self) = @_;
4969        my $k = "svn-remote.$repo_id.added-placeholder";
4970        my $v = eval { command_oneline('config', '--get-all', $k) };
4971        command_noisy('config', '--unset-all', $k) if $v;
4972        foreach (values %added_placeholder) {
4973                command_noisy('config', '--add', $k, $_);
4974        }
4975}
4976
4977package SVN::Git::Editor;
4978use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
4979use strict;
4980use warnings;
4981use Carp qw/croak/;
4982use IO::File;
4983
4984sub new {
4985        my ($class, $opts) = @_;
4986        foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
4987                die "$_ required!\n" unless (defined $opts->{$_});
4988        }
4989
4990        my $pool = SVN::Pool->new;
4991        my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
4992        my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
4993                                     $opts->{r}, $mods);
4994
4995        # $opts->{ra} functions should not be used after this:
4996        my @ce  = $opts->{ra}->get_commit_editor($opts->{log},
4997                                                $opts->{editor_cb}, $pool);
4998        my $self = SVN::Delta::Editor->new(@ce, $pool);
4999        bless $self, $class;
5000        foreach (qw/svn_path r tree_a tree_b/) {
5001                $self->{$_} = $opts->{$_};
5002        }
5003        $self->{url} = $opts->{ra}->{url};
5004        $self->{mods} = $mods;
5005        $self->{types} = $types;
5006        $self->{pool} = $pool;
5007        $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
5008        $self->{rm} = { };
5009        $self->{path_prefix} = length $self->{svn_path} ?
5010                               "$self->{svn_path}/" : '';
5011        $self->{config} = $opts->{config};
5012        $self->{mergeinfo} = $opts->{mergeinfo};
5013        return $self;
5014}
5015
5016sub generate_diff {
5017        my ($tree_a, $tree_b) = @_;
5018        my @diff_tree = qw(diff-tree -z -r);
5019        if ($_cp_similarity) {
5020                push @diff_tree, "-C$_cp_similarity";
5021        } else {
5022                push @diff_tree, '-C';
5023        }
5024        push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
5025        push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
5026        push @diff_tree, $tree_a, $tree_b;
5027        my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
5028        local $/ = "\0";
5029        my $state = 'meta';
5030        my @mods;
5031        while (<$diff_fh>) {
5032                chomp $_; # this gets rid of the trailing "\0"
5033                if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
5034                                        ($::sha1)\s($::sha1)\s
5035                                        ([MTCRAD])\d*$/xo) {
5036                        push @mods, {   mode_a => $1, mode_b => $2,
5037                                        sha1_a => $3, sha1_b => $4,
5038                                        chg => $5 };
5039                        if ($5 =~ /^(?:C|R)$/) {
5040                                $state = 'file_a';
5041                        } else {
5042                                $state = 'file_b';
5043                        }
5044                } elsif ($state eq 'file_a') {
5045                        my $x = $mods[$#mods] or croak "Empty array\n";
5046                        if ($x->{chg} !~ /^(?:C|R)$/) {
5047                                croak "Error parsing $_, $x->{chg}\n";
5048                        }
5049                        $x->{file_a} = $_;
5050                        $state = 'file_b';
5051                } elsif ($state eq 'file_b') {
5052                        my $x = $mods[$#mods] or croak "Empty array\n";
5053                        if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
5054                                croak "Error parsing $_, $x->{chg}\n";
5055                        }
5056                        if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
5057                                croak "Error parsing $_, $x->{chg}\n";
5058                        }
5059                        $x->{file_b} = $_;
5060                        $state = 'meta';
5061                } else {
5062                        croak "Error parsing $_\n";
5063                }
5064        }
5065        command_close_pipe($diff_fh, $ctx);
5066        \@mods;
5067}
5068
5069sub check_diff_paths {
5070        my ($ra, $pfx, $rev, $mods) = @_;
5071        my %types;
5072        $pfx .= '/' if length $pfx;
5073
5074        sub type_diff_paths {
5075                my ($ra, $types, $path, $rev) = @_;
5076                my @p = split m#/+#, $path;
5077                my $c = shift @p;
5078                unless (defined $types->{$c}) {
5079                        $types->{$c} = $ra->check_path($c, $rev);
5080                }
5081                while (@p) {
5082                        $c .= '/' . shift @p;
5083                        next if defined $types->{$c};
5084                        $types->{$c} = $ra->check_path($c, $rev);
5085                }
5086        }
5087
5088        foreach my $m (@$mods) {
5089                foreach my $f (qw/file_a file_b/) {
5090                        next unless defined $m->{$f};
5091                        my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
5092                        if (length $pfx.$dir && ! defined $types{$dir}) {
5093                                type_diff_paths($ra, \%types, $pfx.$dir, $rev);
5094                        }
5095                }
5096        }
5097        \%types;
5098}
5099
5100sub split_path {
5101        return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
5102}
5103
5104sub repo_path {
5105        my ($self, $path) = @_;
5106        if (my $enc = $self->{pathnameencoding}) {
5107                require Encode;
5108                Encode::from_to($path, $enc, 'UTF-8');
5109        }
5110        $self->{path_prefix}.(defined $path ? $path : '');
5111}
5112
5113sub url_path {
5114        my ($self, $path) = @_;
5115        if ($self->{url} =~ m#^https?://#) {
5116                $path =~ s!([^~a-zA-Z0-9_./-])!uc sprintf("%%%02x",ord($1))!eg;
5117        }
5118        $self->{url} . '/' . $self->repo_path($path);
5119}
5120
5121sub rmdirs {
5122        my ($self) = @_;
5123        my $rm = $self->{rm};
5124        delete $rm->{''}; # we never delete the url we're tracking
5125        return unless %$rm;
5126
5127        foreach (keys %$rm) {
5128                my @d = split m#/#, $_;
5129                my $c = shift @d;
5130                $rm->{$c} = 1;
5131                while (@d) {
5132                        $c .= '/' . shift @d;
5133                        $rm->{$c} = 1;
5134                }
5135        }
5136        delete $rm->{$self->{svn_path}};
5137        delete $rm->{''}; # we never delete the url we're tracking
5138        return unless %$rm;
5139
5140        my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
5141                                             $self->{tree_b});
5142        local $/ = "\0";
5143        while (<$fh>) {
5144                chomp;
5145                my @dn = split m#/#, $_;
5146                while (pop @dn) {
5147                        delete $rm->{join '/', @dn};
5148                }
5149                unless (%$rm) {
5150                        close $fh;
5151                        return;
5152                }
5153        }
5154        command_close_pipe($fh, $ctx);
5155
5156        my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
5157        foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
5158                $self->close_directory($bat->{$d}, $p);
5159                my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
5160                print "\tD+\t$d/\n" unless $::_q;
5161                $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
5162                delete $bat->{$d};
5163        }
5164}
5165
5166sub open_or_add_dir {
5167        my ($self, $full_path, $baton, $deletions) = @_;
5168        my $t = $self->{types}->{$full_path};
5169        if (!defined $t) {
5170                die "$full_path not known in r$self->{r} or we have a bug!\n";
5171        }
5172        {
5173                no warnings 'once';
5174                # SVN::Node::none and SVN::Node::file are used only once,
5175                # so we're shutting up Perl's warnings about them.
5176                if ($t == $SVN::Node::none || defined($deletions->{$full_path})) {
5177                        return $self->add_directory($full_path, $baton,
5178                            undef, -1, $self->{pool});
5179                } elsif ($t == $SVN::Node::dir) {
5180                        return $self->open_directory($full_path, $baton,
5181                            $self->{r}, $self->{pool});
5182                } # no warnings 'once'
5183                print STDERR "$full_path already exists in repository at ",
5184                    "r$self->{r} and it is not a directory (",
5185                    ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
5186        } # no warnings 'once'
5187        exit 1;
5188}
5189
5190sub ensure_path {
5191        my ($self, $path, $deletions) = @_;
5192        my $bat = $self->{bat};
5193        my $repo_path = $self->repo_path($path);
5194        return $bat->{''} unless (length $repo_path);
5195
5196        my @p = split m#/+#, $repo_path;
5197        my $c = shift @p;
5198        $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''}, $deletions);
5199        while (@p) {
5200                my $c0 = $c;
5201                $c .= '/' . shift @p;
5202                $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0}, $deletions);
5203        }
5204        return $bat->{$c};
5205}
5206
5207# Subroutine to convert a globbing pattern to a regular expression.
5208# From perl cookbook.
5209sub glob2pat {
5210        my $globstr = shift;
5211        my %patmap = ('*' => '.*', '?' => '.', '[' => '[', ']' => ']');
5212        $globstr =~ s{(.)} { $patmap{$1} || "\Q$1" }ge;
5213        return '^' . $globstr . '$';
5214}
5215
5216sub check_autoprop {
5217        my ($self, $pattern, $properties, $file, $fbat) = @_;
5218        # Convert the globbing pattern to a regular expression.
5219        my $regex = glob2pat($pattern);
5220        # Check if the pattern matches the file name.
5221        if($file =~ m/($regex)/) {
5222                # Parse the list of properties to set.
5223                my @props = split(/;/, $properties);
5224                foreach my $prop (@props) {
5225                        # Parse 'name=value' syntax and set the property.
5226                        if ($prop =~ /([^=]+)=(.*)/) {
5227                                my ($n,$v) = ($1,$2);
5228                                for ($n, $v) {
5229                                        s/^\s+//; s/\s+$//;
5230                                }
5231                                $self->change_file_prop($fbat, $n, $v);
5232                        }
5233                }
5234        }
5235}
5236
5237sub apply_autoprops {
5238        my ($self, $file, $fbat) = @_;
5239        my $conf_t = ${$self->{config}}{'config'};
5240        no warnings 'once';
5241        # Check [miscellany]/enable-auto-props in svn configuration.
5242        if (SVN::_Core::svn_config_get_bool(
5243                $conf_t,
5244                $SVN::_Core::SVN_CONFIG_SECTION_MISCELLANY,
5245                $SVN::_Core::SVN_CONFIG_OPTION_ENABLE_AUTO_PROPS,
5246                0)) {
5247                # Auto-props are enabled.  Enumerate them to look for matches.
5248                my $callback = sub {
5249                        $self->check_autoprop($_[0], $_[1], $file, $fbat);
5250                };
5251                SVN::_Core::svn_config_enumerate(
5252                        $conf_t,
5253                        $SVN::_Core::SVN_CONFIG_SECTION_AUTO_PROPS,
5254                        $callback);
5255        }
5256}
5257
5258sub A {
5259        my ($self, $m, $deletions) = @_;
5260        my ($dir, $file) = split_path($m->{file_b});
5261        my $pbat = $self->ensure_path($dir, $deletions);
5262        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
5263                                        undef, -1);
5264        print "\tA\t$m->{file_b}\n" unless $::_q;
5265        $self->apply_autoprops($file, $fbat);
5266        $self->chg_file($fbat, $m);
5267        $self->close_file($fbat,undef,$self->{pool});
5268}
5269
5270sub C {
5271        my ($self, $m, $deletions) = @_;
5272        my ($dir, $file) = split_path($m->{file_b});
5273        my $pbat = $self->ensure_path($dir, $deletions);
5274        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
5275                                $self->url_path($m->{file_a}), $self->{r});
5276        print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
5277        $self->chg_file($fbat, $m);
5278        $self->close_file($fbat,undef,$self->{pool});
5279}
5280
5281sub delete_entry {
5282        my ($self, $path, $pbat) = @_;
5283        my $rpath = $self->repo_path($path);
5284        my ($dir, $file) = split_path($rpath);
5285        $self->{rm}->{$dir} = 1;
5286        $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
5287}
5288
5289sub R {
5290        my ($self, $m, $deletions) = @_;
5291        my ($dir, $file) = split_path($m->{file_b});
5292        my $pbat = $self->ensure_path($dir, $deletions);
5293        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
5294                                $self->url_path($m->{file_a}), $self->{r});
5295        print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
5296        $self->apply_autoprops($file, $fbat);
5297        $self->chg_file($fbat, $m);
5298        $self->close_file($fbat,undef,$self->{pool});
5299
5300        ($dir, $file) = split_path($m->{file_a});
5301        $pbat = $self->ensure_path($dir, $deletions);
5302        $self->delete_entry($m->{file_a}, $pbat);
5303}
5304
5305sub M {
5306        my ($self, $m, $deletions) = @_;
5307        my ($dir, $file) = split_path($m->{file_b});
5308        my $pbat = $self->ensure_path($dir, $deletions);
5309        my $fbat = $self->open_file($self->repo_path($m->{file_b}),
5310                                $pbat,$self->{r},$self->{pool});
5311        print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
5312        $self->chg_file($fbat, $m);
5313        $self->close_file($fbat,undef,$self->{pool});
5314}
5315
5316sub T { shift->M(@_) }
5317
5318sub change_file_prop {
5319        my ($self, $fbat, $pname, $pval) = @_;
5320        $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
5321}
5322
5323sub change_dir_prop {
5324        my ($self, $pbat, $pname, $pval) = @_;
5325        $self->SUPER::change_dir_prop($pbat, $pname, $pval, $self->{pool});
5326}
5327
5328sub _chg_file_get_blob ($$$$) {
5329        my ($self, $fbat, $m, $which) = @_;
5330        my $fh = $::_repository->temp_acquire("git_blob_$which");
5331        if ($m->{"mode_$which"} =~ /^120/) {
5332                print $fh 'link ' or croak $!;
5333                $self->change_file_prop($fbat,'svn:special','*');
5334        } elsif ($m->{mode_a} =~ /^120/ && $m->{"mode_$which"} !~ /^120/) {
5335                $self->change_file_prop($fbat,'svn:special',undef);
5336        }
5337        my $blob = $m->{"sha1_$which"};
5338        return ($fh,) if ($blob =~ /^0{40}$/);
5339        my $size = $::_repository->cat_blob($blob, $fh);
5340        croak "Failed to read object $blob" if ($size < 0);
5341        $fh->flush == 0 or croak $!;
5342        seek $fh, 0, 0 or croak $!;
5343
5344        my $exp = ::md5sum($fh);
5345        seek $fh, 0, 0 or croak $!;
5346        return ($fh, $exp);
5347}
5348
5349sub chg_file {
5350        my ($self, $fbat, $m) = @_;
5351        if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
5352                $self->change_file_prop($fbat,'svn:executable','*');
5353        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
5354                $self->change_file_prop($fbat,'svn:executable',undef);
5355        }
5356        my ($fh_a, $exp_a) = _chg_file_get_blob $self, $fbat, $m, 'a';
5357        my ($fh_b, $exp_b) = _chg_file_get_blob $self, $fbat, $m, 'b';
5358        my $pool = SVN::Pool->new;
5359        my $atd = $self->apply_textdelta($fbat, $exp_a, $pool);
5360        if (-s $fh_a) {
5361                my $txstream = SVN::TxDelta::new ($fh_a, $fh_b, $pool);
5362                my $res = SVN::TxDelta::send_txstream($txstream, @$atd, $pool);
5363                if (defined $res) {
5364                        die "Unexpected result from send_txstream: $res\n",
5365                            "(SVN::Core::VERSION: $SVN::Core::VERSION)\n";
5366                }
5367        } else {
5368                my $got = SVN::TxDelta::send_stream($fh_b, @$atd, $pool);
5369                die "Checksum mismatch\nexpected: $exp_b\ngot: $got\n"
5370                    if ($got ne $exp_b);
5371        }
5372        Git::temp_release($fh_b, 1);
5373        Git::temp_release($fh_a, 1);
5374        $pool->clear;
5375}
5376
5377sub D {
5378        my ($self, $m, $deletions) = @_;
5379        my ($dir, $file) = split_path($m->{file_b});
5380        my $pbat = $self->ensure_path($dir, $deletions);
5381        print "\tD\t$m->{file_b}\n" unless $::_q;
5382        $self->delete_entry($m->{file_b}, $pbat);
5383}
5384
5385sub close_edit {
5386        my ($self) = @_;
5387        my ($p,$bat) = ($self->{pool}, $self->{bat});
5388        foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
5389                next if $_ eq '';
5390                $self->close_directory($bat->{$_}, $p);
5391        }
5392        $self->close_directory($bat->{''}, $p);
5393        $self->SUPER::close_edit($p);
5394        $p->clear;
5395}
5396
5397sub abort_edit {
5398        my ($self) = @_;
5399        $self->SUPER::abort_edit($self->{pool});
5400}
5401
5402sub DESTROY {
5403        my $self = shift;
5404        $self->SUPER::DESTROY(@_);
5405        $self->{pool}->clear;
5406}
5407
5408# this drives the editor
5409sub apply_diff {
5410        my ($self) = @_;
5411        my $mods = $self->{mods};
5412        my %o = ( D => 0, C => 1, R => 2, A => 3, M => 4, T => 5 );
5413        my %deletions;
5414
5415        foreach my $m (@$mods) {
5416                if ($m->{chg} eq "D") {
5417                        $deletions{$m->{file_b}} = 1;
5418                }
5419        }
5420
5421        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
5422                my $f = $m->{chg};
5423                if (defined $o{$f}) {
5424                        $self->$f($m, \%deletions);
5425                } else {
5426                        fatal("Invalid change type: $f");
5427                }
5428        }
5429
5430        if (defined($self->{mergeinfo})) {
5431                $self->change_dir_prop($self->{bat}{''}, "svn:mergeinfo",
5432                                       $self->{mergeinfo});
5433        }
5434        $self->rmdirs if $_rmdir;
5435        if (@$mods == 0 && !defined($self->{mergeinfo})) {
5436                $self->abort_edit;
5437        } else {
5438                $self->close_edit;
5439        }
5440        return scalar @$mods;
5441}
5442
5443package Git::SVN::Ra;
5444use vars qw/@ISA $config_dir $_ignore_refs_regex $_log_window_size/;
5445use strict;
5446use warnings;
5447my ($ra_invalid, $can_do_switch, %ignored_err, $RA);
5448
5449BEGIN {
5450        # enforce temporary pool usage for some simple functions
5451        no strict 'refs';
5452        for my $f (qw/rev_proplist get_latest_revnum get_uuid get_repos_root
5453                      get_file/) {
5454                my $SUPER = "SUPER::$f";
5455                *$f = sub {
5456                        my $self = shift;
5457                        my $pool = SVN::Pool->new;
5458                        my @ret = $self->$SUPER(@_,$pool);
5459                        $pool->clear;
5460                        wantarray ? @ret : $ret[0];
5461                };
5462        }
5463}
5464
5465sub _auth_providers () {
5466        my @rv = (
5467          SVN::Client::get_simple_provider(),
5468          SVN::Client::get_ssl_server_trust_file_provider(),
5469          SVN::Client::get_simple_prompt_provider(
5470            \&Git::SVN::Prompt::simple, 2),
5471          SVN::Client::get_ssl_client_cert_file_provider(),
5472          SVN::Client::get_ssl_client_cert_prompt_provider(
5473            \&Git::SVN::Prompt::ssl_client_cert, 2),
5474          SVN::Client::get_ssl_client_cert_pw_file_provider(),
5475          SVN::Client::get_ssl_client_cert_pw_prompt_provider(
5476            \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
5477          SVN::Client::get_username_provider(),
5478          SVN::Client::get_ssl_server_trust_prompt_provider(
5479            \&Git::SVN::Prompt::ssl_server_trust),
5480          SVN::Client::get_username_prompt_provider(
5481            \&Git::SVN::Prompt::username, 2)
5482        );
5483
5484        # earlier 1.6.x versions would segfault, and <= 1.5.x didn't have
5485        # this function
5486        if (::compare_svn_version('1.6.12') > 0) {
5487                my $config = SVN::Core::config_get_config($config_dir);
5488                my ($p, @a);
5489                # config_get_config returns all config files from
5490                # ~/.subversion, auth_get_platform_specific_client_providers
5491                # just wants the config "file".
5492                @a = ($config->{'config'}, undef);
5493                $p = SVN::Core::auth_get_platform_specific_client_providers(@a);
5494                # Insert the return value from
5495                # auth_get_platform_specific_providers
5496                unshift @rv, @$p;
5497        }
5498        \@rv;
5499}
5500
5501sub escape_uri_only {
5502        my ($uri) = @_;
5503        my @tmp;
5504        foreach (split m{/}, $uri) {
5505                s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
5506                push @tmp, $_;
5507        }
5508        join('/', @tmp);
5509}
5510
5511sub escape_url {
5512        my ($url) = @_;
5513        if ($url =~ m#^(https?)://([^/]+)(.*)$#) {
5514                my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
5515                $url = "$scheme://$domain$uri";
5516        }
5517        $url;
5518}
5519
5520sub new {
5521        my ($class, $url) = @_;
5522        $url =~ s!/+$!!;
5523        return $RA if ($RA && $RA->{url} eq $url);
5524
5525        ::_req_svn();
5526
5527        SVN::_Core::svn_config_ensure($config_dir, undef);
5528        my ($baton, $callbacks) = SVN::Core::auth_open_helper(_auth_providers);
5529        my $config = SVN::Core::config_get_config($config_dir);
5530        $RA = undef;
5531        my $dont_store_passwords = 1;
5532        my $conf_t = ${$config}{'config'};
5533        {
5534                no warnings 'once';
5535                # The usage of $SVN::_Core::SVN_CONFIG_* variables
5536                # produces warnings that variables are used only once.
5537                # I had not found the better way to shut them up, so
5538                # the warnings of type 'once' are disabled in this block.
5539                if (SVN::_Core::svn_config_get_bool($conf_t,
5540                    $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
5541                    $SVN::_Core::SVN_CONFIG_OPTION_STORE_PASSWORDS,
5542                    1) == 0) {
5543                        SVN::_Core::svn_auth_set_parameter($baton,
5544                            $SVN::_Core::SVN_AUTH_PARAM_DONT_STORE_PASSWORDS,
5545                            bless (\$dont_store_passwords, "_p_void"));
5546                }
5547                if (SVN::_Core::svn_config_get_bool($conf_t,
5548                    $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
5549                    $SVN::_Core::SVN_CONFIG_OPTION_STORE_AUTH_CREDS,
5550                    1) == 0) {
5551                        $Git::SVN::Prompt::_no_auth_cache = 1;
5552                }
5553        } # no warnings 'once'
5554        my $self = SVN::Ra->new(url => escape_url($url), auth => $baton,
5555                              config => $config,
5556                              pool => SVN::Pool->new,
5557                              auth_provider_callbacks => $callbacks);
5558        $self->{url} = $url;
5559        $self->{svn_path} = $url;
5560        $self->{repos_root} = $self->get_repos_root;
5561        $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
5562        $self->{cache} = { check_path => { r => 0, data => {} },
5563                           get_dir => { r => 0, data => {} } };
5564        $RA = bless $self, $class;
5565}
5566
5567sub check_path {
5568        my ($self, $path, $r) = @_;
5569        my $cache = $self->{cache}->{check_path};
5570        if ($r == $cache->{r} && exists $cache->{data}->{$path}) {
5571                return $cache->{data}->{$path};
5572        }
5573        my $pool = SVN::Pool->new;
5574        my $t = $self->SUPER::check_path($path, $r, $pool);
5575        $pool->clear;
5576        if ($r != $cache->{r}) {
5577                %{$cache->{data}} = ();
5578                $cache->{r} = $r;
5579        }
5580        $cache->{data}->{$path} = $t;
5581}
5582
5583sub get_dir {
5584        my ($self, $dir, $r) = @_;
5585        my $cache = $self->{cache}->{get_dir};
5586        if ($r == $cache->{r}) {
5587                if (my $x = $cache->{data}->{$dir}) {
5588                        return wantarray ? @$x : $x->[0];
5589                }
5590        }
5591        my $pool = SVN::Pool->new;
5592        my ($d, undef, $props) = $self->SUPER::get_dir($dir, $r, $pool);
5593        my %dirents = map { $_ => { kind => $d->{$_}->kind } } keys %$d;
5594        $pool->clear;
5595        if ($r != $cache->{r}) {
5596                %{$cache->{data}} = ();
5597                $cache->{r} = $r;
5598        }
5599        $cache->{data}->{$dir} = [ \%dirents, $r, $props ];
5600        wantarray ? (\%dirents, $r, $props) : \%dirents;
5601}
5602
5603sub DESTROY {
5604        # do not call the real DESTROY since we store ourselves in $RA
5605}
5606
5607# get_log(paths, start, end, limit,
5608#         discover_changed_paths, strict_node_history, receiver)
5609sub get_log {
5610        my ($self, @args) = @_;
5611        my $pool = SVN::Pool->new;
5612
5613        # svn_log_changed_path_t objects passed to get_log are likely to be
5614        # overwritten even if only the refs are copied to an external variable,
5615        # so we should dup the structures in their entirety.  Using an
5616        # externally passed pool (instead of our temporary and quickly cleared
5617        # pool in Git::SVN::Ra) does not help matters at all...
5618        my $receiver = pop @args;
5619        my $prefix = "/".$self->{svn_path};
5620        $prefix =~ s#/+($)##;
5621        my $prefix_regex = qr#^\Q$prefix\E#;
5622        push(@args, sub {
5623                my ($paths) = $_[0];
5624                return &$receiver(@_) unless $paths;
5625                $_[0] = ();
5626                foreach my $p (keys %$paths) {
5627                        my $i = $paths->{$p};
5628                        # Make path relative to our url, not repos_root
5629                        $p =~ s/$prefix_regex//;
5630                        my %s = map { $_ => $i->$_; }
5631                                qw/copyfrom_path copyfrom_rev action/;
5632                        if ($s{'copyfrom_path'}) {
5633                                $s{'copyfrom_path'} =~ s/$prefix_regex//;
5634                        }
5635                        $_[0]{$p} = \%s;
5636                }
5637                &$receiver(@_);
5638        });
5639
5640
5641        # the limit parameter was not supported in SVN 1.1.x, so we
5642        # drop it.  Therefore, the receiver callback passed to it
5643        # is made aware of this limitation by being wrapped if
5644        # the limit passed to is being wrapped.
5645        if (::compare_svn_version('1.2.0') <= 0) {
5646                my $limit = splice(@args, 3, 1);
5647                if ($limit > 0) {
5648                        my $receiver = pop @args;
5649                        push(@args, sub { &$receiver(@_) if (--$limit >= 0) });
5650                }
5651        }
5652        my $ret = $self->SUPER::get_log(@args, $pool);
5653        $pool->clear;
5654        $ret;
5655}
5656
5657sub trees_match {
5658        my ($self, $url1, $rev1, $url2, $rev2) = @_;
5659        my $ctx = SVN::Client->new(auth => _auth_providers);
5660        my $out = IO::File->new_tmpfile;
5661
5662        # older SVN (1.1.x) doesn't take $pool as the last parameter for
5663        # $ctx->diff(), so we'll create a default one
5664        my $pool = SVN::Pool->new_default_sub;
5665
5666        $ra_invalid = 1; # this will open a new SVN::Ra connection to $url1
5667        $ctx->diff([], $url1, $rev1, $url2, $rev2, 1, 1, 0, $out, $out);
5668        $out->flush;
5669        my $ret = (($out->stat)[7] == 0);
5670        close $out or croak $!;
5671
5672        $ret;
5673}
5674
5675sub get_commit_editor {
5676        my ($self, $log, $cb, $pool) = @_;
5677
5678        my @lock = (::compare_svn_version('1.2.0') >= 0) ? (undef, 0) : ();
5679        $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
5680}
5681
5682sub gs_do_update {
5683        my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
5684        my $new = ($rev_a == $rev_b);
5685        my $path = $gs->{path};
5686
5687        if ($new && -e $gs->{index}) {
5688                unlink $gs->{index} or die
5689                  "Couldn't unlink index: $gs->{index}: $!\n";
5690        }
5691        my $pool = SVN::Pool->new;
5692        $editor->set_path_strip($path);
5693        my (@pc) = split m#/#, $path;
5694        my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
5695                                        1, $editor, $pool);
5696        my @lock = (::compare_svn_version('1.2.0') >= 0) ? (undef) : ();
5697
5698        # Since we can't rely on svn_ra_reparent being available, we'll
5699        # just have to do some magic with set_path to make it so
5700        # we only want a partial path.
5701        my $sp = '';
5702        my $final = join('/', @pc);
5703        while (@pc) {
5704                $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
5705                $sp .= '/' if length $sp;
5706                $sp .= shift @pc;
5707        }
5708        die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
5709
5710        $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
5711
5712        $reporter->finish_report($pool);
5713        $pool->clear;
5714        $editor->{git_commit_ok};
5715}
5716
5717# this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
5718# svn_ra_reparent didn't work before 1.4)
5719sub gs_do_switch {
5720        my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
5721        my $path = $gs->{path};
5722        my $pool = SVN::Pool->new;
5723
5724        my $full_url = $self->{url};
5725        my $old_url = $full_url;
5726        $full_url .= '/' . $path if length $path;
5727        my ($ra, $reparented);
5728
5729        if ($old_url =~ m#^svn(\+ssh)?://# ||
5730            ($full_url =~ m#^https?://# &&
5731             escape_url($full_url) ne $full_url)) {
5732                $_[0] = undef;
5733                $self = undef;
5734                $RA = undef;
5735                $ra = Git::SVN::Ra->new($full_url);
5736                $ra_invalid = 1;
5737        } elsif ($old_url ne $full_url) {
5738                SVN::_Ra::svn_ra_reparent($self->{session}, $full_url, $pool);
5739                $self->{url} = $full_url;
5740                $reparented = 1;
5741        }
5742
5743        $ra ||= $self;
5744        $url_b = escape_url($url_b);
5745        my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
5746        my @lock = (::compare_svn_version('1.2.0') >= 0) ? (undef) : ();
5747        $reporter->set_path('', $rev_a, 0, @lock, $pool);
5748        $reporter->finish_report($pool);
5749
5750        if ($reparented) {
5751                SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
5752                $self->{url} = $old_url;
5753        }
5754
5755        $pool->clear;
5756        $editor->{git_commit_ok};
5757}
5758
5759sub longest_common_path {
5760        my ($gsv, $globs) = @_;
5761        my %common;
5762        my $common_max = scalar @$gsv;
5763
5764        foreach my $gs (@$gsv) {
5765                my @tmp = split m#/#, $gs->{path};
5766                my $p = '';
5767                foreach (@tmp) {
5768                        $p .= length($p) ? "/$_" : $_;
5769                        $common{$p} ||= 0;
5770                        $common{$p}++;
5771                }
5772        }
5773        $globs ||= [];
5774        $common_max += scalar @$globs;
5775        foreach my $glob (@$globs) {
5776                my @tmp = split m#/#, $glob->{path}->{left};
5777                my $p = '';
5778                foreach (@tmp) {
5779                        $p .= length($p) ? "/$_" : $_;
5780                        $common{$p} ||= 0;
5781                        $common{$p}++;
5782                }
5783        }
5784
5785        my $longest_path = '';
5786        foreach (sort {length $b <=> length $a} keys %common) {
5787                if ($common{$_} == $common_max) {
5788                        $longest_path = $_;
5789                        last;
5790                }
5791        }
5792        $longest_path;
5793}
5794
5795sub gs_fetch_loop_common {
5796        my ($self, $base, $head, $gsv, $globs) = @_;
5797        return if ($base > $head);
5798        my $inc = $_log_window_size;
5799        my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
5800        my $longest_path = longest_common_path($gsv, $globs);
5801        my $ra_url = $self->{url};
5802        my $find_trailing_edge;
5803        while (1) {
5804                my %revs;
5805                my $err;
5806                my $err_handler = $SVN::Error::handler;
5807                $SVN::Error::handler = sub {
5808                        ($err) = @_;
5809                        skip_unknown_revs($err);
5810                };
5811                sub _cb {
5812                        my ($paths, $r, $author, $date, $log) = @_;
5813                        [ $paths,
5814                          { author => $author, date => $date, log => $log } ];
5815                }
5816                $self->get_log([$longest_path], $min, $max, 0, 1, 1,
5817                               sub { $revs{$_[1]} = _cb(@_) });
5818                if ($err) {
5819                        print "Checked through r$max\r";
5820                } else {
5821                        $find_trailing_edge = 1;
5822                }
5823                if ($err and $find_trailing_edge) {
5824                        print STDERR "Path '$longest_path' ",
5825                                     "was probably deleted:\n",
5826                                     $err->expanded_message,
5827                                     "\nWill attempt to follow ",
5828                                     "revisions r$min .. r$max ",
5829                                     "committed before the deletion\n";
5830                        my $hi = $max;
5831                        while (--$hi >= $min) {
5832                                my $ok;
5833                                $self->get_log([$longest_path], $min, $hi,
5834                                               0, 1, 1, sub {
5835                                               $ok = $_[1];
5836                                               $revs{$_[1]} = _cb(@_) });
5837                                if ($ok) {
5838                                        print STDERR "r$min .. r$ok OK\n";
5839                                        last;
5840                                }
5841                        }
5842                        $find_trailing_edge = 0;
5843                }
5844                $SVN::Error::handler = $err_handler;
5845
5846                my %exists = map { $_->{path} => $_ } @$gsv;
5847                foreach my $r (sort {$a <=> $b} keys %revs) {
5848                        my ($paths, $logged) = @{$revs{$r}};
5849
5850                        foreach my $gs ($self->match_globs(\%exists, $paths,
5851                                                           $globs, $r)) {
5852                                if ($gs->rev_map_max >= $r) {
5853                                        next;
5854                                }
5855                                next unless $gs->match_paths($paths, $r);
5856                                $gs->{logged_rev_props} = $logged;
5857                                if (my $last_commit = $gs->last_commit) {
5858                                        $gs->assert_index_clean($last_commit);
5859                                }
5860                                my $log_entry = $gs->do_fetch($paths, $r);
5861                                if ($log_entry) {
5862                                        $gs->do_git_commit($log_entry);
5863                                }
5864                                $INDEX_FILES{$gs->{index}} = 1;
5865                        }
5866                        foreach my $g (@$globs) {
5867                                my $k = "svn-remote.$g->{remote}." .
5868                                        "$g->{t}-maxRev";
5869                                Git::SVN::tmp_config($k, $r);
5870                        }
5871                        if ($ra_invalid) {
5872                                $_[0] = undef;
5873                                $self = undef;
5874                                $RA = undef;
5875                                $self = Git::SVN::Ra->new($ra_url);
5876                                $ra_invalid = undef;
5877                        }
5878                }
5879                # pre-fill the .rev_db since it'll eventually get filled in
5880                # with '0' x40 if something new gets committed
5881                foreach my $gs (@$gsv) {
5882                        next if $gs->rev_map_max >= $max;
5883                        next if defined $gs->rev_map_get($max);
5884                        $gs->rev_map_set($max, 0 x40);
5885                }
5886                foreach my $g (@$globs) {
5887                        my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
5888                        Git::SVN::tmp_config($k, $max);
5889                }
5890                last if $max >= $head;
5891                $min = $max + 1;
5892                $max += $inc;
5893                $max = $head if ($max > $head);
5894        }
5895        Git::SVN::gc();
5896}
5897
5898sub get_dir_globbed {
5899        my ($self, $left, $depth, $r) = @_;
5900
5901        my @x = eval { $self->get_dir($left, $r) };
5902        return unless scalar @x == 3;
5903        my $dirents = $x[0];
5904        my @finalents;
5905        foreach my $de (keys %$dirents) {
5906                next if $dirents->{$de}->{kind} != $SVN::Node::dir;
5907                if ($depth > 1) {
5908                        my @args = ("$left/$de", $depth - 1, $r);
5909                        foreach my $dir ($self->get_dir_globbed(@args)) {
5910                                push @finalents, "$de/$dir";
5911                        }
5912                } else {
5913                        push @finalents, $de;
5914                }
5915        }
5916        @finalents;
5917}
5918
5919# return value: 0 -- don't ignore, 1 -- ignore
5920sub is_ref_ignored {
5921        my ($g, $p) = @_;
5922        my $refname = $g->{ref}->full_path($p);
5923        return 1 if defined($g->{ignore_refs_regex}) &&
5924                    $refname =~ m!$g->{ignore_refs_regex}!;
5925        return 0 unless defined($_ignore_refs_regex);
5926        return 1 if $refname =~ m!$_ignore_refs_regex!o;
5927        return 0;
5928}
5929
5930sub match_globs {
5931        my ($self, $exists, $paths, $globs, $r) = @_;
5932
5933        sub get_dir_check {
5934                my ($self, $exists, $g, $r) = @_;
5935
5936                my @dirs = $self->get_dir_globbed($g->{path}->{left},
5937                                                  $g->{path}->{depth},
5938                                                  $r);
5939
5940                foreach my $de (@dirs) {
5941                        my $p = $g->{path}->full_path($de);
5942                        next if $exists->{$p};
5943                        next if (length $g->{path}->{right} &&
5944                                 ($self->check_path($p, $r) !=
5945                                  $SVN::Node::dir));
5946                        next unless $p =~ /$g->{path}->{regex}/;
5947                        $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
5948                                         $g->{ref}->full_path($de), 1);
5949                }
5950        }
5951        foreach my $g (@$globs) {
5952                if (my $path = $paths->{"/$g->{path}->{left}"}) {
5953                        if ($path->{action} =~ /^[AR]$/) {
5954                                get_dir_check($self, $exists, $g, $r);
5955                        }
5956                }
5957                foreach (keys %$paths) {
5958                        if (/$g->{path}->{left_regex}/ &&
5959                            !/$g->{path}->{regex}/) {
5960                                next if $paths->{$_}->{action} !~ /^[AR]$/;
5961                                get_dir_check($self, $exists, $g, $r);
5962                        }
5963                        next unless /$g->{path}->{regex}/;
5964                        my $p = $1;
5965                        my $pathname = $g->{path}->full_path($p);
5966                        next if is_ref_ignored($g, $p);
5967                        next if $exists->{$pathname};
5968                        next if ($self->check_path($pathname, $r) !=
5969                                 $SVN::Node::dir);
5970                        $exists->{$pathname} = Git::SVN->init(
5971                                              $self->{url}, $pathname, undef,
5972                                              $g->{ref}->full_path($p), 1);
5973                }
5974                my $c = '';
5975                foreach (split m#/#, $g->{path}->{left}) {
5976                        $c .= "/$_";
5977                        next unless ($paths->{$c} &&
5978                                     ($paths->{$c}->{action} =~ /^[AR]$/));
5979                        get_dir_check($self, $exists, $g, $r);
5980                }
5981        }
5982        values %$exists;
5983}
5984
5985sub minimize_url {
5986        my ($self) = @_;
5987        return $self->{url} if ($self->{url} eq $self->{repos_root});
5988        my $url = $self->{repos_root};
5989        my @components = split(m!/!, $self->{svn_path});
5990        my $c = '';
5991        do {
5992                $url .= "/$c" if length $c;
5993                eval {
5994                        my $ra = (ref $self)->new($url);
5995                        my $latest = $ra->get_latest_revnum;
5996                        $ra->get_log("", $latest, 0, 1, 0, 1, sub {});
5997                };
5998        } while ($@ && ($c = shift @components));
5999        $url;
6000}
6001
6002sub can_do_switch {
6003        my $self = shift;
6004        unless (defined $can_do_switch) {
6005                my $pool = SVN::Pool->new;
6006                my $rep = eval {
6007                        $self->do_switch(1, '', 0, $self->{url},
6008                                         SVN::Delta::Editor->new, $pool);
6009                };
6010                if ($@) {
6011                        $can_do_switch = 0;
6012                } else {
6013                        $rep->abort_report($pool);
6014                        $can_do_switch = 1;
6015                }
6016                $pool->clear;
6017        }
6018        $can_do_switch;
6019}
6020
6021sub skip_unknown_revs {
6022        my ($err) = @_;
6023        my $errno = $err->apr_err();
6024        # Maybe the branch we're tracking didn't
6025        # exist when the repo started, so it's
6026        # not an error if it doesn't, just continue
6027        #
6028        # Wonderfully consistent library, eh?
6029        # 160013 - svn:// and file://
6030        # 175002 - http(s)://
6031        # 175007 - http(s):// (this repo required authorization, too...)
6032        #   More codes may be discovered later...
6033        if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
6034                my $err_key = $err->expanded_message;
6035                # revision numbers change every time, filter them out
6036                $err_key =~ s/\d+/\0/g;
6037                $err_key = "$errno\0$err_key";
6038                unless ($ignored_err{$err_key}) {
6039                        warn "W: Ignoring error from SVN, path probably ",
6040                             "does not exist: ($errno): ",
6041                             $err->expanded_message,"\n";
6042                        warn "W: Do not be alarmed at the above message ",
6043                             "git-svn is just searching aggressively for ",
6044                             "old history.\n",
6045                             "This may take a while on large repositories\n";
6046                        $ignored_err{$err_key} = 1;
6047                }
6048                return;
6049        }
6050        die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
6051}
6052
6053package Git::SVN::Log;
6054use strict;
6055use warnings;
6056use POSIX qw/strftime/;
6057use constant commit_log_separator => ('-' x 72) . "\n";
6058use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
6059            %rusers $show_commit $incremental/;
6060my $l_fmt;
6061
6062sub cmt_showable {
6063        my ($c) = @_;
6064        return 1 if defined $c->{r};
6065
6066        # big commit message got truncated by the 16k pretty buffer in rev-list
6067        if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
6068                                $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
6069                @{$c->{l}} = ();
6070                my @log = command(qw/cat-file commit/, $c->{c});
6071
6072                # shift off the headers
6073                shift @log while ($log[0] ne '');
6074                shift @log;
6075
6076                # TODO: make $c->{l} not have a trailing newline in the future
6077                @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
6078
6079                (undef, $c->{r}, undef) = ::extract_metadata(
6080                                (grep(/^git-svn-id: /, @log))[-1]);
6081        }
6082        return defined $c->{r};
6083}
6084
6085sub log_use_color {
6086        return $color || Git->repository->get_colorbool('color.diff');
6087}
6088
6089sub git_svn_log_cmd {
6090        my ($r_min, $r_max, @args) = @_;
6091        my $head = 'HEAD';
6092        my (@files, @log_opts);
6093        foreach my $x (@args) {
6094                if ($x eq '--' || @files) {
6095                        push @files, $x;
6096                } else {
6097                        if (::verify_ref("$x^0")) {
6098                                $head = $x;
6099                        } else {
6100                                push @log_opts, $x;
6101                        }
6102                }
6103        }
6104
6105        my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
6106        $gs ||= Git::SVN->_new;
6107        my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
6108                   $gs->refname);
6109        push @cmd, '-r' unless $non_recursive;
6110        push @cmd, qw/--raw --name-status/ if $verbose;
6111        push @cmd, '--color' if log_use_color();
6112        push @cmd, @log_opts;
6113        if (defined $r_max && $r_max == $r_min) {
6114                push @cmd, '--max-count=1';
6115                if (my $c = $gs->rev_map_get($r_max)) {
6116                        push @cmd, $c;
6117                }
6118        } elsif (defined $r_max) {
6119                if ($r_max < $r_min) {
6120                        ($r_min, $r_max) = ($r_max, $r_min);
6121                }
6122                my (undef, $c_max) = $gs->find_rev_before($r_max, 1, $r_min);
6123                my (undef, $c_min) = $gs->find_rev_after($r_min, 1, $r_max);
6124                # If there are no commits in the range, both $c_max and $c_min
6125                # will be undefined.  If there is at least 1 commit in the
6126                # range, both will be defined.
6127                return () if !defined $c_min || !defined $c_max;
6128                if ($c_min eq $c_max) {
6129                        push @cmd, '--max-count=1', $c_min;
6130                } else {
6131                        push @cmd, '--boundary', "$c_min..$c_max";
6132                }
6133        }
6134        return (@cmd, @files);
6135}
6136
6137# adapted from pager.c
6138sub config_pager {
6139        if (! -t *STDOUT) {
6140                $ENV{GIT_PAGER_IN_USE} = 'false';
6141                $pager = undef;
6142                return;
6143        }
6144        chomp($pager = command_oneline(qw(var GIT_PAGER)));
6145        if ($pager eq 'cat') {
6146                $pager = undef;
6147        }
6148        $ENV{GIT_PAGER_IN_USE} = defined($pager);
6149}
6150
6151sub run_pager {
6152        return unless defined $pager;
6153        pipe my ($rfd, $wfd) or return;
6154        defined(my $pid = fork) or ::fatal "Can't fork: $!";
6155        if (!$pid) {
6156                open STDOUT, '>&', $wfd or
6157                                     ::fatal "Can't redirect to stdout: $!";
6158                return;
6159        }
6160        open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!";
6161        $ENV{LESS} ||= 'FRSX';
6162        exec $pager or ::fatal "Can't run pager: $! ($pager)";
6163}
6164
6165sub format_svn_date {
6166        my $t = shift || time;
6167        my $gmoff = Git::SVN::get_tz($t);
6168        return strftime("%Y-%m-%d %H:%M:%S $gmoff (%a, %d %b %Y)", localtime($t));
6169}
6170
6171sub parse_git_date {
6172        my ($t, $tz) = @_;
6173        # Date::Parse isn't in the standard Perl distro :(
6174        if ($tz =~ s/^\+//) {
6175                $t += tz_to_s_offset($tz);
6176        } elsif ($tz =~ s/^\-//) {
6177                $t -= tz_to_s_offset($tz);
6178        }
6179        return $t;
6180}
6181
6182sub set_local_timezone {
6183        if (defined $TZ) {
6184                $ENV{TZ} = $TZ;
6185        } else {
6186                delete $ENV{TZ};
6187        }
6188}
6189
6190sub tz_to_s_offset {
6191        my ($tz) = @_;
6192        $tz =~ s/(\d\d)$//;
6193        return ($1 * 60) + ($tz * 3600);
6194}
6195
6196sub get_author_info {
6197        my ($dest, $author, $t, $tz) = @_;
6198        $author =~ s/(?:^\s*|\s*$)//g;
6199        $dest->{a_raw} = $author;
6200        my $au;
6201        if ($::_authors) {
6202                $au = $rusers{$author} || undef;
6203        }
6204        if (!$au) {
6205                ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
6206        }
6207        $dest->{t} = $t;
6208        $dest->{tz} = $tz;
6209        $dest->{a} = $au;
6210        $dest->{t_utc} = parse_git_date($t, $tz);
6211}
6212
6213sub process_commit {
6214        my ($c, $r_min, $r_max, $defer) = @_;
6215        if (defined $r_min && defined $r_max) {
6216                if ($r_min == $c->{r} && $r_min == $r_max) {
6217                        show_commit($c);
6218                        return 0;
6219                }
6220                return 1 if $r_min == $r_max;
6221                if ($r_min < $r_max) {
6222                        # we need to reverse the print order
6223                        return 0 if (defined $limit && --$limit < 0);
6224                        push @$defer, $c;
6225                        return 1;
6226                }
6227                if ($r_min != $r_max) {
6228                        return 1 if ($r_min < $c->{r});
6229                        return 1 if ($r_max > $c->{r});
6230                }
6231        }
6232        return 0 if (defined $limit && --$limit < 0);
6233        show_commit($c);
6234        return 1;
6235}
6236
6237sub show_commit {
6238        my $c = shift;
6239        if ($oneline) {
6240                my $x = "\n";
6241                if (my $l = $c->{l}) {
6242                        while ($l->[0] =~ /^\s*$/) { shift @$l }
6243                        $x = $l->[0];
6244                }
6245                $l_fmt ||= 'A' . length($c->{r});
6246                print 'r',pack($l_fmt, $c->{r}),' | ';
6247                print "$c->{c} | " if $show_commit;
6248                print $x;
6249        } else {
6250                show_commit_normal($c);
6251        }
6252}
6253
6254sub show_commit_changed_paths {
6255        my ($c) = @_;
6256        return unless $c->{changed};
6257        print "Changed paths:\n", @{$c->{changed}};
6258}
6259
6260sub show_commit_normal {
6261        my ($c) = @_;
6262        print commit_log_separator, "r$c->{r} | ";
6263        print "$c->{c} | " if $show_commit;
6264        print "$c->{a} | ", format_svn_date($c->{t_utc}), ' | ';
6265        my $nr_line = 0;
6266
6267        if (my $l = $c->{l}) {
6268                while ($l->[$#$l] eq "\n" && $#$l > 0
6269                                          && $l->[($#$l - 1)] eq "\n") {
6270                        pop @$l;
6271                }
6272                $nr_line = scalar @$l;
6273                if (!$nr_line) {
6274                        print "1 line\n\n\n";
6275                } else {
6276                        if ($nr_line == 1) {
6277                                $nr_line = '1 line';
6278                        } else {
6279                                $nr_line .= ' lines';
6280                        }
6281                        print $nr_line, "\n";
6282                        show_commit_changed_paths($c);
6283                        print "\n";
6284                        print $_ foreach @$l;
6285                }
6286        } else {
6287                print "1 line\n";
6288                show_commit_changed_paths($c);
6289                print "\n";
6290
6291        }
6292        foreach my $x (qw/raw stat diff/) {
6293                if ($c->{$x}) {
6294                        print "\n";
6295                        print $_ foreach @{$c->{$x}}
6296                }
6297        }
6298}
6299
6300sub cmd_show_log {
6301        my (@args) = @_;
6302        my ($r_min, $r_max);
6303        my $r_last = -1; # prevent dupes
6304        set_local_timezone();
6305        if (defined $::_revision) {
6306                if ($::_revision =~ /^(\d+):(\d+)$/) {
6307                        ($r_min, $r_max) = ($1, $2);
6308                } elsif ($::_revision =~ /^\d+$/) {
6309                        $r_min = $r_max = $::_revision;
6310                } else {
6311                        ::fatal "-r$::_revision is not supported, use ",
6312                                "standard 'git log' arguments instead";
6313                }
6314        }
6315
6316        config_pager();
6317        @args = git_svn_log_cmd($r_min, $r_max, @args);
6318        if (!@args) {
6319                print commit_log_separator unless $incremental || $oneline;
6320                return;
6321        }
6322        my $log = command_output_pipe(@args);
6323        run_pager();
6324        my (@k, $c, $d, $stat);
6325        my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
6326        while (<$log>) {
6327                if (/^${esc_color}commit (?:- )?($::sha1_short)/o) {
6328                        my $cmt = $1;
6329                        if ($c && cmt_showable($c) && $c->{r} != $r_last) {
6330                                $r_last = $c->{r};
6331                                process_commit($c, $r_min, $r_max, \@k) or
6332                                                                goto out;
6333                        }
6334                        $d = undef;
6335                        $c = { c => $cmt };
6336                } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
6337                        get_author_info($c, $1, $2, $3);
6338                } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
6339                        # ignore
6340                } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
6341                        push @{$c->{raw}}, $_;
6342                } elsif (/^${esc_color}[ACRMDT]\t/) {
6343                        # we could add $SVN->{svn_path} here, but that requires
6344                        # remote access at the moment (repo_path_split)...
6345                        s#^(${esc_color})([ACRMDT])\t#$1   $2 #o;
6346                        push @{$c->{changed}}, $_;
6347                } elsif (/^${esc_color}diff /o) {
6348                        $d = 1;
6349                        push @{$c->{diff}}, $_;
6350                } elsif ($d) {
6351                        push @{$c->{diff}}, $_;
6352                } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
6353                          $esc_color*[\+\-]*$esc_color$/x) {
6354                        $stat = 1;
6355                        push @{$c->{stat}}, $_;
6356                } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
6357                        push @{$c->{stat}}, $_;
6358                        $stat = undef;
6359                } elsif (/^${esc_color}    (git-svn-id:.+)$/o) {
6360                        ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
6361                } elsif (s/^${esc_color}    //o) {
6362                        push @{$c->{l}}, $_;
6363                }
6364        }
6365        if ($c && defined $c->{r} && $c->{r} != $r_last) {
6366                $r_last = $c->{r};
6367                process_commit($c, $r_min, $r_max, \@k);
6368        }
6369        if (@k) {
6370                ($r_min, $r_max) = ($r_max, $r_min);
6371                process_commit($_, $r_min, $r_max) foreach reverse @k;
6372        }
6373out:
6374        close $log;
6375        print commit_log_separator unless $incremental || $oneline;
6376}
6377
6378sub cmd_blame {
6379        my $path = pop;
6380
6381        config_pager();
6382        run_pager();
6383
6384        my ($fh, $ctx, $rev);
6385
6386        if ($_git_format) {
6387                ($fh, $ctx) = command_output_pipe('blame', @_, $path);
6388                while (my $line = <$fh>) {
6389                        if ($line =~ /^\^?([[:xdigit:]]+)\s/) {
6390                                # Uncommitted edits show up as a rev ID of
6391                                # all zeros, which we can't look up with
6392                                # cmt_metadata
6393                                if ($1 !~ /^0+$/) {
6394                                        (undef, $rev, undef) =
6395                                                ::cmt_metadata($1);
6396                                        $rev = '0' if (!$rev);
6397                                } else {
6398                                        $rev = '0';
6399                                }
6400                                $rev = sprintf('%-10s', $rev);
6401                                $line =~ s/^\^?[[:xdigit:]]+(\s)/$rev$1/;
6402                        }
6403                        print $line;
6404                }
6405        } else {
6406                ($fh, $ctx) = command_output_pipe('blame', '-p', @_, 'HEAD',
6407                                                  '--', $path);
6408                my ($sha1);
6409                my %authors;
6410                my @buffer;
6411                my %dsha; #distinct sha keys
6412
6413                while (my $line = <$fh>) {
6414                        push @buffer, $line;
6415                        if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
6416                                $dsha{$1} = 1;
6417                        }
6418                }
6419
6420                my $s2r = ::cmt_sha2rev_batch([keys %dsha]);
6421
6422                foreach my $line (@buffer) {
6423                        if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
6424                                $rev = $s2r->{$1};
6425                                $rev = '0' if (!$rev)
6426                        }
6427                        elsif ($line =~ /^author (.*)/) {
6428                                $authors{$rev} = $1;
6429                                $authors{$rev} =~ s/\s/_/g;
6430                        }
6431                        elsif ($line =~ /^\t(.*)$/) {
6432                                printf("%6s %10s %s\n", $rev, $authors{$rev}, $1);
6433                        }
6434                }
6435        }
6436        command_close_pipe($fh, $ctx);
6437}
6438
6439package Git::SVN::Migration;
6440# these version numbers do NOT correspond to actual version numbers
6441# of git nor git-svn.  They are just relative.
6442#
6443# v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
6444#
6445# v1 layout: .git/$id/info/url, refs/remotes/$id
6446#
6447# v2 layout: .git/svn/$id/info/url, refs/remotes/$id
6448#
6449# v3 layout: .git/svn/$id, refs/remotes/$id
6450#            - info/url may remain for backwards compatibility
6451#            - this is what we migrate up to this layout automatically,
6452#            - this will be used by git svn init on single branches
6453# v3.1 layout (auto migrated):
6454#            - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
6455#              for backwards compatibility
6456#
6457# v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
6458#            - this is only created for newly multi-init-ed
6459#              repositories.  Similar in spirit to the
6460#              --use-separate-remotes option in git-clone (now default)
6461#            - we do not automatically migrate to this (following
6462#              the example set by core git)
6463#
6464# v5 layout: .rev_db.$UUID => .rev_map.$UUID
6465#            - newer, more-efficient format that uses 24-bytes per record
6466#              with no filler space.
6467#            - use xxd -c24 < .rev_map.$UUID to view and debug
6468#            - This is a one-way migration, repositories updated to the
6469#              new format will not be able to use old git-svn without
6470#              rebuilding the .rev_db.  Rebuilding the rev_db is not
6471#              possible if noMetadata or useSvmProps are set; but should
6472#              be no problem for users that use the (sensible) defaults.
6473use strict;
6474use warnings;
6475use Carp qw/croak/;
6476use File::Path qw/mkpath/;
6477use File::Basename qw/dirname basename/;
6478use vars qw/$_minimize/;
6479
6480sub migrate_from_v0 {
6481        my $git_dir = $ENV{GIT_DIR};
6482        return undef unless -d $git_dir;
6483        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
6484        my $migrated = 0;
6485        while (<$fh>) {
6486                chomp;
6487                my ($id, $orig_ref) = ($_, $_);
6488                next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
6489                next unless -f "$git_dir/$id/info/url";
6490                my $new_ref = "refs/remotes/$id";
6491                if (::verify_ref("$new_ref^0")) {
6492                        print STDERR "W: $orig_ref is probably an old ",
6493                                     "branch used by an ancient version of ",
6494                                     "git-svn.\n",
6495                                     "However, $new_ref also exists.\n",
6496                                     "We will not be able ",
6497                                     "to use this branch until this ",
6498                                     "ambiguity is resolved.\n";
6499                        next;
6500                }
6501                print STDERR "Migrating from v0 layout...\n" if !$migrated;
6502                print STDERR "Renaming ref: $orig_ref => $new_ref\n";
6503                command_noisy('update-ref', $new_ref, $orig_ref);
6504                command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
6505                $migrated++;
6506        }
6507        command_close_pipe($fh, $ctx);
6508        print STDERR "Done migrating from v0 layout...\n" if $migrated;
6509        $migrated;
6510}
6511
6512sub migrate_from_v1 {
6513        my $git_dir = $ENV{GIT_DIR};
6514        my $migrated = 0;
6515        return $migrated unless -d $git_dir;
6516        my $svn_dir = "$git_dir/svn";
6517
6518        # just in case somebody used 'svn' as their $id at some point...
6519        return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
6520
6521        print STDERR "Migrating from a git-svn v1 layout...\n";
6522        mkpath([$svn_dir]);
6523        print STDERR "Data from a previous version of git-svn exists, but\n\t",
6524                     "$svn_dir\n\t(required for this version ",
6525                     "($::VERSION) of git-svn) does not exist.\n";
6526        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
6527        while (<$fh>) {
6528                my $x = $_;
6529                next unless $x =~ s#^refs/remotes/##;
6530                chomp $x;
6531                next unless -f "$git_dir/$x/info/url";
6532                my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
6533                next unless $u;
6534                my $dn = dirname("$git_dir/svn/$x");
6535                mkpath([$dn]) unless -d $dn;
6536                if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
6537                        mkpath(["$git_dir/svn/svn"]);
6538                        print STDERR " - $git_dir/$x/info => ",
6539                                        "$git_dir/svn/$x/info\n";
6540                        rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
6541                               croak "$!: $x";
6542                        # don't worry too much about these, they probably
6543                        # don't exist with repos this old (save for index,
6544                        # and we can easily regenerate that)
6545                        foreach my $f (qw/unhandled.log index .rev_db/) {
6546                                rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
6547                        }
6548                } else {
6549                        print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
6550                        rename "$git_dir/$x", "$git_dir/svn/$x" or
6551                               croak "$!: $x";
6552                }
6553                $migrated++;
6554        }
6555        command_close_pipe($fh, $ctx);
6556        print STDERR "Done migrating from a git-svn v1 layout\n";
6557        $migrated;
6558}
6559
6560sub read_old_urls {
6561        my ($l_map, $pfx, $path) = @_;
6562        my @dir;
6563        foreach (<$path/*>) {
6564                if (-r "$_/info/url") {
6565                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
6566                        my $ref_id = $pfx . basename $_;
6567                        my $url = ::file_to_s("$_/info/url");
6568                        $l_map->{$ref_id} = $url;
6569                } elsif (-d $_) {
6570                        push @dir, $_;
6571                }
6572        }
6573        foreach (@dir) {
6574                my $x = $_;
6575                $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
6576                read_old_urls($l_map, $x, $_);
6577        }
6578}
6579
6580sub migrate_from_v2 {
6581        my @cfg = command(qw/config -l/);
6582        return if grep /^svn-remote\..+\.url=/, @cfg;
6583        my %l_map;
6584        read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
6585        my $migrated = 0;
6586
6587        foreach my $ref_id (sort keys %l_map) {
6588                eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
6589                if ($@) {
6590                        Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
6591                }
6592                $migrated++;
6593        }
6594        $migrated;
6595}
6596
6597sub minimize_connections {
6598        my $r = Git::SVN::read_all_remotes();
6599        my $new_urls = {};
6600        my $root_repos = {};
6601        foreach my $repo_id (keys %$r) {
6602                my $url = $r->{$repo_id}->{url} or next;
6603                my $fetch = $r->{$repo_id}->{fetch} or next;
6604                my $ra = Git::SVN::Ra->new($url);
6605
6606                # skip existing cases where we already connect to the root
6607                if (($ra->{url} eq $ra->{repos_root}) ||
6608                    ($ra->{repos_root} eq $repo_id)) {
6609                        $root_repos->{$ra->{url}} = $repo_id;
6610                        next;
6611                }
6612
6613                my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
6614                my $root_path = $ra->{url};
6615                $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
6616                foreach my $path (keys %$fetch) {
6617                        my $ref_id = $fetch->{$path};
6618                        my $gs = Git::SVN->new($ref_id, $repo_id, $path);
6619
6620                        # make sure we can read when connecting to
6621                        # a higher level of a repository
6622                        my ($last_rev, undef) = $gs->last_rev_commit;
6623                        if (!defined $last_rev) {
6624                                $last_rev = eval {
6625                                        $root_ra->get_latest_revnum;
6626                                };
6627                                next if $@;
6628                        }
6629                        my $new = $root_path;
6630                        $new .= length $path ? "/$path" : '';
6631                        eval {
6632                                $root_ra->get_log([$new], $last_rev, $last_rev,
6633                                                  0, 0, 1, sub { });
6634                        };
6635                        next if $@;
6636                        $new_urls->{$ra->{repos_root}}->{$new} =
6637                                { ref_id => $ref_id,
6638                                  old_repo_id => $repo_id,
6639                                  old_path => $path };
6640                }
6641        }
6642
6643        my @emptied;
6644        foreach my $url (keys %$new_urls) {
6645                # see if we can re-use an existing [svn-remote "repo_id"]
6646                # instead of creating a(n ugly) new section:
6647                my $repo_id = $root_repos->{$url} || $url;
6648
6649                my $fetch = $new_urls->{$url};
6650                foreach my $path (keys %$fetch) {
6651                        my $x = $fetch->{$path};
6652                        Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
6653                        my $pfx = "svn-remote.$x->{old_repo_id}";
6654
6655                        my $old_fetch = quotemeta("$x->{old_path}:".
6656                                                  "$x->{ref_id}");
6657                        command_noisy(qw/config --unset/,
6658                                      "$pfx.fetch", '^'. $old_fetch . '$');
6659                        delete $r->{$x->{old_repo_id}}->
6660                               {fetch}->{$x->{old_path}};
6661                        if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
6662                                command_noisy(qw/config --unset/,
6663                                              "$pfx.url");
6664                                push @emptied, $x->{old_repo_id}
6665                        }
6666                }
6667        }
6668        if (@emptied) {
6669                my $file = $ENV{GIT_CONFIG} || "$ENV{GIT_DIR}/config";
6670                print STDERR <<EOF;
6671The following [svn-remote] sections in your config file ($file) are empty
6672and can be safely removed:
6673EOF
6674                print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
6675        }
6676}
6677
6678sub migration_check {
6679        migrate_from_v0();
6680        migrate_from_v1();
6681        migrate_from_v2();
6682        minimize_connections() if $_minimize;
6683}
6684
6685package Git::IndexInfo;
6686use strict;
6687use warnings;
6688use Git qw/command_input_pipe command_close_pipe/;
6689
6690sub new {
6691        my ($class) = @_;
6692        my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
6693        bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
6694}
6695
6696sub remove {
6697        my ($self, $path) = @_;
6698        if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
6699                return ++$self->{nr};
6700        }
6701        undef;
6702}
6703
6704sub update {
6705        my ($self, $mode, $hash, $path) = @_;
6706        if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
6707                return ++$self->{nr};
6708        }
6709        undef;
6710}
6711
6712sub DESTROY {
6713        my ($self) = @_;
6714        command_close_pipe($self->{gui}, $self->{ctx});
6715}
6716
6717package Git::SVN::GlobSpec;
6718use strict;
6719use warnings;
6720
6721sub new {
6722        my ($class, $glob, $pattern_ok) = @_;
6723        my $re = $glob;
6724        $re =~ s!/+$!!g; # no need for trailing slashes
6725        my (@left, @right, @patterns);
6726        my $state = "left";
6727        my $die_msg = "Only one set of wildcard directories " .
6728                                "(e.g. '*' or '*/*/*') is supported: '$glob'\n";
6729        for my $part (split(m|/|, $glob)) {
6730                if ($part =~ /\*/ && $part ne "*") {
6731                        die "Invalid pattern in '$glob': $part\n";
6732                } elsif ($pattern_ok && $part =~ /[{}]/ &&
6733                         $part !~ /^\{[^{}]+\}/) {
6734                        die "Invalid pattern in '$glob': $part\n";
6735                }
6736                if ($part eq "*") {
6737                        die $die_msg if $state eq "right";
6738                        $state = "pattern";
6739                        push(@patterns, "[^/]*");
6740                } elsif ($pattern_ok && $part =~ /^\{(.*)\}$/) {
6741                        die $die_msg if $state eq "right";
6742                        $state = "pattern";
6743                        my $p = quotemeta($1);
6744                        $p =~ s/\\,/|/g;
6745                        push(@patterns, "(?:$p)");
6746                } else {
6747                        if ($state eq "left") {
6748                                push(@left, $part);
6749                        } else {
6750                                push(@right, $part);
6751                                $state = "right";
6752                        }
6753                }
6754        }
6755        my $depth = @patterns;
6756        if ($depth == 0) {
6757                die "One '*' is needed in glob: '$glob'\n";
6758        }
6759        my $left = join('/', @left);
6760        my $right = join('/', @right);
6761        $re = join('/', @patterns);
6762        $re = join('\/',
6763                   grep(length, quotemeta($left), "($re)", quotemeta($right)));
6764        my $left_re = qr/^\/\Q$left\E(\/|$)/;
6765        bless { left => $left, right => $right, left_regex => $left_re,
6766                regex => qr/$re/, glob => $glob, depth => $depth }, $class;
6767}
6768
6769sub full_path {
6770        my ($self, $path) = @_;
6771        return (length $self->{left} ? "$self->{left}/" : '') .
6772               $path . (length $self->{right} ? "/$self->{right}" : '');
6773}
6774
6775__END__
6776
6777Data structures:
6778
6779
6780$remotes = { # returned by read_all_remotes()
6781        'svn' => {
6782                # svn-remote.svn.url=https://svn.musicpd.org
6783                url => 'https://svn.musicpd.org',
6784                # svn-remote.svn.fetch=mpd/trunk:trunk
6785                fetch => {
6786                        'mpd/trunk' => 'trunk',
6787                },
6788                # svn-remote.svn.tags=mpd/tags/*:tags/*
6789                tags => {
6790                        path => {
6791                                left => 'mpd/tags',
6792                                right => '',
6793                                regex => qr!mpd/tags/([^/]+)$!,
6794                                glob => 'tags/*',
6795                        },
6796                        ref => {
6797                                left => 'tags',
6798                                right => '',
6799                                regex => qr!tags/([^/]+)$!,
6800                                glob => 'tags/*',
6801                        },
6802                }
6803        }
6804};
6805
6806$log_entry hashref as returned by libsvn_log_entry()
6807{
6808        log => 'whitespace-formatted log entry
6809',                                              # trailing newline is preserved
6810        revision => '8',                        # integer
6811        date => '2004-02-24T17:01:44.108345Z',  # commit date
6812        author => 'committer name'
6813};
6814
6815
6816# this is generated by generate_diff();
6817@mods = array of diff-index line hashes, each element represents one line
6818        of diff-index output
6819
6820diff-index line ($m hash)
6821{
6822        mode_a => first column of diff-index output, no leading ':',
6823        mode_b => second column of diff-index output,
6824        sha1_b => sha1sum of the final blob,
6825        chg => change type [MCRADT],
6826        file_a => original file name of a file (iff chg is 'C' or 'R')
6827        file_b => new/current file name of a file (any chg)
6828}
6829;
6830
6831# retval of read_url_paths{,_all}();
6832$l_map = {
6833        # repository root url
6834        'https://svn.musicpd.org' => {
6835                # repository path               # GIT_SVN_ID
6836                'mpd/trunk'             =>      'trunk',
6837                'mpd/tags/0.11.5'       =>      'tags/0.11.5',
6838        },
6839}
6840
6841Notes:
6842        I don't trust the each() function on unless I created %hash myself
6843        because the internal iterator may not have started at base.