git-svn.perlon commit ident: don't write fallback username into git_default_name (060d4bb)
   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, $_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                          %fc_opts } ],
 259        'commit-diff' => [ \&cmd_commit_diff,
 260                           'Commit a diff between two trees',
 261                        { 'message|m=s' => \$_message,
 262                          'file|F=s' => \$_file,
 263                          'revision|r=s' => \$_revision,
 264                        %cmt_opts } ],
 265        'info' => [ \&cmd_info,
 266                    "Show info about the latest SVN revision
 267                     on the current branch",
 268                    { 'url' => \$_url, } ],
 269        'blame' => [ \&Git::SVN::Log::cmd_blame,
 270                    "Show what revision and author last modified each line of a file",
 271                    { 'git-format' => \$_git_format } ],
 272        'reset' => [ \&cmd_reset,
 273                     "Undo fetches back to the specified SVN revision",
 274                     { 'revision|r=s' => \$_revision,
 275                       'parent|p' => \$_fetch_parent } ],
 276        'gc' => [ \&cmd_gc,
 277                  "Compress unhandled.log files in .git/svn and remove " .
 278                  "index files in .git/svn",
 279                {} ],
 280);
 281
 282use Term::ReadLine;
 283package FakeTerm;
 284sub new {
 285        my ($class, $reason) = @_;
 286        return bless \$reason, shift;
 287}
 288sub readline {
 289        my $self = shift;
 290        die "Cannot use readline on FakeTerm: $$self";
 291}
 292package main;
 293
 294my $term = eval {
 295        $ENV{"GIT_SVN_NOTTY"}
 296                ? new Term::ReadLine 'git-svn', \*STDIN, \*STDOUT
 297                : new Term::ReadLine 'git-svn';
 298};
 299if ($@) {
 300        $term = new FakeTerm "$@: going non-interactive";
 301}
 302
 303my $cmd;
 304for (my $i = 0; $i < @ARGV; $i++) {
 305        if (defined $cmd{$ARGV[$i]}) {
 306                $cmd = $ARGV[$i];
 307                splice @ARGV, $i, 1;
 308                last;
 309        } elsif ($ARGV[$i] eq 'help') {
 310                $cmd = $ARGV[$i+1];
 311                usage(0);
 312        }
 313};
 314
 315# make sure we're always running at the top-level working directory
 316unless ($cmd && $cmd =~ /(?:clone|init|multi-init)$/) {
 317        unless (-d $ENV{GIT_DIR}) {
 318                if ($git_dir_user_set) {
 319                        die "GIT_DIR=$ENV{GIT_DIR} explicitly set, ",
 320                            "but it is not a directory\n";
 321                }
 322                my $git_dir = delete $ENV{GIT_DIR};
 323                my $cdup = undef;
 324                git_cmd_try {
 325                        $cdup = command_oneline(qw/rev-parse --show-cdup/);
 326                        $git_dir = '.' unless ($cdup);
 327                        chomp $cdup if ($cdup);
 328                        $cdup = "." unless ($cdup && length $cdup);
 329                } "Already at toplevel, but $git_dir not found\n";
 330                chdir $cdup or die "Unable to chdir up to '$cdup'\n";
 331                unless (-d $git_dir) {
 332                        die "$git_dir still not found after going to ",
 333                            "'$cdup'\n";
 334                }
 335                $ENV{GIT_DIR} = $git_dir;
 336        }
 337        $_repository = Git->repository(Repository => $ENV{GIT_DIR});
 338}
 339
 340my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
 341
 342read_git_config(\%opts);
 343if ($cmd && ($cmd eq 'log' || $cmd eq 'blame')) {
 344        Getopt::Long::Configure('pass_through');
 345}
 346my $rv = GetOptions(%opts, 'h|H' => \$_help, 'version|V' => \$_version,
 347                    'minimize-connections' => \$Git::SVN::Migration::_minimize,
 348                    'id|i=s' => \$Git::SVN::default_ref_id,
 349                    'svn-remote|remote|R=s' => sub {
 350                       $Git::SVN::no_reuse_existing = 1;
 351                       $Git::SVN::default_repo_id = $_[1] });
 352exit 1 if (!$rv && $cmd && $cmd ne 'log');
 353
 354usage(0) if $_help;
 355version() if $_version;
 356usage(1) unless defined $cmd;
 357load_authors() if $_authors;
 358if (defined $_authors_prog) {
 359        $_authors_prog = "'" . File::Spec->rel2abs($_authors_prog) . "'";
 360}
 361
 362unless ($cmd =~ /^(?:clone|init|multi-init|commit-diff)$/) {
 363        Git::SVN::Migration::migration_check();
 364}
 365Git::SVN::init_vars();
 366eval {
 367        Git::SVN::verify_remotes_sanity();
 368        $cmd{$cmd}->[0]->(@ARGV);
 369};
 370fatal $@ if $@;
 371post_fetch_checkout();
 372exit 0;
 373
 374####################### primary functions ######################
 375sub usage {
 376        my $exit = shift || 0;
 377        my $fd = $exit ? \*STDERR : \*STDOUT;
 378        print $fd <<"";
 379git-svn - bidirectional operations between a single Subversion tree and git
 380Usage: git svn <command> [options] [arguments]\n
 381
 382        print $fd "Available commands:\n" unless $cmd;
 383
 384        foreach (sort keys %cmd) {
 385                next if $cmd && $cmd ne $_;
 386                next if /^multi-/; # don't show deprecated commands
 387                print $fd '  ',pack('A17',$_),$cmd{$_}->[1],"\n";
 388                foreach (sort keys %{$cmd{$_}->[2]}) {
 389                        # mixed-case options are for .git/config only
 390                        next if /[A-Z]/ && /^[a-z]+$/i;
 391                        # prints out arguments as they should be passed:
 392                        my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
 393                        print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
 394                                                        "--$_" : "-$_" }
 395                                                split /\|/,$_)," $x\n";
 396                }
 397        }
 398        print $fd <<"";
 399\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
 400arbitrary identifier if you're tracking multiple SVN branches/repositories in
 401one git repository and want to keep them separate.  See git-svn(1) for more
 402information.
 403
 404        exit $exit;
 405}
 406
 407sub version {
 408        ::_req_svn();
 409        print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
 410        exit 0;
 411}
 412
 413sub ask {
 414        my ($prompt, %arg) = @_;
 415        my $valid_re = $arg{valid_re};
 416        my $default = $arg{default};
 417        my $resp;
 418        my $i = 0;
 419
 420        if ( !( defined($term->IN)
 421            && defined( fileno($term->IN) )
 422            && defined( $term->OUT )
 423            && defined( fileno($term->OUT) ) ) ){
 424                return defined($default) ? $default : undef;
 425        }
 426
 427        while ($i++ < 10) {
 428                $resp = $term->readline($prompt);
 429                if (!defined $resp) { # EOF
 430                        print "\n";
 431                        return defined $default ? $default : undef;
 432                }
 433                if ($resp eq '' and defined $default) {
 434                        return $default;
 435                }
 436                if (!defined $valid_re or $resp =~ /$valid_re/) {
 437                        return $resp;
 438                }
 439        }
 440        return undef;
 441}
 442
 443sub do_git_init_db {
 444        unless (-d $ENV{GIT_DIR}) {
 445                my @init_db = ('init');
 446                push @init_db, "--template=$_template" if defined $_template;
 447                if (defined $_shared) {
 448                        if ($_shared =~ /[a-z]/) {
 449                                push @init_db, "--shared=$_shared";
 450                        } else {
 451                                push @init_db, "--shared";
 452                        }
 453                }
 454                command_noisy(@init_db);
 455                $_repository = Git->repository(Repository => ".git");
 456        }
 457        my $set;
 458        my $pfx = "svn-remote.$Git::SVN::default_repo_id";
 459        foreach my $i (keys %icv) {
 460                die "'$set' and '$i' cannot both be set\n" if $set;
 461                next unless defined $icv{$i};
 462                command_noisy('config', "$pfx.$i", $icv{$i});
 463                $set = $i;
 464        }
 465        my $ignore_paths_regex = \$SVN::Git::Fetcher::_ignore_regex;
 466        command_noisy('config', "$pfx.ignore-paths", $$ignore_paths_regex)
 467                if defined $$ignore_paths_regex;
 468        my $ignore_refs_regex = \$Git::SVN::Ra::_ignore_refs_regex;
 469        command_noisy('config', "$pfx.ignore-refs", $$ignore_refs_regex)
 470                if defined $$ignore_refs_regex;
 471
 472        if (defined $SVN::Git::Fetcher::_preserve_empty_dirs) {
 473                my $fname = \$SVN::Git::Fetcher::_placeholder_filename;
 474                command_noisy('config', "$pfx.preserve-empty-dirs", 'true');
 475                command_noisy('config', "$pfx.placeholder-filename", $$fname);
 476        }
 477}
 478
 479sub init_subdir {
 480        my $repo_path = shift or return;
 481        mkpath([$repo_path]) unless -d $repo_path;
 482        chdir $repo_path or die "Couldn't chdir to $repo_path: $!\n";
 483        $ENV{GIT_DIR} = '.git';
 484        $_repository = Git->repository(Repository => $ENV{GIT_DIR});
 485}
 486
 487sub cmd_clone {
 488        my ($url, $path) = @_;
 489        if (!defined $path &&
 490            (defined $_trunk || @_branches || @_tags ||
 491             defined $_stdlayout) &&
 492            $url !~ m#^[a-z\+]+://#) {
 493                $path = $url;
 494        }
 495        $path = basename($url) if !defined $path || !length $path;
 496        my $authors_absolute = $_authors ? File::Spec->rel2abs($_authors) : "";
 497        cmd_init($url, $path);
 498        command_oneline('config', 'svn.authorsfile', $authors_absolute)
 499            if $_authors;
 500        Git::SVN::fetch_all($Git::SVN::default_repo_id);
 501}
 502
 503sub cmd_init {
 504        if (defined $_stdlayout) {
 505                $_trunk = 'trunk' if (!defined $_trunk);
 506                @_tags = 'tags' if (! @_tags);
 507                @_branches = 'branches' if (! @_branches);
 508        }
 509        if (defined $_trunk || @_branches || @_tags) {
 510                return cmd_multi_init(@_);
 511        }
 512        my $url = shift or die "SVN repository location required ",
 513                               "as a command-line argument\n";
 514        $url = canonicalize_url($url);
 515        init_subdir(@_);
 516        do_git_init_db();
 517
 518        if ($Git::SVN::_minimize_url eq 'unset') {
 519                $Git::SVN::_minimize_url = 0;
 520        }
 521
 522        Git::SVN->init($url);
 523}
 524
 525sub cmd_fetch {
 526        if (grep /^\d+=./, @_) {
 527                die "'<rev>=<commit>' fetch arguments are ",
 528                    "no longer supported.\n";
 529        }
 530        my ($remote) = @_;
 531        if (@_ > 1) {
 532                die "Usage: $0 fetch [--all] [--parent] [svn-remote]\n";
 533        }
 534        $Git::SVN::no_reuse_existing = undef;
 535        if ($_fetch_parent) {
 536                my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 537                unless ($gs) {
 538                        die "Unable to determine upstream SVN information from ",
 539                            "working tree history\n";
 540                }
 541                # just fetch, don't checkout.
 542                $_no_checkout = 'true';
 543                $_fetch_all ? $gs->fetch_all : $gs->fetch;
 544        } elsif ($_fetch_all) {
 545                cmd_multi_fetch();
 546        } else {
 547                $remote ||= $Git::SVN::default_repo_id;
 548                Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
 549        }
 550}
 551
 552sub cmd_set_tree {
 553        my (@commits) = @_;
 554        if ($_stdin || !@commits) {
 555                print "Reading from stdin...\n";
 556                @commits = ();
 557                while (<STDIN>) {
 558                        if (/\b($sha1_short)\b/o) {
 559                                unshift @commits, $1;
 560                        }
 561                }
 562        }
 563        my @revs;
 564        foreach my $c (@commits) {
 565                my @tmp = command('rev-parse',$c);
 566                if (scalar @tmp == 1) {
 567                        push @revs, $tmp[0];
 568                } elsif (scalar @tmp > 1) {
 569                        push @revs, reverse(command('rev-list',@tmp));
 570                } else {
 571                        fatal "Failed to rev-parse $c";
 572                }
 573        }
 574        my $gs = Git::SVN->new;
 575        my ($r_last, $cmt_last) = $gs->last_rev_commit;
 576        $gs->fetch;
 577        if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
 578                fatal "There are new revisions that were fetched ",
 579                      "and need to be merged (or acknowledged) ",
 580                      "before committing.\nlast rev: $r_last\n",
 581                      " current: $gs->{last_rev}";
 582        }
 583        $gs->set_tree($_) foreach @revs;
 584        print "Done committing ",scalar @revs," revisions to SVN\n";
 585        unlink $gs->{index};
 586}
 587
 588sub split_merge_info_range {
 589        my ($range) = @_;
 590        if ($range =~ /(\d+)-(\d+)/) {
 591                return (int($1), int($2));
 592        } else {
 593                return (int($range), int($range));
 594        }
 595}
 596
 597sub combine_ranges {
 598        my ($in) = @_;
 599
 600        my @fnums = ();
 601        my @arr = split(/,/, $in);
 602        for my $element (@arr) {
 603                my ($start, $end) = split_merge_info_range($element);
 604                push @fnums, $start;
 605        }
 606
 607        my @sorted = @arr [ sort {
 608                $fnums[$a] <=> $fnums[$b]
 609        } 0..$#arr ];
 610
 611        my @return = ();
 612        my $last = -1;
 613        my $first = -1;
 614        for my $element (@sorted) {
 615                my ($start, $end) = split_merge_info_range($element);
 616
 617                if ($last == -1) {
 618                        $first = $start;
 619                        $last = $end;
 620                        next;
 621                }
 622                if ($start <= $last+1) {
 623                        if ($end > $last) {
 624                                $last = $end;
 625                        }
 626                        next;
 627                }
 628                if ($first == $last) {
 629                        push @return, "$first";
 630                } else {
 631                        push @return, "$first-$last";
 632                }
 633                $first = $start;
 634                $last = $end;
 635        }
 636
 637        if ($first != -1) {
 638                if ($first == $last) {
 639                        push @return, "$first";
 640                } else {
 641                        push @return, "$first-$last";
 642                }
 643        }
 644
 645        return join(',', @return);
 646}
 647
 648sub merge_revs_into_hash {
 649        my ($hash, $minfo) = @_;
 650        my @lines = split(' ', $minfo);
 651
 652        for my $line (@lines) {
 653                my ($branchpath, $revs) = split(/:/, $line);
 654
 655                if (exists($hash->{$branchpath})) {
 656                        # Merge the two revision sets
 657                        my $combined = "$hash->{$branchpath},$revs";
 658                        $hash->{$branchpath} = combine_ranges($combined);
 659                } else {
 660                        # Just do range combining for consolidation
 661                        $hash->{$branchpath} = combine_ranges($revs);
 662                }
 663        }
 664}
 665
 666sub merge_merge_info {
 667        my ($mergeinfo_one, $mergeinfo_two) = @_;
 668        my %result_hash = ();
 669
 670        merge_revs_into_hash(\%result_hash, $mergeinfo_one);
 671        merge_revs_into_hash(\%result_hash, $mergeinfo_two);
 672
 673        my $result = '';
 674        # Sort below is for consistency's sake
 675        for my $branchname (sort keys(%result_hash)) {
 676                my $revlist = $result_hash{$branchname};
 677                $result .= "$branchname:$revlist\n"
 678        }
 679        return $result;
 680}
 681
 682sub populate_merge_info {
 683        my ($d, $gs, $uuid, $linear_refs, $rewritten_parent) = @_;
 684
 685        my %parentshash;
 686        read_commit_parents(\%parentshash, $d);
 687        my @parents = @{$parentshash{$d}};
 688        if ($#parents > 0) {
 689                # Merge commit
 690                my $all_parents_ok = 1;
 691                my $aggregate_mergeinfo = '';
 692                my $rooturl = $gs->repos_root;
 693
 694                if (defined($rewritten_parent)) {
 695                        # Replace first parent with newly-rewritten version
 696                        shift @parents;
 697                        unshift @parents, $rewritten_parent;
 698                }
 699
 700                foreach my $parent (@parents) {
 701                        my ($branchurl, $svnrev, $paruuid) =
 702                                cmt_metadata($parent);
 703
 704                        unless (defined($svnrev)) {
 705                                # Should have been caught be preflight check
 706                                fatal "merge commit $d has ancestor $parent, but that change "
 707                     ."does not have git-svn metadata!";
 708                        }
 709                        unless ($branchurl =~ /^\Q$rooturl\E(.*)/) {
 710                                fatal "commit $parent git-svn metadata changed mid-run!";
 711                        }
 712                        my $branchpath = $1;
 713
 714                        my $ra = Git::SVN::Ra->new($branchurl);
 715                        my (undef, undef, $props) =
 716                                $ra->get_dir(canonicalize_path("."), $svnrev);
 717                        my $par_mergeinfo = $props->{'svn:mergeinfo'};
 718                        unless (defined $par_mergeinfo) {
 719                                $par_mergeinfo = '';
 720                        }
 721                        # Merge previous mergeinfo values
 722                        $aggregate_mergeinfo =
 723                                merge_merge_info($aggregate_mergeinfo,
 724                                                                 $par_mergeinfo, 0);
 725
 726                        next if $parent eq $parents[0]; # Skip first parent
 727                        # Add new changes being placed in tree by merge
 728                        my @cmd = (qw/rev-list --reverse/,
 729                                           $parent, qw/--not/);
 730                        foreach my $par (@parents) {
 731                                unless ($par eq $parent) {
 732                                        push @cmd, $par;
 733                                }
 734                        }
 735                        my @revsin = ();
 736                        my ($revlist, $ctx) = command_output_pipe(@cmd);
 737                        while (<$revlist>) {
 738                                my $irev = $_;
 739                                chomp $irev;
 740                                my (undef, $csvnrev, undef) =
 741                                        cmt_metadata($irev);
 742                                unless (defined $csvnrev) {
 743                                        # A child is missing SVN annotations...
 744                                        # this might be OK, or might not be.
 745                                        warn "W:child $irev is merged into revision "
 746                                                 ."$d but does not have git-svn metadata. "
 747                                                 ."This means git-svn cannot determine the "
 748                                                 ."svn revision numbers to place into the "
 749                                                 ."svn:mergeinfo property. You must ensure "
 750                                                 ."a branch is entirely committed to "
 751                                                 ."SVN before merging it in order for "
 752                                                 ."svn:mergeinfo population to function "
 753                                                 ."properly";
 754                                }
 755                                push @revsin, $csvnrev;
 756                        }
 757                        command_close_pipe($revlist, $ctx);
 758
 759                        last unless $all_parents_ok;
 760
 761                        # We now have a list of all SVN revnos which are
 762                        # merged by this particular parent. Integrate them.
 763                        next if $#revsin == -1;
 764                        my $newmergeinfo = "$branchpath:" . join(',', @revsin);
 765                        $aggregate_mergeinfo =
 766                                merge_merge_info($aggregate_mergeinfo,
 767                                                                 $newmergeinfo, 1);
 768                }
 769                if ($all_parents_ok and $aggregate_mergeinfo) {
 770                        return $aggregate_mergeinfo;
 771                }
 772        }
 773
 774        return undef;
 775}
 776
 777sub cmd_dcommit {
 778        my $head = shift;
 779        command_noisy(qw/update-index --refresh/);
 780        git_cmd_try { command_oneline(qw/diff-index --quiet HEAD/) }
 781                'Cannot dcommit with a dirty index.  Commit your changes first, '
 782                . "or stash them with `git stash'.\n";
 783        $head ||= 'HEAD';
 784
 785        my $old_head;
 786        if ($head ne 'HEAD') {
 787                $old_head = eval {
 788                        command_oneline([qw/symbolic-ref -q HEAD/])
 789                };
 790                if ($old_head) {
 791                        $old_head =~ s{^refs/heads/}{};
 792                } else {
 793                        $old_head = eval { command_oneline(qw/rev-parse HEAD/) };
 794                }
 795                command(['checkout', $head], STDERR => 0);
 796        }
 797
 798        my @refs;
 799        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD', \@refs);
 800        unless ($gs) {
 801                die "Unable to determine upstream SVN information from ",
 802                    "$head history.\nPerhaps the repository is empty.";
 803        }
 804
 805        if (defined $_commit_url) {
 806                $url = $_commit_url;
 807        } else {
 808                $url = eval { command_oneline('config', '--get',
 809                              "svn-remote.$gs->{repo_id}.commiturl") };
 810                if (!$url) {
 811                        $url = $gs->full_pushurl
 812                }
 813        }
 814
 815        my $last_rev = $_revision if defined $_revision;
 816        if ($url) {
 817                print "Committing to $url ...\n";
 818        }
 819        my ($linear_refs, $parents) = linearize_history($gs, \@refs);
 820        if ($_no_rebase && scalar(@$linear_refs) > 1) {
 821                warn "Attempting to commit more than one change while ",
 822                     "--no-rebase is enabled.\n",
 823                     "If these changes depend on each other, re-running ",
 824                     "without --no-rebase may be required."
 825        }
 826
 827        if (defined $_interactive){
 828                my $ask_default = "y";
 829                foreach my $d (@$linear_refs){
 830                        my ($fh, $ctx) = command_output_pipe(qw(show --summary), "$d");
 831                        while (<$fh>){
 832                                print $_;
 833                        }
 834                        command_close_pipe($fh, $ctx);
 835                        $_ = ask("Commit this patch to SVN? ([y]es (default)|[n]o|[q]uit|[a]ll): ",
 836                                 valid_re => qr/^(?:yes|y|no|n|quit|q|all|a)/i,
 837                                 default => $ask_default);
 838                        die "Commit this patch reply required" unless defined $_;
 839                        if (/^[nq]/i) {
 840                                exit(0);
 841                        } elsif (/^a/i) {
 842                                last;
 843                        }
 844                }
 845        }
 846
 847        my $expect_url = $url;
 848
 849        my $push_merge_info = eval {
 850                command_oneline(qw/config --get svn.pushmergeinfo/)
 851                };
 852        if (not defined($push_merge_info)
 853                        or $push_merge_info eq "false"
 854                        or $push_merge_info eq "no"
 855                        or $push_merge_info eq "never") {
 856                $push_merge_info = 0;
 857        }
 858
 859        unless (defined($_merge_info) || ! $push_merge_info) {
 860                # Preflight check of changes to ensure no issues with mergeinfo
 861                # This includes check for uncommitted-to-SVN parents
 862                # (other than the first parent, which we will handle),
 863                # information from different SVN repos, and paths
 864                # which are not underneath this repository root.
 865                my $rooturl = $gs->repos_root;
 866                foreach my $d (@$linear_refs) {
 867                        my %parentshash;
 868                        read_commit_parents(\%parentshash, $d);
 869                        my @realparents = @{$parentshash{$d}};
 870                        if ($#realparents > 0) {
 871                                # Merge commit
 872                                shift @realparents; # Remove/ignore first parent
 873                                foreach my $parent (@realparents) {
 874                                        my ($branchurl, $svnrev, $paruuid) = cmt_metadata($parent);
 875                                        unless (defined $paruuid) {
 876                                                # A parent is missing SVN annotations...
 877                                                # abort the whole operation.
 878                                                fatal "$parent is merged into revision $d, "
 879                                                         ."but does not have git-svn metadata. "
 880                                                         ."Either dcommit the branch or use a "
 881                                                         ."local cherry-pick, FF merge, or rebase "
 882                                                         ."instead of an explicit merge commit.";
 883                                        }
 884
 885                                        unless ($paruuid eq $uuid) {
 886                                                # Parent has SVN metadata from different repository
 887                                                fatal "merge parent $parent for change $d has "
 888                                                         ."git-svn uuid $paruuid, while current change "
 889                                                         ."has uuid $uuid!";
 890                                        }
 891
 892                                        unless ($branchurl =~ /^\Q$rooturl\E(.*)/) {
 893                                                # This branch is very strange indeed.
 894                                                fatal "merge parent $parent for $d is on branch "
 895                                                         ."$branchurl, which is not under the "
 896                                                         ."git-svn root $rooturl!";
 897                                        }
 898                                }
 899                        }
 900                }
 901        }
 902
 903        my $rewritten_parent;
 904        Git::SVN::remove_username($expect_url);
 905        if (defined($_merge_info)) {
 906                $_merge_info =~ tr{ }{\n};
 907        }
 908        while (1) {
 909                my $d = shift @$linear_refs or last;
 910                unless (defined $last_rev) {
 911                        (undef, $last_rev, undef) = cmt_metadata("$d~1");
 912                        unless (defined $last_rev) {
 913                                fatal "Unable to extract revision information ",
 914                                      "from commit $d~1";
 915                        }
 916                }
 917                if ($_dry_run) {
 918                        print "diff-tree $d~1 $d\n";
 919                } else {
 920                        my $cmt_rev;
 921
 922                        unless (defined($_merge_info) || ! $push_merge_info) {
 923                                $_merge_info = populate_merge_info($d, $gs,
 924                                                             $uuid,
 925                                                             $linear_refs,
 926                                                             $rewritten_parent);
 927                        }
 928
 929                        my %ed_opts = ( r => $last_rev,
 930                                        log => get_commit_entry($d)->{log},
 931                                        ra => Git::SVN::Ra->new($url),
 932                                        config => SVN::Core::config_get_config(
 933                                                $Git::SVN::Ra::config_dir
 934                                        ),
 935                                        tree_a => "$d~1",
 936                                        tree_b => $d,
 937                                        editor_cb => sub {
 938                                               print "Committed r$_[0]\n";
 939                                               $cmt_rev = $_[0];
 940                                        },
 941                                        mergeinfo => $_merge_info,
 942                                        svn_path => '');
 943                        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
 944                                print "No changes\n$d~1 == $d\n";
 945                        } elsif ($parents->{$d} && @{$parents->{$d}}) {
 946                                $gs->{inject_parents_dcommit}->{$cmt_rev} =
 947                                                               $parents->{$d};
 948                        }
 949                        $_fetch_all ? $gs->fetch_all : $gs->fetch;
 950                        $last_rev = $cmt_rev;
 951                        next if $_no_rebase;
 952
 953                        # we always want to rebase against the current HEAD,
 954                        # not any head that was passed to us
 955                        my @diff = command('diff-tree', $d,
 956                                           $gs->refname, '--');
 957                        my @finish;
 958                        if (@diff) {
 959                                @finish = rebase_cmd();
 960                                print STDERR "W: $d and ", $gs->refname,
 961                                             " differ, using @finish:\n",
 962                                             join("\n", @diff), "\n";
 963                        } else {
 964                                print "No changes between current HEAD and ",
 965                                      $gs->refname,
 966                                      "\nResetting to the latest ",
 967                                      $gs->refname, "\n";
 968                                @finish = qw/reset --mixed/;
 969                        }
 970                        command_noisy(@finish, $gs->refname);
 971
 972                        $rewritten_parent = command_oneline(qw/rev-parse HEAD/);
 973
 974                        if (@diff) {
 975                                @refs = ();
 976                                my ($url_, $rev_, $uuid_, $gs_) =
 977                                              working_head_info('HEAD', \@refs);
 978                                my ($linear_refs_, $parents_) =
 979                                              linearize_history($gs_, \@refs);
 980                                if (scalar(@$linear_refs) !=
 981                                    scalar(@$linear_refs_)) {
 982                                        fatal "# of revisions changed ",
 983                                          "\nbefore:\n",
 984                                          join("\n", @$linear_refs),
 985                                          "\n\nafter:\n",
 986                                          join("\n", @$linear_refs_), "\n",
 987                                          'If you are attempting to commit ',
 988                                          "merges, try running:\n\t",
 989                                          'git rebase --interactive',
 990                                          '--preserve-merges ',
 991                                          $gs->refname,
 992                                          "\nBefore dcommitting";
 993                                }
 994                                if ($url_ ne $expect_url) {
 995                                        if ($url_ eq $gs->metadata_url) {
 996                                                print
 997                                                  "Accepting rewritten URL:",
 998                                                  " $url_\n";
 999                                        } else {
1000                                                fatal
1001                                                  "URL mismatch after rebase:",
1002                                                  " $url_ != $expect_url";
1003                                        }
1004                                }
1005                                if ($uuid_ ne $uuid) {
1006                                        fatal "uuid mismatch after rebase: ",
1007                                              "$uuid_ != $uuid";
1008                                }
1009                                # remap parents
1010                                my (%p, @l, $i);
1011                                for ($i = 0; $i < scalar @$linear_refs; $i++) {
1012                                        my $new = $linear_refs_->[$i] or next;
1013                                        $p{$new} =
1014                                                $parents->{$linear_refs->[$i]};
1015                                        push @l, $new;
1016                                }
1017                                $parents = \%p;
1018                                $linear_refs = \@l;
1019                        }
1020                }
1021        }
1022
1023        if ($old_head) {
1024                my $new_head = command_oneline(qw/rev-parse HEAD/);
1025                my $new_is_symbolic = eval {
1026                        command_oneline(qw/symbolic-ref -q HEAD/);
1027                };
1028                if ($new_is_symbolic) {
1029                        print "dcommitted the branch ", $head, "\n";
1030                } else {
1031                        print "dcommitted on a detached HEAD because you gave ",
1032                              "a revision argument.\n",
1033                              "The rewritten commit is: ", $new_head, "\n";
1034                }
1035                command(['checkout', $old_head], STDERR => 0);
1036        }
1037
1038        unlink $gs->{index};
1039}
1040
1041sub cmd_branch {
1042        my ($branch_name, $head) = @_;
1043
1044        unless (defined $branch_name && length $branch_name) {
1045                die(($_tag ? "tag" : "branch") . " name required\n");
1046        }
1047        $head ||= 'HEAD';
1048
1049        my (undef, $rev, undef, $gs) = working_head_info($head);
1050        my $src = $gs->full_pushurl;
1051
1052        my $remote = Git::SVN::read_all_remotes()->{$gs->{repo_id}};
1053        my $allglobs = $remote->{ $_tag ? 'tags' : 'branches' };
1054        my $glob;
1055        if ($#{$allglobs} == 0) {
1056                $glob = $allglobs->[0];
1057        } else {
1058                unless(defined $_branch_dest) {
1059                        die "Multiple ",
1060                            $_tag ? "tag" : "branch",
1061                            " paths defined for Subversion repository.\n",
1062                            "You must specify where you want to create the ",
1063                            $_tag ? "tag" : "branch",
1064                            " with the --destination argument.\n";
1065                }
1066                foreach my $g (@{$allglobs}) {
1067                        # SVN::Git::Editor could probably be moved to Git.pm..
1068                        my $re = SVN::Git::Editor::glob2pat($g->{path}->{left});
1069                        if ($_branch_dest =~ /$re/) {
1070                                $glob = $g;
1071                                last;
1072                        }
1073                }
1074                unless (defined $glob) {
1075                        my $dest_re = qr/\b\Q$_branch_dest\E\b/;
1076                        foreach my $g (@{$allglobs}) {
1077                                $g->{path}->{left} =~ /$dest_re/ or next;
1078                                if (defined $glob) {
1079                                        die "Ambiguous destination: ",
1080                                            $_branch_dest, "\nmatches both '",
1081                                            $glob->{path}->{left}, "' and '",
1082                                            $g->{path}->{left}, "'\n";
1083                                }
1084                                $glob = $g;
1085                        }
1086                        unless (defined $glob) {
1087                                die "Unknown ",
1088                                    $_tag ? "tag" : "branch",
1089                                    " destination $_branch_dest\n";
1090                        }
1091                }
1092        }
1093        my ($lft, $rgt) = @{ $glob->{path} }{qw/left right/};
1094        my $url;
1095        if (defined $_commit_url) {
1096                $url = $_commit_url;
1097        } else {
1098                $url = eval { command_oneline('config', '--get',
1099                        "svn-remote.$gs->{repo_id}.commiturl") };
1100                if (!$url) {
1101                        $url = $remote->{pushurl} || $remote->{url};
1102                }
1103        }
1104        my $dst = join '/', $url, $lft, $branch_name, ($rgt || ());
1105
1106        if ($dst =~ /^https:/ && $src =~ /^http:/) {
1107                $src=~s/^http:/https:/;
1108        }
1109
1110        ::_req_svn();
1111
1112        my $ctx = SVN::Client->new(
1113                auth    => Git::SVN::Ra::_auth_providers(),
1114                log_msg => sub {
1115                        ${ $_[0] } = defined $_message
1116                                ? $_message
1117                                : 'Create ' . ($_tag ? 'tag ' : 'branch ' )
1118                                . $branch_name;
1119                },
1120        );
1121
1122        eval {
1123                $ctx->ls($dst, 'HEAD', 0);
1124        } and die "branch ${branch_name} already exists\n";
1125
1126        print "Copying ${src} at r${rev} to ${dst}...\n";
1127        $ctx->copy($src, $rev, $dst)
1128                unless $_dry_run;
1129
1130        $gs->fetch_all;
1131}
1132
1133sub cmd_find_rev {
1134        my $revision_or_hash = shift or die "SVN or git revision required ",
1135                                            "as a command-line argument\n";
1136        my $result;
1137        if ($revision_or_hash =~ /^r\d+$/) {
1138                my $head = shift;
1139                $head ||= 'HEAD';
1140                my @refs;
1141                my (undef, undef, $uuid, $gs) = working_head_info($head, \@refs);
1142                unless ($gs) {
1143                        die "Unable to determine upstream SVN information from ",
1144                            "$head history\n";
1145                }
1146                my $desired_revision = substr($revision_or_hash, 1);
1147                $result = $gs->rev_map_get($desired_revision, $uuid);
1148        } else {
1149                my (undef, $rev, undef) = cmt_metadata($revision_or_hash);
1150                $result = $rev;
1151        }
1152        print "$result\n" if $result;
1153}
1154
1155sub auto_create_empty_directories {
1156        my ($gs) = @_;
1157        my $var = eval { command_oneline('config', '--get', '--bool',
1158                                         "svn-remote.$gs->{repo_id}.automkdirs") };
1159        # By default, create empty directories by consulting the unhandled log,
1160        # but allow setting it to 'false' to skip it.
1161        return !($var && $var eq 'false');
1162}
1163
1164sub cmd_rebase {
1165        command_noisy(qw/update-index --refresh/);
1166        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1167        unless ($gs) {
1168                die "Unable to determine upstream SVN information from ",
1169                    "working tree history\n";
1170        }
1171        if ($_dry_run) {
1172                print "Remote Branch: " . $gs->refname . "\n";
1173                print "SVN URL: " . $url . "\n";
1174                return;
1175        }
1176        if (command(qw/diff-index HEAD --/)) {
1177                print STDERR "Cannot rebase with uncommited changes:\n";
1178                command_noisy('status');
1179                exit 1;
1180        }
1181        unless ($_local) {
1182                # rebase will checkout for us, so no need to do it explicitly
1183                $_no_checkout = 'true';
1184                $_fetch_all ? $gs->fetch_all : $gs->fetch;
1185        }
1186        command_noisy(rebase_cmd(), $gs->refname);
1187        if (auto_create_empty_directories($gs)) {
1188                $gs->mkemptydirs;
1189        }
1190}
1191
1192sub cmd_show_ignore {
1193        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1194        $gs ||= Git::SVN->new;
1195        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
1196        $gs->prop_walk($gs->{path}, $r, sub {
1197                my ($gs, $path, $props) = @_;
1198                print STDOUT "\n# $path\n";
1199                my $s = $props->{'svn:ignore'} or return;
1200                $s =~ s/[\r\n]+/\n/g;
1201                $s =~ s/^\n+//;
1202                chomp $s;
1203                $s =~ s#^#$path#gm;
1204                print STDOUT "$s\n";
1205        });
1206}
1207
1208sub cmd_show_externals {
1209        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1210        $gs ||= Git::SVN->new;
1211        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
1212        $gs->prop_walk($gs->{path}, $r, sub {
1213                my ($gs, $path, $props) = @_;
1214                print STDOUT "\n# $path\n";
1215                my $s = $props->{'svn:externals'} or return;
1216                $s =~ s/[\r\n]+/\n/g;
1217                chomp $s;
1218                $s =~ s#^#$path#gm;
1219                print STDOUT "$s\n";
1220        });
1221}
1222
1223sub cmd_create_ignore {
1224        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1225        $gs ||= Git::SVN->new;
1226        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
1227        $gs->prop_walk($gs->{path}, $r, sub {
1228                my ($gs, $path, $props) = @_;
1229                # $path is of the form /path/to/dir/
1230                $path = '.' . $path;
1231                # SVN can have attributes on empty directories,
1232                # which git won't track
1233                mkpath([$path]) unless -d $path;
1234                my $ignore = $path . '.gitignore';
1235                my $s = $props->{'svn:ignore'} or return;
1236                open(GITIGNORE, '>', $ignore)
1237                  or fatal("Failed to open `$ignore' for writing: $!");
1238                $s =~ s/[\r\n]+/\n/g;
1239                $s =~ s/^\n+//;
1240                chomp $s;
1241                # Prefix all patterns so that the ignore doesn't apply
1242                # to sub-directories.
1243                $s =~ s#^#/#gm;
1244                print GITIGNORE "$s\n";
1245                close(GITIGNORE)
1246                  or fatal("Failed to close `$ignore': $!");
1247                command_noisy('add', '-f', $ignore);
1248        });
1249}
1250
1251sub cmd_mkdirs {
1252        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1253        $gs ||= Git::SVN->new;
1254        $gs->mkemptydirs($_revision);
1255}
1256
1257sub canonicalize_path {
1258        my ($path) = @_;
1259        my $dot_slash_added = 0;
1260        if (substr($path, 0, 1) ne "/") {
1261                $path = "./" . $path;
1262                $dot_slash_added = 1;
1263        }
1264        # File::Spec->canonpath doesn't collapse x/../y into y (for a
1265        # good reason), so let's do this manually.
1266        $path =~ s#/+#/#g;
1267        $path =~ s#/\.(?:/|$)#/#g;
1268        $path =~ s#/[^/]+/\.\.##g;
1269        $path =~ s#/$##g;
1270        $path =~ s#^\./## if $dot_slash_added;
1271        $path =~ s#^/##;
1272        $path =~ s#^\.$##;
1273        return $path;
1274}
1275
1276sub canonicalize_url {
1277        my ($url) = @_;
1278        $url =~ s#^([^:]+://[^/]*/)(.*)$#$1 . canonicalize_path($2)#e;
1279        return $url;
1280}
1281
1282# get_svnprops(PATH)
1283# ------------------
1284# Helper for cmd_propget and cmd_proplist below.
1285sub get_svnprops {
1286        my $path = shift;
1287        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1288        $gs ||= Git::SVN->new;
1289
1290        # prefix THE PATH by the sub-directory from which the user
1291        # invoked us.
1292        $path = $cmd_dir_prefix . $path;
1293        fatal("No such file or directory: $path") unless -e $path;
1294        my $is_dir = -d $path ? 1 : 0;
1295        $path = $gs->{path} . '/' . $path;
1296
1297        # canonicalize the path (otherwise libsvn will abort or fail to
1298        # find the file)
1299        $path = canonicalize_path($path);
1300
1301        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
1302        my $props;
1303        if ($is_dir) {
1304                (undef, undef, $props) = $gs->ra->get_dir($path, $r);
1305        }
1306        else {
1307                (undef, $props) = $gs->ra->get_file($path, $r, undef);
1308        }
1309        return $props;
1310}
1311
1312# cmd_propget (PROP, PATH)
1313# ------------------------
1314# Print the SVN property PROP for PATH.
1315sub cmd_propget {
1316        my ($prop, $path) = @_;
1317        $path = '.' if not defined $path;
1318        usage(1) if not defined $prop;
1319        my $props = get_svnprops($path);
1320        if (not defined $props->{$prop}) {
1321                fatal("`$path' does not have a `$prop' SVN property.");
1322        }
1323        print $props->{$prop} . "\n";
1324}
1325
1326# cmd_proplist (PATH)
1327# -------------------
1328# Print the list of SVN properties for PATH.
1329sub cmd_proplist {
1330        my $path = shift;
1331        $path = '.' if not defined $path;
1332        my $props = get_svnprops($path);
1333        print "Properties on '$path':\n";
1334        foreach (sort keys %{$props}) {
1335                print "  $_\n";
1336        }
1337}
1338
1339sub cmd_multi_init {
1340        my $url = shift;
1341        unless (defined $_trunk || @_branches || @_tags) {
1342                usage(1);
1343        }
1344
1345        $_prefix = '' unless defined $_prefix;
1346        if (defined $url) {
1347                $url = canonicalize_url($url);
1348                init_subdir(@_);
1349        }
1350        do_git_init_db();
1351        if (defined $_trunk) {
1352                $_trunk =~ s#^/+##;
1353                my $trunk_ref = 'refs/remotes/' . $_prefix . 'trunk';
1354                # try both old-style and new-style lookups:
1355                my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
1356                unless ($gs_trunk) {
1357                        my ($trunk_url, $trunk_path) =
1358                                              complete_svn_url($url, $_trunk);
1359                        $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
1360                                                   undef, $trunk_ref);
1361                }
1362        }
1363        return unless @_branches || @_tags;
1364        my $ra = $url ? Git::SVN::Ra->new($url) : undef;
1365        foreach my $path (@_branches) {
1366                complete_url_ls_init($ra, $path, '--branches/-b', $_prefix);
1367        }
1368        foreach my $path (@_tags) {
1369                complete_url_ls_init($ra, $path, '--tags/-t', $_prefix.'tags/');
1370        }
1371}
1372
1373sub cmd_multi_fetch {
1374        $Git::SVN::no_reuse_existing = undef;
1375        my $remotes = Git::SVN::read_all_remotes();
1376        foreach my $repo_id (sort keys %$remotes) {
1377                if ($remotes->{$repo_id}->{url}) {
1378                        Git::SVN::fetch_all($repo_id, $remotes);
1379                }
1380        }
1381}
1382
1383# this command is special because it requires no metadata
1384sub cmd_commit_diff {
1385        my ($ta, $tb, $url) = @_;
1386        my $usage = "Usage: $0 commit-diff -r<revision> ".
1387                    "<tree-ish> <tree-ish> [<URL>]";
1388        fatal($usage) if (!defined $ta || !defined $tb);
1389        my $svn_path = '';
1390        if (!defined $url) {
1391                my $gs = eval { Git::SVN->new };
1392                if (!$gs) {
1393                        fatal("Needed URL or usable git-svn --id in ",
1394                              "the command-line\n", $usage);
1395                }
1396                $url = $gs->{url};
1397                $svn_path = $gs->{path};
1398        }
1399        unless (defined $_revision) {
1400                fatal("-r|--revision is a required argument\n", $usage);
1401        }
1402        if (defined $_message && defined $_file) {
1403                fatal("Both --message/-m and --file/-F specified ",
1404                      "for the commit message.\n",
1405                      "I have no idea what you mean");
1406        }
1407        if (defined $_file) {
1408                $_message = file_to_s($_file);
1409        } else {
1410                $_message ||= get_commit_entry($tb)->{log};
1411        }
1412        my $ra ||= Git::SVN::Ra->new($url);
1413        my $r = $_revision;
1414        if ($r eq 'HEAD') {
1415                $r = $ra->get_latest_revnum;
1416        } elsif ($r !~ /^\d+$/) {
1417                die "revision argument: $r not understood by git-svn\n";
1418        }
1419        my %ed_opts = ( r => $r,
1420                        log => $_message,
1421                        ra => $ra,
1422                        tree_a => $ta,
1423                        tree_b => $tb,
1424                        editor_cb => sub { print "Committed r$_[0]\n" },
1425                        svn_path => $svn_path );
1426        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
1427                print "No changes\n$ta == $tb\n";
1428        }
1429}
1430
1431sub escape_uri_only {
1432        my ($uri) = @_;
1433        my @tmp;
1434        foreach (split m{/}, $uri) {
1435                s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
1436                push @tmp, $_;
1437        }
1438        join('/', @tmp);
1439}
1440
1441sub escape_url {
1442        my ($url) = @_;
1443        if ($url =~ m#^([^:]+)://([^/]*)(.*)$#) {
1444                my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
1445                $url = "$scheme://$domain$uri";
1446        }
1447        $url;
1448}
1449
1450sub cmd_info {
1451        my $path = canonicalize_path(defined($_[0]) ? $_[0] : ".");
1452        my $fullpath = canonicalize_path($cmd_dir_prefix . $path);
1453        if (exists $_[1]) {
1454                die "Too many arguments specified\n";
1455        }
1456
1457        my ($file_type, $diff_status) = find_file_type_and_diff_status($path);
1458
1459        if (!$file_type && !$diff_status) {
1460                print STDERR "svn: '$path' is not under version control\n";
1461                exit 1;
1462        }
1463
1464        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1465        unless ($gs) {
1466                die "Unable to determine upstream SVN information from ",
1467                    "working tree history\n";
1468        }
1469
1470        # canonicalize_path() will return "" to make libsvn 1.5.x happy,
1471        $path = "." if $path eq "";
1472
1473        my $full_url = $url . ($fullpath eq "" ? "" : "/$fullpath");
1474
1475        if ($_url) {
1476                print escape_url($full_url), "\n";
1477                return;
1478        }
1479
1480        my $result = "Path: $path\n";
1481        $result .= "Name: " . basename($path) . "\n" if $file_type ne "dir";
1482        $result .= "URL: " . escape_url($full_url) . "\n";
1483
1484        eval {
1485                my $repos_root = $gs->repos_root;
1486                Git::SVN::remove_username($repos_root);
1487                $result .= "Repository Root: " . escape_url($repos_root) . "\n";
1488        };
1489        if ($@) {
1490                $result .= "Repository Root: (offline)\n";
1491        }
1492        ::_req_svn();
1493        $result .= "Repository UUID: $uuid\n" unless $diff_status eq "A" &&
1494                (::compare_svn_version('1.5.4') <= 0 || $file_type ne "dir");
1495        $result .= "Revision: " . ($diff_status eq "A" ? 0 : $rev) . "\n";
1496
1497        $result .= "Node Kind: " .
1498                   ($file_type eq "dir" ? "directory" : "file") . "\n";
1499
1500        my $schedule = $diff_status eq "A"
1501                       ? "add"
1502                       : ($diff_status eq "D" ? "delete" : "normal");
1503        $result .= "Schedule: $schedule\n";
1504
1505        if ($diff_status eq "A") {
1506                print $result, "\n";
1507                return;
1508        }
1509
1510        my ($lc_author, $lc_rev, $lc_date_utc);
1511        my @args = Git::SVN::Log::git_svn_log_cmd($rev, $rev, "--", $fullpath);
1512        my $log = command_output_pipe(@args);
1513        my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
1514        while (<$log>) {
1515                if (/^${esc_color}author (.+) <[^>]+> (\d+) ([\-\+]?\d+)$/o) {
1516                        $lc_author = $1;
1517                        $lc_date_utc = Git::SVN::Log::parse_git_date($2, $3);
1518                } elsif (/^${esc_color}    (git-svn-id:.+)$/o) {
1519                        (undef, $lc_rev, undef) = ::extract_metadata($1);
1520                }
1521        }
1522        close $log;
1523
1524        Git::SVN::Log::set_local_timezone();
1525
1526        $result .= "Last Changed Author: $lc_author\n";
1527        $result .= "Last Changed Rev: $lc_rev\n";
1528        $result .= "Last Changed Date: " .
1529                   Git::SVN::Log::format_svn_date($lc_date_utc) . "\n";
1530
1531        if ($file_type ne "dir") {
1532                my $text_last_updated_date =
1533                    ($diff_status eq "D" ? $lc_date_utc : (stat $path)[9]);
1534                $result .=
1535                    "Text Last Updated: " .
1536                    Git::SVN::Log::format_svn_date($text_last_updated_date) .
1537                    "\n";
1538                my $checksum;
1539                if ($diff_status eq "D") {
1540                        my ($fh, $ctx) =
1541                            command_output_pipe(qw(cat-file blob), "HEAD:$path");
1542                        if ($file_type eq "link") {
1543                                my $file_name = <$fh>;
1544                                $checksum = md5sum("link $file_name");
1545                        } else {
1546                                $checksum = md5sum($fh);
1547                        }
1548                        command_close_pipe($fh, $ctx);
1549                } elsif ($file_type eq "link") {
1550                        my $file_name =
1551                            command(qw(cat-file blob), "HEAD:$path");
1552                        $checksum =
1553                            md5sum("link " . $file_name);
1554                } else {
1555                        open FILE, "<", $path or die $!;
1556                        $checksum = md5sum(\*FILE);
1557                        close FILE or die $!;
1558                }
1559                $result .= "Checksum: " . $checksum . "\n";
1560        }
1561
1562        print $result, "\n";
1563}
1564
1565sub cmd_reset {
1566        my $target = shift || $_revision or die "SVN revision required\n";
1567        $target = $1 if $target =~ /^r(\d+)$/;
1568        $target =~ /^\d+$/ or die "Numeric SVN revision expected\n";
1569        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1570        unless ($gs) {
1571                die "Unable to determine upstream SVN information from ".
1572                    "history\n";
1573        }
1574        my ($r, $c) = $gs->find_rev_before($target, not $_fetch_parent);
1575        die "Cannot find SVN revision $target\n" unless defined($c);
1576        $gs->rev_map_set($r, $c, 'reset', $uuid);
1577        print "r$r = $c ($gs->{ref_id})\n";
1578}
1579
1580sub cmd_gc {
1581        if (!$can_compress) {
1582                warn "Compress::Zlib could not be found; unhandled.log " .
1583                     "files will not be compressed.\n";
1584        }
1585        find({ wanted => \&gc_directory, no_chdir => 1}, "$ENV{GIT_DIR}/svn");
1586}
1587
1588########################### utility functions #########################
1589
1590sub rebase_cmd {
1591        my @cmd = qw/rebase/;
1592        push @cmd, '-v' if $_verbose;
1593        push @cmd, qw/--merge/ if $_merge;
1594        push @cmd, "--strategy=$_strategy" if $_strategy;
1595        @cmd;
1596}
1597
1598sub post_fetch_checkout {
1599        return if $_no_checkout;
1600        my $gs = $Git::SVN::_head or return;
1601        return if verify_ref('refs/heads/master^0');
1602
1603        # look for "trunk" ref if it exists
1604        my $remote = Git::SVN::read_all_remotes()->{$gs->{repo_id}};
1605        my $fetch = $remote->{fetch};
1606        if ($fetch) {
1607                foreach my $p (keys %$fetch) {
1608                        basename($fetch->{$p}) eq 'trunk' or next;
1609                        $gs = Git::SVN->new($fetch->{$p}, $gs->{repo_id}, $p);
1610                        last;
1611                }
1612        }
1613
1614        my $valid_head = verify_ref('HEAD^0');
1615        command_noisy(qw(update-ref refs/heads/master), $gs->refname);
1616        return if ($valid_head || !verify_ref('HEAD^0'));
1617
1618        return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
1619        my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
1620        return if -f $index;
1621
1622        return if command_oneline(qw/rev-parse --is-inside-work-tree/) eq 'false';
1623        return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
1624        command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
1625        print STDERR "Checked out HEAD:\n  ",
1626                     $gs->full_url, " r", $gs->last_rev, "\n";
1627        if (auto_create_empty_directories($gs)) {
1628                $gs->mkemptydirs($gs->last_rev);
1629        }
1630}
1631
1632sub complete_svn_url {
1633        my ($url, $path) = @_;
1634        $path =~ s#/+$##;
1635        if ($path !~ m#^[a-z\+]+://#) {
1636                if (!defined $url || $url !~ m#^[a-z\+]+://#) {
1637                        fatal("E: '$path' is not a complete URL ",
1638                              "and a separate URL is not specified");
1639                }
1640                return ($url, $path);
1641        }
1642        return ($path, '');
1643}
1644
1645sub complete_url_ls_init {
1646        my ($ra, $repo_path, $switch, $pfx) = @_;
1647        unless ($repo_path) {
1648                print STDERR "W: $switch not specified\n";
1649                return;
1650        }
1651        $repo_path =~ s#/+$##;
1652        if ($repo_path =~ m#^[a-z\+]+://#) {
1653                $ra = Git::SVN::Ra->new($repo_path);
1654                $repo_path = '';
1655        } else {
1656                $repo_path =~ s#^/+##;
1657                unless ($ra) {
1658                        fatal("E: '$repo_path' is not a complete URL ",
1659                              "and a separate URL is not specified");
1660                }
1661        }
1662        my $url = $ra->{url};
1663        my $gs = Git::SVN->init($url, undef, undef, undef, 1);
1664        my $k = "svn-remote.$gs->{repo_id}.url";
1665        my $orig_url = eval { command_oneline(qw/config --get/, $k) };
1666        if ($orig_url && ($orig_url ne $gs->{url})) {
1667                die "$k already set: $orig_url\n",
1668                    "wanted to set to: $gs->{url}\n";
1669        }
1670        command_oneline('config', $k, $gs->{url}) unless $orig_url;
1671        my $remote_path = "$gs->{path}/$repo_path";
1672        $remote_path =~ s{%([0-9A-F]{2})}{chr hex($1)}ieg;
1673        $remote_path =~ s#/+#/#g;
1674        $remote_path =~ s#^/##g;
1675        $remote_path .= "/*" if $remote_path !~ /\*/;
1676        my ($n) = ($switch =~ /^--(\w+)/);
1677        if (length $pfx && $pfx !~ m#/$#) {
1678                die "--prefix='$pfx' must have a trailing slash '/'\n";
1679        }
1680        command_noisy('config',
1681                      '--add',
1682                      "svn-remote.$gs->{repo_id}.$n",
1683                      "$remote_path:refs/remotes/$pfx*" .
1684                        ('/*' x (($remote_path =~ tr/*/*/) - 1)) );
1685}
1686
1687sub verify_ref {
1688        my ($ref) = @_;
1689        eval { command_oneline([ 'rev-parse', '--verify', $ref ],
1690                               { STDERR => 0 }); };
1691}
1692
1693sub get_tree_from_treeish {
1694        my ($treeish) = @_;
1695        # $treeish can be a symbolic ref, too:
1696        my $type = command_oneline(qw/cat-file -t/, $treeish);
1697        my $expected;
1698        while ($type eq 'tag') {
1699                ($treeish, $type) = command(qw/cat-file tag/, $treeish);
1700        }
1701        if ($type eq 'commit') {
1702                $expected = (grep /^tree /, command(qw/cat-file commit/,
1703                                                    $treeish))[0];
1704                ($expected) = ($expected =~ /^tree ($sha1)$/o);
1705                die "Unable to get tree from $treeish\n" unless $expected;
1706        } elsif ($type eq 'tree') {
1707                $expected = $treeish;
1708        } else {
1709                die "$treeish is a $type, expected tree, tag or commit\n";
1710        }
1711        return $expected;
1712}
1713
1714sub get_commit_entry {
1715        my ($treeish) = shift;
1716        my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
1717        my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
1718        my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
1719        open my $log_fh, '>', $commit_editmsg or croak $!;
1720
1721        my $type = command_oneline(qw/cat-file -t/, $treeish);
1722        if ($type eq 'commit' || $type eq 'tag') {
1723                my ($msg_fh, $ctx) = command_output_pipe('cat-file',
1724                                                         $type, $treeish);
1725                my $in_msg = 0;
1726                my $author;
1727                my $saw_from = 0;
1728                my $msgbuf = "";
1729                while (<$msg_fh>) {
1730                        if (!$in_msg) {
1731                                $in_msg = 1 if (/^\s*$/);
1732                                $author = $1 if (/^author (.*>)/);
1733                        } elsif (/^git-svn-id: /) {
1734                                # skip this for now, we regenerate the
1735                                # correct one on re-fetch anyways
1736                                # TODO: set *:merge properties or like...
1737                        } else {
1738                                if (/^From:/ || /^Signed-off-by:/) {
1739                                        $saw_from = 1;
1740                                }
1741                                $msgbuf .= $_;
1742                        }
1743                }
1744                $msgbuf =~ s/\s+$//s;
1745                if ($Git::SVN::_add_author_from && defined($author)
1746                    && !$saw_from) {
1747                        $msgbuf .= "\n\nFrom: $author";
1748                }
1749                print $log_fh $msgbuf or croak $!;
1750                command_close_pipe($msg_fh, $ctx);
1751        }
1752        close $log_fh or croak $!;
1753
1754        if ($_edit || ($type eq 'tree')) {
1755                chomp(my $editor = command_oneline(qw(var GIT_EDITOR)));
1756                system('sh', '-c', $editor.' "$@"', $editor, $commit_editmsg);
1757        }
1758        rename $commit_editmsg, $commit_msg or croak $!;
1759        {
1760                require Encode;
1761                # SVN requires messages to be UTF-8 when entering the repo
1762                local $/;
1763                open $log_fh, '<', $commit_msg or croak $!;
1764                binmode $log_fh;
1765                chomp($log_entry{log} = <$log_fh>);
1766
1767                my $enc = Git::config('i18n.commitencoding') || 'UTF-8';
1768                my $msg = $log_entry{log};
1769
1770                eval { $msg = Encode::decode($enc, $msg, 1) };
1771                if ($@) {
1772                        die "Could not decode as $enc:\n", $msg,
1773                            "\nPerhaps you need to set i18n.commitencoding\n";
1774                }
1775
1776                eval { $msg = Encode::encode('UTF-8', $msg, 1) };
1777                die "Could not encode as UTF-8:\n$msg\n" if $@;
1778
1779                $log_entry{log} = $msg;
1780
1781                close $log_fh or croak $!;
1782        }
1783        unlink $commit_msg;
1784        \%log_entry;
1785}
1786
1787sub s_to_file {
1788        my ($str, $file, $mode) = @_;
1789        open my $fd,'>',$file or croak $!;
1790        print $fd $str,"\n" or croak $!;
1791        close $fd or croak $!;
1792        chmod ($mode &~ umask, $file) if (defined $mode);
1793}
1794
1795sub file_to_s {
1796        my $file = shift;
1797        open my $fd,'<',$file or croak "$!: file: $file\n";
1798        local $/;
1799        my $ret = <$fd>;
1800        close $fd or croak $!;
1801        $ret =~ s/\s*$//s;
1802        return $ret;
1803}
1804
1805# '<svn username> = real-name <email address>' mapping based on git-svnimport:
1806sub load_authors {
1807        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1808        my $log = $cmd eq 'log';
1809        while (<$authors>) {
1810                chomp;
1811                next unless /^(.+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
1812                my ($user, $name, $email) = ($1, $2, $3);
1813                if ($log) {
1814                        $Git::SVN::Log::rusers{"$name <$email>"} = $user;
1815                } else {
1816                        $users{$user} = [$name, $email];
1817                }
1818        }
1819        close $authors or croak $!;
1820}
1821
1822# convert GetOpt::Long specs for use by git-config
1823sub read_git_config {
1824        my $opts = shift;
1825        my @config_only;
1826        foreach my $o (keys %$opts) {
1827                # if we have mixedCase and a long option-only, then
1828                # it's a config-only variable that we don't need for
1829                # the command-line.
1830                push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
1831                my $v = $opts->{$o};
1832                my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
1833                $key =~ s/-//g;
1834                my $arg = 'git config';
1835                $arg .= ' --int' if ($o =~ /[:=]i$/);
1836                $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
1837                if (ref $v eq 'ARRAY') {
1838                        chomp(my @tmp = `$arg --get-all svn.$key`);
1839                        @$v = @tmp if @tmp;
1840                } else {
1841                        chomp(my $tmp = `$arg --get svn.$key`);
1842                        if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
1843                                $$v = $tmp;
1844                        }
1845                }
1846        }
1847        delete @$opts{@config_only} if @config_only;
1848}
1849
1850sub extract_metadata {
1851        my $id = shift or return (undef, undef, undef);
1852        my ($url, $rev, $uuid) = ($id =~ /^\s*git-svn-id:\s+(.*)\@(\d+)
1853                                                        \s([a-f\d\-]+)$/ix);
1854        if (!defined $rev || !$uuid || !$url) {
1855                # some of the original repositories I made had
1856                # identifiers like this:
1857                ($rev, $uuid) = ($id =~/^\s*git-svn-id:\s(\d+)\@([a-f\d\-]+)/i);
1858        }
1859        return ($url, $rev, $uuid);
1860}
1861
1862sub cmt_metadata {
1863        return extract_metadata((grep(/^git-svn-id: /,
1864                command(qw/cat-file commit/, shift)))[-1]);
1865}
1866
1867sub cmt_sha2rev_batch {
1868        my %s2r;
1869        my ($pid, $in, $out, $ctx) = command_bidi_pipe(qw/cat-file --batch/);
1870        my $list = shift;
1871
1872        foreach my $sha (@{$list}) {
1873                my $first = 1;
1874                my $size = 0;
1875                print $out $sha, "\n";
1876
1877                while (my $line = <$in>) {
1878                        if ($first && $line =~ /^[[:xdigit:]]{40}\smissing$/) {
1879                                last;
1880                        } elsif ($first &&
1881                               $line =~ /^[[:xdigit:]]{40}\scommit\s(\d+)$/) {
1882                                $first = 0;
1883                                $size = $1;
1884                                next;
1885                        } elsif ($line =~ /^(git-svn-id: )/) {
1886                                my (undef, $rev, undef) =
1887                                                      extract_metadata($line);
1888                                $s2r{$sha} = $rev;
1889                        }
1890
1891                        $size -= length($line);
1892                        last if ($size == 0);
1893                }
1894        }
1895
1896        command_close_bidi_pipe($pid, $in, $out, $ctx);
1897
1898        return \%s2r;
1899}
1900
1901sub working_head_info {
1902        my ($head, $refs) = @_;
1903        my @args = qw/rev-list --first-parent --pretty=medium/;
1904        my ($fh, $ctx) = command_output_pipe(@args, $head);
1905        my $hash;
1906        my %max;
1907        while (<$fh>) {
1908                if ( m{^commit ($::sha1)$} ) {
1909                        unshift @$refs, $hash if $hash and $refs;
1910                        $hash = $1;
1911                        next;
1912                }
1913                next unless s{^\s*(git-svn-id:)}{$1};
1914                my ($url, $rev, $uuid) = extract_metadata($_);
1915                if (defined $url && defined $rev) {
1916                        next if $max{$url} and $max{$url} < $rev;
1917                        if (my $gs = Git::SVN->find_by_url($url)) {
1918                                my $c = $gs->rev_map_get($rev, $uuid);
1919                                if ($c && $c eq $hash) {
1920                                        close $fh; # break the pipe
1921                                        return ($url, $rev, $uuid, $gs);
1922                                } else {
1923                                        $max{$url} ||= $gs->rev_map_max;
1924                                }
1925                        }
1926                }
1927        }
1928        command_close_pipe($fh, $ctx);
1929        (undef, undef, undef, undef);
1930}
1931
1932sub read_commit_parents {
1933        my ($parents, $c) = @_;
1934        chomp(my $p = command_oneline(qw/rev-list --parents -1/, $c));
1935        $p =~ s/^($c)\s*// or die "rev-list --parents -1 $c failed!\n";
1936        @{$parents->{$c}} = split(/ /, $p);
1937}
1938
1939sub linearize_history {
1940        my ($gs, $refs) = @_;
1941        my %parents;
1942        foreach my $c (@$refs) {
1943                read_commit_parents(\%parents, $c);
1944        }
1945
1946        my @linear_refs;
1947        my %skip = ();
1948        my $last_svn_commit = $gs->last_commit;
1949        foreach my $c (reverse @$refs) {
1950                next if $c eq $last_svn_commit;
1951                last if $skip{$c};
1952
1953                unshift @linear_refs, $c;
1954                $skip{$c} = 1;
1955
1956                # we only want the first parent to diff against for linear
1957                # history, we save the rest to inject when we finalize the
1958                # svn commit
1959                my $fp_a = verify_ref("$c~1");
1960                my $fp_b = shift @{$parents{$c}} if $parents{$c};
1961                if (!$fp_a || !$fp_b) {
1962                        die "Commit $c\n",
1963                            "has no parent commit, and therefore ",
1964                            "nothing to diff against.\n",
1965                            "You should be working from a repository ",
1966                            "originally created by git-svn\n";
1967                }
1968                if ($fp_a ne $fp_b) {
1969                        die "$c~1 = $fp_a, however parsing commit $c ",
1970                            "revealed that:\n$c~1 = $fp_b\nBUG!\n";
1971                }
1972
1973                foreach my $p (@{$parents{$c}}) {
1974                        $skip{$p} = 1;
1975                }
1976        }
1977        (\@linear_refs, \%parents);
1978}
1979
1980sub find_file_type_and_diff_status {
1981        my ($path) = @_;
1982        return ('dir', '') if $path eq '';
1983
1984        my $diff_output =
1985            command_oneline(qw(diff --cached --name-status --), $path) || "";
1986        my $diff_status = (split(' ', $diff_output))[0] || "";
1987
1988        my $ls_tree = command_oneline(qw(ls-tree HEAD), $path) || "";
1989
1990        return (undef, undef) if !$diff_status && !$ls_tree;
1991
1992        if ($diff_status eq "A") {
1993                return ("link", $diff_status) if -l $path;
1994                return ("dir", $diff_status) if -d $path;
1995                return ("file", $diff_status);
1996        }
1997
1998        my $mode = (split(' ', $ls_tree))[0] || "";
1999
2000        return ("link", $diff_status) if $mode eq "120000";
2001        return ("dir", $diff_status) if $mode eq "040000";
2002        return ("file", $diff_status);
2003}
2004
2005sub md5sum {
2006        my $arg = shift;
2007        my $ref = ref $arg;
2008        my $md5 = Digest::MD5->new();
2009        if ($ref eq 'GLOB' || $ref eq 'IO::File' || $ref eq 'File::Temp') {
2010                $md5->addfile($arg) or croak $!;
2011        } elsif ($ref eq 'SCALAR') {
2012                $md5->add($$arg) or croak $!;
2013        } elsif (!$ref) {
2014                $md5->add($arg) or croak $!;
2015        } else {
2016                ::fatal "Can't provide MD5 hash for unknown ref type: '", $ref, "'";
2017        }
2018        return $md5->hexdigest();
2019}
2020
2021sub gc_directory {
2022        if ($can_compress && -f $_ && basename($_) eq "unhandled.log") {
2023                my $out_filename = $_ . ".gz";
2024                open my $in_fh, "<", $_ or die "Unable to open $_: $!\n";
2025                binmode $in_fh;
2026                my $gz = Compress::Zlib::gzopen($out_filename, "ab") or
2027                                die "Unable to open $out_filename: $!\n";
2028
2029                my $res;
2030                while ($res = sysread($in_fh, my $str, 1024)) {
2031                        $gz->gzwrite($str) or
2032                                die "Unable to write: ".$gz->gzerror()."!\n";
2033                }
2034                unlink $_ or die "unlink $File::Find::name: $!\n";
2035        } elsif (-f $_ && basename($_) eq "index") {
2036                unlink $_ or die "unlink $_: $!\n";
2037        }
2038}
2039
2040package Git::SVN;
2041use strict;
2042use warnings;
2043use Fcntl qw/:DEFAULT :seek/;
2044use constant rev_map_fmt => 'NH40';
2045use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
2046            $_repack $_repack_flags $_use_svm_props $_head
2047            $_use_svnsync_props $no_reuse_existing $_minimize_url
2048            $_use_log_author $_add_author_from $_localtime/;
2049use Carp qw/croak/;
2050use File::Path qw/mkpath/;
2051use File::Copy qw/copy/;
2052use IPC::Open3;
2053use Time::Local;
2054use Memoize;  # core since 5.8.0, Jul 2002
2055use Memoize::Storable;
2056use POSIX qw(:signal_h);
2057
2058my ($_gc_nr, $_gc_period);
2059
2060# properties that we do not log:
2061my %SKIP_PROP;
2062BEGIN {
2063        %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
2064                                        svn:special svn:executable
2065                                        svn:entry:committed-rev
2066                                        svn:entry:last-author
2067                                        svn:entry:uuid
2068                                        svn:entry:committed-date/;
2069
2070        # some options are read globally, but can be overridden locally
2071        # per [svn-remote "..."] section.  Command-line options will *NOT*
2072        # override options set in an [svn-remote "..."] section
2073        no strict 'refs';
2074        for my $option (qw/follow_parent no_metadata use_svm_props
2075                           use_svnsync_props/) {
2076                my $key = $option;
2077                $key =~ tr/_//d;
2078                my $prop = "-$option";
2079                *$option = sub {
2080                        my ($self) = @_;
2081                        return $self->{$prop} if exists $self->{$prop};
2082                        my $k = "svn-remote.$self->{repo_id}.$key";
2083                        eval { command_oneline(qw/config --get/, $k) };
2084                        if ($@) {
2085                                $self->{$prop} = ${"Git::SVN::_$option"};
2086                        } else {
2087                                my $v = command_oneline(qw/config --bool/,$k);
2088                                $self->{$prop} = $v eq 'false' ? 0 : 1;
2089                        }
2090                        return $self->{$prop};
2091                }
2092        }
2093}
2094
2095
2096my (%LOCKFILES, %INDEX_FILES);
2097END {
2098        unlink keys %LOCKFILES if %LOCKFILES;
2099        unlink keys %INDEX_FILES if %INDEX_FILES;
2100}
2101
2102sub resolve_local_globs {
2103        my ($url, $fetch, $glob_spec) = @_;
2104        return unless defined $glob_spec;
2105        my $ref = $glob_spec->{ref};
2106        my $path = $glob_spec->{path};
2107        foreach (command(qw#for-each-ref --format=%(refname) refs/#)) {
2108                next unless m#^$ref->{regex}$#;
2109                my $p = $1;
2110                my $pathname = desanitize_refname($path->full_path($p));
2111                my $refname = desanitize_refname($ref->full_path($p));
2112                if (my $existing = $fetch->{$pathname}) {
2113                        if ($existing ne $refname) {
2114                                die "Refspec conflict:\n",
2115                                    "existing: $existing\n",
2116                                    " globbed: $refname\n";
2117                        }
2118                        my $u = (::cmt_metadata("$refname"))[0];
2119                        $u =~ s!^\Q$url\E(/|$)!! or die
2120                          "$refname: '$url' not found in '$u'\n";
2121                        if ($pathname ne $u) {
2122                                warn "W: Refspec glob conflict ",
2123                                     "(ref: $refname):\n",
2124                                     "expected path: $pathname\n",
2125                                     "    real path: $u\n",
2126                                     "Continuing ahead with $u\n";
2127                                next;
2128                        }
2129                } else {
2130                        $fetch->{$pathname} = $refname;
2131                }
2132        }
2133}
2134
2135sub parse_revision_argument {
2136        my ($base, $head) = @_;
2137        if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
2138                return ($base, $head);
2139        }
2140        return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
2141        return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
2142        return ($head, $head) if ($::_revision eq 'HEAD');
2143        return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
2144        return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
2145        die "revision argument: $::_revision not understood by git-svn\n";
2146}
2147
2148sub fetch_all {
2149        my ($repo_id, $remotes) = @_;
2150        if (ref $repo_id) {
2151                my $gs = $repo_id;
2152                $repo_id = undef;
2153                $repo_id = $gs->{repo_id};
2154        }
2155        $remotes ||= read_all_remotes();
2156        my $remote = $remotes->{$repo_id} or
2157                     die "[svn-remote \"$repo_id\"] unknown\n";
2158        my $fetch = $remote->{fetch};
2159        my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
2160        my (@gs, @globs);
2161        my $ra = Git::SVN::Ra->new($url);
2162        my $uuid = $ra->get_uuid;
2163        my $head = $ra->get_latest_revnum;
2164
2165        # ignore errors, $head revision may not even exist anymore
2166        eval { $ra->get_log("", $head, 0, 1, 0, 1, sub { $head = $_[1] }) };
2167        warn "W: $@\n" if $@;
2168
2169        my $base = defined $fetch ? $head : 0;
2170
2171        # read the max revs for wildcard expansion (branches/*, tags/*)
2172        foreach my $t (qw/branches tags/) {
2173                defined $remote->{$t} or next;
2174                push @globs, @{$remote->{$t}};
2175
2176                my $max_rev = eval { tmp_config(qw/--int --get/,
2177                                         "svn-remote.$repo_id.${t}-maxRev") };
2178                if (defined $max_rev && ($max_rev < $base)) {
2179                        $base = $max_rev;
2180                } elsif (!defined $max_rev) {
2181                        $base = 0;
2182                }
2183        }
2184
2185        if ($fetch) {
2186                foreach my $p (sort keys %$fetch) {
2187                        my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
2188                        my $lr = $gs->rev_map_max;
2189                        if (defined $lr) {
2190                                $base = $lr if ($lr < $base);
2191                        }
2192                        push @gs, $gs;
2193                }
2194        }
2195
2196        ($base, $head) = parse_revision_argument($base, $head);
2197        $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
2198}
2199
2200sub read_all_remotes {
2201        my $r = {};
2202        my $use_svm_props = eval { command_oneline(qw/config --bool
2203            svn.useSvmProps/) };
2204        $use_svm_props = $use_svm_props eq 'true' if $use_svm_props;
2205        my $svn_refspec = qr{\s*(.*?)\s*:\s*(.+?)\s*};
2206        foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
2207                if (m!^(.+)\.fetch=$svn_refspec$!) {
2208                        my ($remote, $local_ref, $remote_ref) = ($1, $2, $3);
2209                        die("svn-remote.$remote: remote ref '$remote_ref' "
2210                            . "must start with 'refs/'\n")
2211                                unless $remote_ref =~ m{^refs/};
2212                        $local_ref = uri_decode($local_ref);
2213                        $r->{$remote}->{fetch}->{$local_ref} = $remote_ref;
2214                        $r->{$remote}->{svm} = {} if $use_svm_props;
2215                } elsif (m!^(.+)\.usesvmprops=\s*(.*)\s*$!) {
2216                        $r->{$1}->{svm} = {};
2217                } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
2218                        $r->{$1}->{url} = $2;
2219                } elsif (m!^(.+)\.pushurl=\s*(.*)\s*$!) {
2220                        $r->{$1}->{pushurl} = $2;
2221                } elsif (m!^(.+)\.ignore-refs=\s*(.*)\s*$!) {
2222                        $r->{$1}->{ignore_refs_regex} = $2;
2223                } elsif (m!^(.+)\.(branches|tags)=$svn_refspec$!) {
2224                        my ($remote, $t, $local_ref, $remote_ref) =
2225                                                             ($1, $2, $3, $4);
2226                        die("svn-remote.$remote: remote ref '$remote_ref' ($t) "
2227                            . "must start with 'refs/'\n")
2228                                unless $remote_ref =~ m{^refs/};
2229                        $local_ref = uri_decode($local_ref);
2230                        my $rs = {
2231                            t => $t,
2232                            remote => $remote,
2233                            path => Git::SVN::GlobSpec->new($local_ref, 1),
2234                            ref => Git::SVN::GlobSpec->new($remote_ref, 0) };
2235                        if (length($rs->{ref}->{right}) != 0) {
2236                                die "The '*' glob character must be the last ",
2237                                    "character of '$remote_ref'\n";
2238                        }
2239                        push @{ $r->{$remote}->{$t} }, $rs;
2240                }
2241        }
2242
2243        map {
2244                if (defined $r->{$_}->{svm}) {
2245                        my $svm;
2246                        eval {
2247                                my $section = "svn-remote.$_";
2248                                $svm = {
2249                                        source => tmp_config('--get',
2250                                            "$section.svm-source"),
2251                                        replace => tmp_config('--get',
2252                                            "$section.svm-replace"),
2253                                }
2254                        };
2255                        $r->{$_}->{svm} = $svm;
2256                }
2257        } keys %$r;
2258
2259        foreach my $remote (keys %$r) {
2260                foreach ( grep { defined $_ }
2261                          map { $r->{$remote}->{$_} } qw(branches tags) ) {
2262                        foreach my $rs ( @$_ ) {
2263                                $rs->{ignore_refs_regex} =
2264                                    $r->{$remote}->{ignore_refs_regex};
2265                        }
2266                }
2267        }
2268
2269        $r;
2270}
2271
2272sub init_vars {
2273        $_gc_nr = $_gc_period = 1000;
2274        if (defined $_repack || defined $_repack_flags) {
2275               warn "Repack options are obsolete; they have no effect.\n";
2276        }
2277}
2278
2279sub verify_remotes_sanity {
2280        return unless -d $ENV{GIT_DIR};
2281        my %seen;
2282        foreach (command(qw/config -l/)) {
2283                if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
2284                        if ($seen{$1}) {
2285                                die "Remote ref refs/remote/$1 is tracked by",
2286                                    "\n  \"$_\"\nand\n  \"$seen{$1}\"\n",
2287                                    "Please resolve this ambiguity in ",
2288                                    "your git configuration file before ",
2289                                    "continuing\n";
2290                        }
2291                        $seen{$1} = $_;
2292                }
2293        }
2294}
2295
2296sub find_existing_remote {
2297        my ($url, $remotes) = @_;
2298        return undef if $no_reuse_existing;
2299        my $existing;
2300        foreach my $repo_id (keys %$remotes) {
2301                my $u = $remotes->{$repo_id}->{url} or next;
2302                next if $u ne $url;
2303                $existing = $repo_id;
2304                last;
2305        }
2306        $existing;
2307}
2308
2309sub init_remote_config {
2310        my ($self, $url, $no_write) = @_;
2311        $url =~ s!/+$!!; # strip trailing slash
2312        my $r = read_all_remotes();
2313        my $existing = find_existing_remote($url, $r);
2314        if ($existing) {
2315                unless ($no_write) {
2316                        print STDERR "Using existing ",
2317                                     "[svn-remote \"$existing\"]\n";
2318                }
2319                $self->{repo_id} = $existing;
2320        } elsif ($_minimize_url) {
2321                my $min_url = Git::SVN::Ra->new($url)->minimize_url;
2322                $existing = find_existing_remote($min_url, $r);
2323                if ($existing) {
2324                        unless ($no_write) {
2325                                print STDERR "Using existing ",
2326                                             "[svn-remote \"$existing\"]\n";
2327                        }
2328                        $self->{repo_id} = $existing;
2329                }
2330                if ($min_url ne $url) {
2331                        unless ($no_write) {
2332                                print STDERR "Using higher level of URL: ",
2333                                             "$url => $min_url\n";
2334                        }
2335                        my $old_path = $self->{path};
2336                        $self->{path} = $url;
2337                        $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
2338                        if (length $old_path) {
2339                                $self->{path} .= "/$old_path";
2340                        }
2341                        $url = $min_url;
2342                }
2343        }
2344        my $orig_url;
2345        if (!$existing) {
2346                # verify that we aren't overwriting anything:
2347                $orig_url = eval {
2348                        command_oneline('config', '--get',
2349                                        "svn-remote.$self->{repo_id}.url")
2350                };
2351                if ($orig_url && ($orig_url ne $url)) {
2352                        die "svn-remote.$self->{repo_id}.url already set: ",
2353                            "$orig_url\nwanted to set to: $url\n";
2354                }
2355        }
2356        my ($xrepo_id, $xpath) = find_ref($self->refname);
2357        if (!$no_write && defined $xpath) {
2358                die "svn-remote.$xrepo_id.fetch already set to track ",
2359                    "$xpath:", $self->refname, "\n";
2360        }
2361        unless ($no_write) {
2362                command_noisy('config',
2363                              "svn-remote.$self->{repo_id}.url", $url);
2364                $self->{path} =~ s{^/}{};
2365                $self->{path} =~ s{%([0-9A-F]{2})}{chr hex($1)}ieg;
2366                command_noisy('config', '--add',
2367                              "svn-remote.$self->{repo_id}.fetch",
2368                              "$self->{path}:".$self->refname);
2369        }
2370        $self->{url} = $url;
2371}
2372
2373sub find_by_url { # repos_root and, path are optional
2374        my ($class, $full_url, $repos_root, $path) = @_;
2375
2376        return undef unless defined $full_url;
2377        remove_username($full_url);
2378        remove_username($repos_root) if defined $repos_root;
2379        my $remotes = read_all_remotes();
2380        if (defined $full_url && defined $repos_root && !defined $path) {
2381                $path = $full_url;
2382                $path =~ s#^\Q$repos_root\E(?:/|$)##;
2383        }
2384        foreach my $repo_id (keys %$remotes) {
2385                my $u = $remotes->{$repo_id}->{url} or next;
2386                remove_username($u);
2387                next if defined $repos_root && $repos_root ne $u;
2388
2389                my $fetch = $remotes->{$repo_id}->{fetch} || {};
2390                foreach my $t (qw/branches tags/) {
2391                        foreach my $globspec (@{$remotes->{$repo_id}->{$t}}) {
2392                                resolve_local_globs($u, $fetch, $globspec);
2393                        }
2394                }
2395                my $p = $path;
2396                my $rwr = rewrite_root({repo_id => $repo_id});
2397                my $svm = $remotes->{$repo_id}->{svm}
2398                        if defined $remotes->{$repo_id}->{svm};
2399                unless (defined $p) {
2400                        $p = $full_url;
2401                        my $z = $u;
2402                        my $prefix = '';
2403                        if ($rwr) {
2404                                $z = $rwr;
2405                                remove_username($z);
2406                        } elsif (defined $svm) {
2407                                $z = $svm->{source};
2408                                $prefix = $svm->{replace};
2409                                $prefix =~ s#^\Q$u\E(?:/|$)##;
2410                                $prefix =~ s#/$##;
2411                        }
2412                        $p =~ s#^\Q$z\E(?:/|$)#$prefix# or next;
2413                }
2414                foreach my $f (keys %$fetch) {
2415                        next if $f ne $p;
2416                        return Git::SVN->new($fetch->{$f}, $repo_id, $f);
2417                }
2418        }
2419        undef;
2420}
2421
2422sub init {
2423        my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
2424        my $self = _new($class, $repo_id, $ref_id, $path);
2425        if (defined $url) {
2426                $self->init_remote_config($url, $no_write);
2427        }
2428        $self;
2429}
2430
2431sub find_ref {
2432        my ($ref_id) = @_;
2433        foreach (command(qw/config -l/)) {
2434                next unless m!^svn-remote\.(.+)\.fetch=
2435                              \s*(.*?)\s*:\s*(.+?)\s*$!x;
2436                my ($repo_id, $path, $ref) = ($1, $2, $3);
2437                if ($ref eq $ref_id) {
2438                        $path = '' if ($path =~ m#^\./?#);
2439                        return ($repo_id, $path);
2440                }
2441        }
2442        (undef, undef, undef);
2443}
2444
2445sub new {
2446        my ($class, $ref_id, $repo_id, $path) = @_;
2447        if (defined $ref_id && !defined $repo_id && !defined $path) {
2448                ($repo_id, $path) = find_ref($ref_id);
2449                if (!defined $repo_id) {
2450                        die "Could not find a \"svn-remote.*.fetch\" key ",
2451                            "in the repository configuration matching: ",
2452                            "$ref_id\n";
2453                }
2454        }
2455        my $self = _new($class, $repo_id, $ref_id, $path);
2456        if (!defined $self->{path} || !length $self->{path}) {
2457                my $fetch = command_oneline('config', '--get',
2458                                            "svn-remote.$repo_id.fetch",
2459                                            ":$ref_id\$") or
2460                     die "Failed to read \"svn-remote.$repo_id.fetch\" ",
2461                         "\":$ref_id\$\" in config\n";
2462                ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
2463        }
2464        $self->{path} =~ s{/+}{/}g;
2465        $self->{path} =~ s{\A/}{};
2466        $self->{path} =~ s{/\z}{};
2467        $self->{url} = command_oneline('config', '--get',
2468                                       "svn-remote.$repo_id.url") or
2469                  die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
2470        $self->{pushurl} = eval { command_oneline('config', '--get',
2471                                  "svn-remote.$repo_id.pushurl") };
2472        $self->rebuild;
2473        $self;
2474}
2475
2476sub refname {
2477        my ($refname) = $_[0]->{ref_id} ;
2478
2479        # It cannot end with a slash /, we'll throw up on this because
2480        # SVN can't have directories with a slash in their name, either:
2481        if ($refname =~ m{/$}) {
2482                die "ref: '$refname' ends with a trailing slash, this is ",
2483                    "not permitted by git nor Subversion\n";
2484        }
2485
2486        # It cannot have ASCII control character space, tilde ~, caret ^,
2487        # colon :, question-mark ?, asterisk *, space, or open bracket [
2488        # anywhere.
2489        #
2490        # Additionally, % must be escaped because it is used for escaping
2491        # and we want our escaped refname to be reversible
2492        $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
2493
2494        # no slash-separated component can begin with a dot .
2495        # /.* becomes /%2E*
2496        $refname =~ s{/\.}{/%2E}g;
2497
2498        # It cannot have two consecutive dots .. anywhere
2499        # .. becomes %2E%2E
2500        $refname =~ s{\.\.}{%2E%2E}g;
2501
2502        # trailing dots and .lock are not allowed
2503        # .$ becomes %2E and .lock becomes %2Elock
2504        $refname =~ s{\.(?=$|lock$)}{%2E};
2505
2506        # the sequence @{ is used to access the reflog
2507        # @{ becomes %40{
2508        $refname =~ s{\@\{}{%40\{}g;
2509
2510        return $refname;
2511}
2512
2513sub desanitize_refname {
2514        my ($refname) = @_;
2515        $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
2516        return $refname;
2517}
2518
2519sub svm_uuid {
2520        my ($self) = @_;
2521        return $self->{svm}->{uuid} if $self->svm;
2522        $self->ra;
2523        unless ($self->{svm}) {
2524                die "SVM UUID not cached, and reading remotely failed\n";
2525        }
2526        $self->{svm}->{uuid};
2527}
2528
2529sub svm {
2530        my ($self) = @_;
2531        return $self->{svm} if $self->{svm};
2532        my $svm;
2533        # see if we have it in our config, first:
2534        eval {
2535                my $section = "svn-remote.$self->{repo_id}";
2536                $svm = {
2537                  source => tmp_config('--get', "$section.svm-source"),
2538                  uuid => tmp_config('--get', "$section.svm-uuid"),
2539                  replace => tmp_config('--get', "$section.svm-replace"),
2540                }
2541        };
2542        if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
2543                $self->{svm} = $svm;
2544        }
2545        $self->{svm};
2546}
2547
2548sub _set_svm_vars {
2549        my ($self, $ra) = @_;
2550        return $ra if $self->svm;
2551
2552        my @err = ( "useSvmProps set, but failed to read SVM properties\n",
2553                    "(svm:source, svm:uuid) ",
2554                    "from the following URLs:\n" );
2555        sub read_svm_props {
2556                my ($self, $ra, $path, $r) = @_;
2557                my $props = ($ra->get_dir($path, $r))[2];
2558                my $src = $props->{'svm:source'};
2559                my $uuid = $props->{'svm:uuid'};
2560                return undef if (!$src || !$uuid);
2561
2562                chomp($src, $uuid);
2563
2564                $uuid =~ m{^[0-9a-f\-]{30,}$}i
2565                    or die "doesn't look right - svm:uuid is '$uuid'\n";
2566
2567                # the '!' is used to mark the repos_root!/relative/path
2568                $src =~ s{/?!/?}{/};
2569                $src =~ s{/+$}{}; # no trailing slashes please
2570                # username is of no interest
2571                $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
2572
2573                my $replace = $ra->{url};
2574                $replace .= "/$path" if length $path;
2575
2576                my $section = "svn-remote.$self->{repo_id}";
2577                tmp_config("$section.svm-source", $src);
2578                tmp_config("$section.svm-replace", $replace);
2579                tmp_config("$section.svm-uuid", $uuid);
2580                $self->{svm} = {
2581                        source => $src,
2582                        uuid => $uuid,
2583                        replace => $replace
2584                };
2585        }
2586
2587        my $r = $ra->get_latest_revnum;
2588        my $path = $self->{path};
2589        my %tried;
2590        while (length $path) {
2591                unless ($tried{"$self->{url}/$path"}) {
2592                        return $ra if $self->read_svm_props($ra, $path, $r);
2593                        $tried{"$self->{url}/$path"} = 1;
2594                }
2595                $path =~ s#/?[^/]+$##;
2596        }
2597        die "Path: '$path' should be ''\n" if $path ne '';
2598        return $ra if $self->read_svm_props($ra, $path, $r);
2599        $tried{"$self->{url}/$path"} = 1;
2600
2601        if ($ra->{repos_root} eq $self->{url}) {
2602                die @err, (map { "  $_\n" } keys %tried), "\n";
2603        }
2604
2605        # nope, make sure we're connected to the repository root:
2606        my $ok;
2607        my @tried_b;
2608        $path = $ra->{svn_path};
2609        $ra = Git::SVN::Ra->new($ra->{repos_root});
2610        while (length $path) {
2611                unless ($tried{"$ra->{url}/$path"}) {
2612                        $ok = $self->read_svm_props($ra, $path, $r);
2613                        last if $ok;
2614                        $tried{"$ra->{url}/$path"} = 1;
2615                }
2616                $path =~ s#/?[^/]+$##;
2617        }
2618        die "Path: '$path' should be ''\n" if $path ne '';
2619        $ok ||= $self->read_svm_props($ra, $path, $r);
2620        $tried{"$ra->{url}/$path"} = 1;
2621        if (!$ok) {
2622                die @err, (map { "  $_\n" } keys %tried), "\n";
2623        }
2624        Git::SVN::Ra->new($self->{url});
2625}
2626
2627sub svnsync {
2628        my ($self) = @_;
2629        return $self->{svnsync} if $self->{svnsync};
2630
2631        if ($self->no_metadata) {
2632                die "Can't have both 'noMetadata' and ",
2633                    "'useSvnsyncProps' options set!\n";
2634        }
2635        if ($self->rewrite_root) {
2636                die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
2637                    "options set!\n";
2638        }
2639        if ($self->rewrite_uuid) {
2640                die "Can't have both 'useSvnsyncProps' and 'rewriteUUID' ",
2641                    "options set!\n";
2642        }
2643
2644        my $svnsync;
2645        # see if we have it in our config, first:
2646        eval {
2647                my $section = "svn-remote.$self->{repo_id}";
2648
2649                my $url = tmp_config('--get', "$section.svnsync-url");
2650                ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2651                   die "doesn't look right - svn:sync-from-url is '$url'\n";
2652
2653                my $uuid = tmp_config('--get', "$section.svnsync-uuid");
2654                ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2655                   die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2656
2657                $svnsync = { url => $url, uuid => $uuid }
2658        };
2659        if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
2660                return $self->{svnsync} = $svnsync;
2661        }
2662
2663        my $err = "useSvnsyncProps set, but failed to read " .
2664                  "svnsync property: svn:sync-from-";
2665        my $rp = $self->ra->rev_proplist(0);
2666
2667        my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
2668        ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2669                   die "doesn't look right - svn:sync-from-url is '$url'\n";
2670
2671        my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
2672        ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2673                   die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2674
2675        my $section = "svn-remote.$self->{repo_id}";
2676        tmp_config('--add', "$section.svnsync-uuid", $uuid);
2677        tmp_config('--add', "$section.svnsync-url", $url);
2678        return $self->{svnsync} = { url => $url, uuid => $uuid };
2679}
2680
2681# this allows us to memoize our SVN::Ra UUID locally and avoid a
2682# remote lookup (useful for 'git svn log').
2683sub ra_uuid {
2684        my ($self) = @_;
2685        unless ($self->{ra_uuid}) {
2686                my $key = "svn-remote.$self->{repo_id}.uuid";
2687                my $uuid = eval { tmp_config('--get', $key) };
2688                if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/i) {
2689                        $self->{ra_uuid} = $uuid;
2690                } else {
2691                        die "ra_uuid called without URL\n" unless $self->{url};
2692                        $self->{ra_uuid} = $self->ra->get_uuid;
2693                        tmp_config('--add', $key, $self->{ra_uuid});
2694                }
2695        }
2696        $self->{ra_uuid};
2697}
2698
2699sub _set_repos_root {
2700        my ($self, $repos_root) = @_;
2701        my $k = "svn-remote.$self->{repo_id}.reposRoot";
2702        $repos_root ||= $self->ra->{repos_root};
2703        tmp_config($k, $repos_root);
2704        $repos_root;
2705}
2706
2707sub repos_root {
2708        my ($self) = @_;
2709        my $k = "svn-remote.$self->{repo_id}.reposRoot";
2710        eval { tmp_config('--get', $k) } || $self->_set_repos_root;
2711}
2712
2713sub ra {
2714        my ($self) = shift;
2715        my $ra = Git::SVN::Ra->new($self->{url});
2716        $self->_set_repos_root($ra->{repos_root});
2717        if ($self->use_svm_props && !$self->{svm}) {
2718                if ($self->no_metadata) {
2719                        die "Can't have both 'noMetadata' and ",
2720                            "'useSvmProps' options set!\n";
2721                } elsif ($self->use_svnsync_props) {
2722                        die "Can't have both 'useSvnsyncProps' and ",
2723                            "'useSvmProps' options set!\n";
2724                }
2725                $ra = $self->_set_svm_vars($ra);
2726                $self->{-want_revprops} = 1;
2727        }
2728        $ra;
2729}
2730
2731# prop_walk(PATH, REV, SUB)
2732# -------------------------
2733# Recursively traverse PATH at revision REV and invoke SUB for each
2734# directory that contains a SVN property.  SUB will be invoked as
2735# follows:  &SUB(gs, path, props);  where `gs' is this instance of
2736# Git::SVN, `path' the path to the directory where the properties
2737# `props' were found.  The `path' will be relative to point of checkout,
2738# that is, if url://repo/trunk is the current Git branch, and that
2739# directory contains a sub-directory `d', SUB will be invoked with `/d/'
2740# as `path' (note the trailing `/').
2741sub prop_walk {
2742        my ($self, $path, $rev, $sub) = @_;
2743
2744        $path =~ s#^/##;
2745        my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev);
2746        $path =~ s#^/*#/#g;
2747        my $p = $path;
2748        # Strip the irrelevant part of the path.
2749        $p =~ s#^/+\Q$self->{path}\E(/|$)#/#;
2750        # Ensure the path is terminated by a `/'.
2751        $p =~ s#/*$#/#;
2752
2753        # The properties contain all the internal SVN stuff nobody
2754        # (usually) cares about.
2755        my $interesting_props = 0;
2756        foreach (keys %{$props}) {
2757                # If it doesn't start with `svn:', it must be a
2758                # user-defined property.
2759                ++$interesting_props and next if $_ !~ /^svn:/;
2760                # FIXME: Fragile, if SVN adds new public properties,
2761                # this needs to be updated.
2762                ++$interesting_props if /^svn:(?:ignore|keywords|executable
2763                                                 |eol-style|mime-type
2764                                                 |externals|needs-lock)$/x;
2765        }
2766        &$sub($self, $p, $props) if $interesting_props;
2767
2768        foreach (sort keys %$dirent) {
2769                next if $dirent->{$_}->{kind} != $SVN::Node::dir;
2770                $self->prop_walk($self->{path} . $p . $_, $rev, $sub);
2771        }
2772}
2773
2774sub last_rev { ($_[0]->last_rev_commit)[0] }
2775sub last_commit { ($_[0]->last_rev_commit)[1] }
2776
2777# returns the newest SVN revision number and newest commit SHA1
2778sub last_rev_commit {
2779        my ($self) = @_;
2780        if (defined $self->{last_rev} && defined $self->{last_commit}) {
2781                return ($self->{last_rev}, $self->{last_commit});
2782        }
2783        my $c = ::verify_ref($self->refname.'^0');
2784        if ($c && !$self->use_svm_props && !$self->no_metadata) {
2785                my $rev = (::cmt_metadata($c))[1];
2786                if (defined $rev) {
2787                        ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
2788                        return ($rev, $c);
2789                }
2790        }
2791        my $map_path = $self->map_path;
2792        unless (-e $map_path) {
2793                ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
2794                return (undef, undef);
2795        }
2796        my ($rev, $commit) = $self->rev_map_max(1);
2797        ($self->{last_rev}, $self->{last_commit}) = ($rev, $commit);
2798        return ($rev, $commit);
2799}
2800
2801sub get_fetch_range {
2802        my ($self, $min, $max) = @_;
2803        $max ||= $self->ra->get_latest_revnum;
2804        $min ||= $self->rev_map_max;
2805        (++$min, $max);
2806}
2807
2808sub tmp_config {
2809        my (@args) = @_;
2810        my $old_def_config = "$ENV{GIT_DIR}/svn/config";
2811        my $config = "$ENV{GIT_DIR}/svn/.metadata";
2812        if (! -f $config && -f $old_def_config) {
2813                rename $old_def_config, $config or
2814                       die "Failed rename $old_def_config => $config: $!\n";
2815        }
2816        my $old_config = $ENV{GIT_CONFIG};
2817        $ENV{GIT_CONFIG} = $config;
2818        $@ = undef;
2819        my @ret = eval {
2820                unless (-f $config) {
2821                        mkfile($config);
2822                        open my $fh, '>', $config or
2823                            die "Can't open $config: $!\n";
2824                        print $fh "; This file is used internally by ",
2825                                  "git-svn\n" or die
2826                                  "Couldn't write to $config: $!\n";
2827                        print $fh "; You should not have to edit it\n" or
2828                              die "Couldn't write to $config: $!\n";
2829                        close $fh or die "Couldn't close $config: $!\n";
2830                }
2831                command('config', @args);
2832        };
2833        my $err = $@;
2834        if (defined $old_config) {
2835                $ENV{GIT_CONFIG} = $old_config;
2836        } else {
2837                delete $ENV{GIT_CONFIG};
2838        }
2839        die $err if $err;
2840        wantarray ? @ret : $ret[0];
2841}
2842
2843sub tmp_index_do {
2844        my ($self, $sub) = @_;
2845        my $old_index = $ENV{GIT_INDEX_FILE};
2846        $ENV{GIT_INDEX_FILE} = $self->{index};
2847        $@ = undef;
2848        my @ret = eval {
2849                my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
2850                mkpath([$dir]) unless -d $dir;
2851                &$sub;
2852        };
2853        my $err = $@;
2854        if (defined $old_index) {
2855                $ENV{GIT_INDEX_FILE} = $old_index;
2856        } else {
2857                delete $ENV{GIT_INDEX_FILE};
2858        }
2859        die $err if $err;
2860        wantarray ? @ret : $ret[0];
2861}
2862
2863sub assert_index_clean {
2864        my ($self, $treeish) = @_;
2865
2866        $self->tmp_index_do(sub {
2867                command_noisy('read-tree', $treeish) unless -e $self->{index};
2868                my $x = command_oneline('write-tree');
2869                my ($y) = (command(qw/cat-file commit/, $treeish) =~
2870                           /^tree ($::sha1)/mo);
2871                return if $y eq $x;
2872
2873                warn "Index mismatch: $y != $x\nrereading $treeish\n";
2874                unlink $self->{index} or die "unlink $self->{index}: $!\n";
2875                command_noisy('read-tree', $treeish);
2876                $x = command_oneline('write-tree');
2877                if ($y ne $x) {
2878                        ::fatal "trees ($treeish) $y != $x\n",
2879                                "Something is seriously wrong...";
2880                }
2881        });
2882}
2883
2884sub get_commit_parents {
2885        my ($self, $log_entry) = @_;
2886        my (%seen, @ret, @tmp);
2887        # legacy support for 'set-tree'; this is only used by set_tree_cb:
2888        if (my $ip = $self->{inject_parents}) {
2889                if (my $commit = delete $ip->{$log_entry->{revision}}) {
2890                        push @tmp, $commit;
2891                }
2892        }
2893        if (my $cur = ::verify_ref($self->refname.'^0')) {
2894                push @tmp, $cur;
2895        }
2896        if (my $ipd = $self->{inject_parents_dcommit}) {
2897                if (my $commit = delete $ipd->{$log_entry->{revision}}) {
2898                        push @tmp, @$commit;
2899                }
2900        }
2901        push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
2902        while (my $p = shift @tmp) {
2903                next if $seen{$p};
2904                $seen{$p} = 1;
2905                push @ret, $p;
2906        }
2907        @ret;
2908}
2909
2910sub rewrite_root {
2911        my ($self) = @_;
2912        return $self->{-rewrite_root} if exists $self->{-rewrite_root};
2913        my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
2914        my $rwr = eval { command_oneline(qw/config --get/, $k) };
2915        if ($rwr) {
2916                $rwr =~ s#/+$##;
2917                if ($rwr !~ m#^[a-z\+]+://#) {
2918                        die "$rwr is not a valid URL (key: $k)\n";
2919                }
2920        }
2921        $self->{-rewrite_root} = $rwr;
2922}
2923
2924sub rewrite_uuid {
2925        my ($self) = @_;
2926        return $self->{-rewrite_uuid} if exists $self->{-rewrite_uuid};
2927        my $k = "svn-remote.$self->{repo_id}.rewriteUUID";
2928        my $rwid = eval { command_oneline(qw/config --get/, $k) };
2929        if ($rwid) {
2930                $rwid =~ s#/+$##;
2931                if ($rwid !~ m#^[a-f0-9]{8}-(?:[a-f0-9]{4}-){3}[a-f0-9]{12}$#) {
2932                        die "$rwid is not a valid UUID (key: $k)\n";
2933                }
2934        }
2935        $self->{-rewrite_uuid} = $rwid;
2936}
2937
2938sub metadata_url {
2939        my ($self) = @_;
2940        ($self->rewrite_root || $self->{url}) .
2941           (length $self->{path} ? '/' . $self->{path} : '');
2942}
2943
2944sub full_url {
2945        my ($self) = @_;
2946        $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
2947}
2948
2949sub full_pushurl {
2950        my ($self) = @_;
2951        if ($self->{pushurl}) {
2952                return $self->{pushurl} . (length $self->{path} ? '/' .
2953                       $self->{path} : '');
2954        } else {
2955                return $self->full_url;
2956        }
2957}
2958
2959sub set_commit_header_env {
2960        my ($log_entry) = @_;
2961        my %env;
2962        foreach my $ned (qw/NAME EMAIL DATE/) {
2963                foreach my $ac (qw/AUTHOR COMMITTER/) {
2964                        $env{"GIT_${ac}_${ned}"} = $ENV{"GIT_${ac}_${ned}"};
2965                }
2966        }
2967
2968        $ENV{GIT_AUTHOR_NAME} = $log_entry->{name};
2969        $ENV{GIT_AUTHOR_EMAIL} = $log_entry->{email};
2970        $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
2971
2972        $ENV{GIT_COMMITTER_NAME} = (defined $log_entry->{commit_name})
2973                                                ? $log_entry->{commit_name}
2974                                                : $log_entry->{name};
2975        $ENV{GIT_COMMITTER_EMAIL} = (defined $log_entry->{commit_email})
2976                                                ? $log_entry->{commit_email}
2977                                                : $log_entry->{email};
2978        \%env;
2979}
2980
2981sub restore_commit_header_env {
2982        my ($env) = @_;
2983        foreach my $ned (qw/NAME EMAIL DATE/) {
2984                foreach my $ac (qw/AUTHOR COMMITTER/) {
2985                        my $k = "GIT_${ac}_${ned}";
2986                        if (defined $env->{$k}) {
2987                                $ENV{$k} = $env->{$k};
2988                        } else {
2989                                delete $ENV{$k};
2990                        }
2991                }
2992        }
2993}
2994
2995sub gc {
2996        command_noisy('gc', '--auto');
2997};
2998
2999sub do_git_commit {
3000        my ($self, $log_entry) = @_;
3001        my $lr = $self->last_rev;
3002        if (defined $lr && $lr >= $log_entry->{revision}) {
3003                die "Last fetched revision of ", $self->refname,
3004                    " was r$lr, but we are about to fetch: ",
3005                    "r$log_entry->{revision}!\n";
3006        }
3007        if (my $c = $self->rev_map_get($log_entry->{revision})) {
3008                croak "$log_entry->{revision} = $c already exists! ",
3009                      "Why are we refetching it?\n";
3010        }
3011        my $old_env = set_commit_header_env($log_entry);
3012        my $tree = $log_entry->{tree};
3013        if (!defined $tree) {
3014                $tree = $self->tmp_index_do(sub {
3015                                            command_oneline('write-tree') });
3016        }
3017        die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
3018
3019        my @exec = ('git', 'commit-tree', $tree);
3020        foreach ($self->get_commit_parents($log_entry)) {
3021                push @exec, '-p', $_;
3022        }
3023        defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
3024                                                                   or croak $!;
3025        binmode $msg_fh;
3026
3027        # we always get UTF-8 from SVN, but we may want our commits in
3028        # a different encoding.
3029        if (my $enc = Git::config('i18n.commitencoding')) {
3030                require Encode;
3031                Encode::from_to($log_entry->{log}, 'UTF-8', $enc);
3032        }
3033        print $msg_fh $log_entry->{log} or croak $!;
3034        restore_commit_header_env($old_env);
3035        unless ($self->no_metadata) {
3036                print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
3037                              or croak $!;
3038        }
3039        $msg_fh->flush == 0 or croak $!;
3040        close $msg_fh or croak $!;
3041        chomp(my $commit = do { local $/; <$out_fh> });
3042        close $out_fh or croak $!;
3043        waitpid $pid, 0;
3044        croak $? if $?;
3045        if ($commit !~ /^$::sha1$/o) {
3046                die "Failed to commit, invalid sha1: $commit\n";
3047        }
3048
3049        $self->rev_map_set($log_entry->{revision}, $commit, 1);
3050
3051        $self->{last_rev} = $log_entry->{revision};
3052        $self->{last_commit} = $commit;
3053        print "r$log_entry->{revision}" unless $::_q > 1;
3054        if (defined $log_entry->{svm_revision}) {
3055                 print " (\@$log_entry->{svm_revision})" unless $::_q > 1;
3056                 $self->rev_map_set($log_entry->{svm_revision}, $commit,
3057                                   0, $self->svm_uuid);
3058        }
3059        print " = $commit ($self->{ref_id})\n" unless $::_q > 1;
3060        if (--$_gc_nr == 0) {
3061                $_gc_nr = $_gc_period;
3062                gc();
3063        }
3064        return $commit;
3065}
3066
3067sub match_paths {
3068        my ($self, $paths, $r) = @_;
3069        return 1 if $self->{path} eq '';
3070        if (my $path = $paths->{"/$self->{path}"}) {
3071                return ($path->{action} eq 'D') ? 0 : 1;
3072        }
3073        $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
3074        if (grep /$self->{path_regex}/, keys %$paths) {
3075                return 1;
3076        }
3077        my $c = '';
3078        foreach (split m#/#, $self->{path}) {
3079                $c .= "/$_";
3080                next unless ($paths->{$c} &&
3081                             ($paths->{$c}->{action} =~ /^[AR]$/));
3082                if ($self->ra->check_path($self->{path}, $r) ==
3083                    $SVN::Node::dir) {
3084                        return 1;
3085                }
3086        }
3087        return 0;
3088}
3089
3090sub find_parent_branch {
3091        my ($self, $paths, $rev) = @_;
3092        return undef unless $self->follow_parent;
3093        unless (defined $paths) {
3094                my $err_handler = $SVN::Error::handler;
3095                $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
3096                $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1,
3097                                   sub { $paths = $_[0] });
3098                $SVN::Error::handler = $err_handler;
3099        }
3100        return undef unless defined $paths;
3101
3102        # look for a parent from another branch:
3103        my @b_path_components = split m#/#, $self->{path};
3104        my @a_path_components;
3105        my $i;
3106        while (@b_path_components) {
3107                $i = $paths->{'/'.join('/', @b_path_components)};
3108                last if $i && defined $i->{copyfrom_path};
3109                unshift(@a_path_components, pop(@b_path_components));
3110        }
3111        return undef unless defined $i && defined $i->{copyfrom_path};
3112        my $branch_from = $i->{copyfrom_path};
3113        if (@a_path_components) {
3114                print STDERR "branch_from: $branch_from => ";
3115                $branch_from .= '/'.join('/', @a_path_components);
3116                print STDERR $branch_from, "\n";
3117        }
3118        my $r = $i->{copyfrom_rev};
3119        my $repos_root = $self->ra->{repos_root};
3120        my $url = $self->ra->{url};
3121        my $new_url = $url . $branch_from;
3122        print STDERR  "Found possible branch point: ",
3123                      "$new_url => ", $self->full_url, ", $r\n"
3124                      unless $::_q > 1;
3125        $branch_from =~ s#^/##;
3126        my $gs = $self->other_gs($new_url, $url,
3127                                 $branch_from, $r, $self->{ref_id});
3128        my ($r0, $parent) = $gs->find_rev_before($r, 1);
3129        {
3130                my ($base, $head);
3131                if (!defined $r0 || !defined $parent) {
3132                        ($base, $head) = parse_revision_argument(0, $r);
3133                } else {
3134                        if ($r0 < $r) {
3135                                $gs->ra->get_log([$gs->{path}], $r0 + 1, $r, 1,
3136                                        0, 1, sub { $base = $_[1] - 1 });
3137                        }
3138                }
3139                if (defined $base && $base <= $r) {
3140                        $gs->fetch($base, $r);
3141                }
3142                ($r0, $parent) = $gs->find_rev_before($r, 1);
3143        }
3144        if (defined $r0 && defined $parent) {
3145                print STDERR "Found branch parent: ($self->{ref_id}) $parent\n"
3146                             unless $::_q > 1;
3147                my $ed;
3148                if ($self->ra->can_do_switch) {
3149                        $self->assert_index_clean($parent);
3150                        print STDERR "Following parent with do_switch\n"
3151                                     unless $::_q > 1;
3152                        # do_switch works with svn/trunk >= r22312, but that
3153                        # is not included with SVN 1.4.3 (the latest version
3154                        # at the moment), so we can't rely on it
3155                        $self->{last_rev} = $r0;
3156                        $self->{last_commit} = $parent;
3157                        $ed = SVN::Git::Fetcher->new($self, $gs->{path});
3158                        $gs->ra->gs_do_switch($r0, $rev, $gs,
3159                                              $self->full_url, $ed)
3160                          or die "SVN connection failed somewhere...\n";
3161                } elsif ($self->ra->trees_match($new_url, $r0,
3162                                                $self->full_url, $rev)) {
3163                        print STDERR "Trees match:\n",
3164                                     "  $new_url\@$r0\n",
3165                                     "  ${\$self->full_url}\@$rev\n",
3166                                     "Following parent with no changes\n"
3167                                     unless $::_q > 1;
3168                        $self->tmp_index_do(sub {
3169                            command_noisy('read-tree', $parent);
3170                        });
3171                        $self->{last_commit} = $parent;
3172                } else {
3173                        print STDERR "Following parent with do_update\n"
3174                                     unless $::_q > 1;
3175                        $ed = SVN::Git::Fetcher->new($self);
3176                        $self->ra->gs_do_update($rev, $rev, $self, $ed)
3177                          or die "SVN connection failed somewhere...\n";
3178                }
3179                print STDERR "Successfully followed parent\n" unless $::_q > 1;
3180                return $self->make_log_entry($rev, [$parent], $ed);
3181        }
3182        return undef;
3183}
3184
3185sub do_fetch {
3186        my ($self, $paths, $rev) = @_;
3187        my $ed;
3188        my ($last_rev, @parents);
3189        if (my $lc = $self->last_commit) {
3190                # we can have a branch that was deleted, then re-added
3191                # under the same name but copied from another path, in
3192                # which case we'll have multiple parents (we don't
3193                # want to break the original ref, nor lose copypath info):
3194                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
3195                        push @{$log_entry->{parents}}, $lc;
3196                        return $log_entry;
3197                }
3198                $ed = SVN::Git::Fetcher->new($self);
3199                $last_rev = $self->{last_rev};
3200                $ed->{c} = $lc;
3201                @parents = ($lc);
3202        } else {
3203                $last_rev = $rev;
3204                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
3205                        return $log_entry;
3206                }
3207                $ed = SVN::Git::Fetcher->new($self);
3208        }
3209        unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
3210                die "SVN connection failed somewhere...\n";
3211        }
3212        $self->make_log_entry($rev, \@parents, $ed);
3213}
3214
3215sub mkemptydirs {
3216        my ($self, $r) = @_;
3217
3218        sub scan {
3219                my ($r, $empty_dirs, $line) = @_;
3220                if (defined $r && $line =~ /^r(\d+)$/) {
3221                        return 0 if $1 > $r;
3222                } elsif ($line =~ /^  \+empty_dir: (.+)$/) {
3223                        $empty_dirs->{$1} = 1;
3224                } elsif ($line =~ /^  \-empty_dir: (.+)$/) {
3225                        my @d = grep {m[^\Q$1\E(/|$)]} (keys %$empty_dirs);
3226                        delete @$empty_dirs{@d};
3227                }
3228                1; # continue
3229        };
3230
3231        my %empty_dirs = ();
3232        my $gz_file = "$self->{dir}/unhandled.log.gz";
3233        if (-f $gz_file) {
3234                if (!$can_compress) {
3235                        warn "Compress::Zlib could not be found; ",
3236                             "empty directories in $gz_file will not be read\n";
3237                } else {
3238                        my $gz = Compress::Zlib::gzopen($gz_file, "rb") or
3239                                die "Unable to open $gz_file: $!\n";
3240                        my $line;
3241                        while ($gz->gzreadline($line) > 0) {
3242                                scan($r, \%empty_dirs, $line) or last;
3243                        }
3244                        $gz->gzclose;
3245                }
3246        }
3247
3248        if (open my $fh, '<', "$self->{dir}/unhandled.log") {
3249                binmode $fh or croak "binmode: $!";
3250                while (<$fh>) {
3251                        scan($r, \%empty_dirs, $_) or last;
3252                }
3253                close $fh;
3254        }
3255
3256        my $strip = qr/\A\Q$self->{path}\E(?:\/|$)/;
3257        foreach my $d (sort keys %empty_dirs) {
3258                $d = uri_decode($d);
3259                $d =~ s/$strip//;
3260                next unless length($d);
3261                next if -d $d;
3262                if (-e $d) {
3263                        warn "$d exists but is not a directory\n";
3264                } else {
3265                        print "creating empty directory: $d\n";
3266                        mkpath([$d]);
3267                }
3268        }
3269}
3270
3271sub get_untracked {
3272        my ($self, $ed) = @_;
3273        my @out;
3274        my $h = $ed->{empty};
3275        foreach (sort keys %$h) {
3276                my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
3277                push @out, "  $act: " . uri_encode($_);
3278                warn "W: $act: $_\n";
3279        }
3280        foreach my $t (qw/dir_prop file_prop/) {
3281                $h = $ed->{$t} or next;
3282                foreach my $path (sort keys %$h) {
3283                        my $ppath = $path eq '' ? '.' : $path;
3284                        foreach my $prop (sort keys %{$h->{$path}}) {
3285                                next if $SKIP_PROP{$prop};
3286                                my $v = $h->{$path}->{$prop};
3287                                my $t_ppath_prop = "$t: " .
3288                                                    uri_encode($ppath) . ' ' .
3289                                                    uri_encode($prop);
3290                                if (defined $v) {
3291                                        push @out, "  +$t_ppath_prop " .
3292                                                   uri_encode($v);
3293                                } else {
3294                                        push @out, "  -$t_ppath_prop";
3295                                }
3296                        }
3297                }
3298        }
3299        foreach my $t (qw/absent_file absent_directory/) {
3300                $h = $ed->{$t} or next;
3301                foreach my $parent (sort keys %$h) {
3302                        foreach my $path (sort @{$h->{$parent}}) {
3303                                push @out, "  $t: " .
3304                                           uri_encode("$parent/$path");
3305                                warn "W: $t: $parent/$path ",
3306                                     "Insufficient permissions?\n";
3307                        }
3308                }
3309        }
3310        \@out;
3311}
3312
3313sub get_tz {
3314        # some systmes don't handle or mishandle %z, so be creative.
3315        my $t = shift || time;
3316        my $gm = timelocal(gmtime($t));
3317        my $sign = qw( + + - )[ $t <=> $gm ];
3318        return sprintf("%s%02d%02d", $sign, (gmtime(abs($t - $gm)))[2,1]);
3319}
3320
3321# parse_svn_date(DATE)
3322# --------------------
3323# Given a date (in UTC) from Subversion, return a string in the format
3324# "<TZ Offset> <local date/time>" that Git will use.
3325#
3326# By default the parsed date will be in UTC; if $Git::SVN::_localtime
3327# is true we'll convert it to the local timezone instead.
3328sub parse_svn_date {
3329        my $date = shift || return '+0000 1970-01-01 00:00:00';
3330        my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
3331                                            (\d\d)\:(\d\d)\:(\d\d)\.\d*Z$/x) or
3332                                         croak "Unable to parse date: $date\n";
3333        my $parsed_date;    # Set next.
3334
3335        if ($Git::SVN::_localtime) {
3336                # Translate the Subversion datetime to an epoch time.
3337                # Begin by switching ourselves to $date's timezone, UTC.
3338                my $old_env_TZ = $ENV{TZ};
3339                $ENV{TZ} = 'UTC';
3340
3341                my $epoch_in_UTC =
3342                    POSIX::strftime('%s', $S, $M, $H, $d, $m - 1, $Y - 1900);
3343
3344                # Determine our local timezone (including DST) at the
3345                # time of $epoch_in_UTC.  $Git::SVN::Log::TZ stored the
3346                # value of TZ, if any, at the time we were run.
3347                if (defined $Git::SVN::Log::TZ) {
3348                        $ENV{TZ} = $Git::SVN::Log::TZ;
3349                } else {
3350                        delete $ENV{TZ};
3351                }
3352
3353                my $our_TZ = get_tz();
3354
3355                # This converts $epoch_in_UTC into our local timezone.
3356                my ($sec, $min, $hour, $mday, $mon, $year,
3357                    $wday, $yday, $isdst) = localtime($epoch_in_UTC);
3358
3359                $parsed_date = sprintf('%s %04d-%02d-%02d %02d:%02d:%02d',
3360                                       $our_TZ, $year + 1900, $mon + 1,
3361                                       $mday, $hour, $min, $sec);
3362
3363                # Reset us to the timezone in effect when we entered
3364                # this routine.
3365                if (defined $old_env_TZ) {
3366                        $ENV{TZ} = $old_env_TZ;
3367                } else {
3368                        delete $ENV{TZ};
3369                }
3370        } else {
3371                $parsed_date = "+0000 $Y-$m-$d $H:$M:$S";
3372        }
3373
3374        return $parsed_date;
3375}
3376
3377sub other_gs {
3378        my ($self, $new_url, $url,
3379            $branch_from, $r, $old_ref_id) = @_;
3380        my $gs = Git::SVN->find_by_url($new_url, $url, $branch_from);
3381        unless ($gs) {
3382                my $ref_id = $old_ref_id;
3383                $ref_id =~ s/\@\d+-*$//;
3384                $ref_id .= "\@$r";
3385                # just grow a tail if we're not unique enough :x
3386                $ref_id .= '-' while find_ref($ref_id);
3387                my ($u, $p, $repo_id) = ($new_url, '', $ref_id);
3388                if ($u =~ s#^\Q$url\E(/|$)##) {
3389                        $p = $u;
3390                        $u = $url;
3391                        $repo_id = $self->{repo_id};
3392                }
3393                while (1) {
3394                        # It is possible to tag two different subdirectories at
3395                        # the same revision.  If the url for an existing ref
3396                        # does not match, we must either find a ref with a
3397                        # matching url or create a new ref by growing a tail.
3398                        $gs = Git::SVN->init($u, $p, $repo_id, $ref_id, 1);
3399                        my (undef, $max_commit) = $gs->rev_map_max(1);
3400                        last if (!$max_commit);
3401                        my ($url) = ::cmt_metadata($max_commit);
3402                        last if ($url eq $gs->metadata_url);
3403                        $ref_id .= '-';
3404                }
3405                print STDERR "Initializing parent: $ref_id\n" unless $::_q > 1;
3406        }
3407        $gs
3408}
3409
3410sub call_authors_prog {
3411        my ($orig_author) = @_;
3412        $orig_author = command_oneline('rev-parse', '--sq-quote', $orig_author);
3413        my $author = `$::_authors_prog $orig_author`;
3414        if ($? != 0) {
3415                die "$::_authors_prog failed with exit code $?\n"
3416        }
3417        if ($author =~ /^\s*(.+?)\s*<(.*)>\s*$/) {
3418                my ($name, $email) = ($1, $2);
3419                $email = undef if length $2 == 0;
3420                return [$name, $email];
3421        } else {
3422                die "Author: $orig_author: $::_authors_prog returned "
3423                        . "invalid author format: $author\n";
3424        }
3425}
3426
3427sub check_author {
3428        my ($author) = @_;
3429        if (!defined $author || length $author == 0) {
3430                $author = '(no author)';
3431        }
3432        if (!defined $::users{$author}) {
3433                if (defined $::_authors_prog) {
3434                        $::users{$author} = call_authors_prog($author);
3435                } elsif (defined $::_authors) {
3436                        die "Author: $author not defined in $::_authors file\n";
3437                }
3438        }
3439        $author;
3440}
3441
3442sub find_extra_svk_parents {
3443        my ($self, $ed, $tickets, $parents) = @_;
3444        # aha!  svk:merge property changed...
3445        my @tickets = split "\n", $tickets;
3446        my @known_parents;
3447        for my $ticket ( @tickets ) {
3448                my ($uuid, $path, $rev) = split /:/, $ticket;
3449                if ( $uuid eq $self->ra_uuid ) {
3450                        my $url = $self->{url};
3451                        my $repos_root = $url;
3452                        my $branch_from = $path;
3453                        $branch_from =~ s{^/}{};
3454                        my $gs = $self->other_gs($repos_root."/".$branch_from,
3455                                                 $url,
3456                                                 $branch_from,
3457                                                 $rev,
3458                                                 $self->{ref_id});
3459                        if ( my $commit = $gs->rev_map_get($rev, $uuid) ) {
3460                                # wahey!  we found it, but it might be
3461                                # an old one (!)
3462                                push @known_parents, [ $rev, $commit ];
3463                        }
3464                }
3465        }
3466        # Ordering matters; highest-numbered commit merge tickets
3467        # first, as they may account for later merge ticket additions
3468        # or changes.
3469        @known_parents = map {$_->[1]} sort {$b->[0] <=> $a->[0]} @known_parents;
3470        for my $parent ( @known_parents ) {
3471                my @cmd = ('rev-list', $parent, map { "^$_" } @$parents );
3472                my ($msg_fh, $ctx) = command_output_pipe(@cmd);
3473                my $new;
3474                while ( <$msg_fh> ) {
3475                        $new=1;last;
3476                }
3477                command_close_pipe($msg_fh, $ctx);
3478                if ( $new ) {
3479                        print STDERR
3480                            "Found merge parent (svk:merge ticket): $parent\n";
3481                        push @$parents, $parent;
3482                }
3483        }
3484}
3485
3486sub lookup_svn_merge {
3487        my $uuid = shift;
3488        my $url = shift;
3489        my $merge = shift;
3490
3491        my ($source, $revs) = split ":", $merge;
3492        my $path = $source;
3493        $path =~ s{^/}{};
3494        my $gs = Git::SVN->find_by_url($url.$source, $url, $path);
3495        if ( !$gs ) {
3496                warn "Couldn't find revmap for $url$source\n";
3497                return;
3498        }
3499        my @ranges = split ",", $revs;
3500        my ($tip, $tip_commit);
3501        my @merged_commit_ranges;
3502        # find the tip
3503        for my $range ( @ranges ) {
3504                my ($bottom, $top) = split "-", $range;
3505                $top ||= $bottom;
3506                my $bottom_commit = $gs->find_rev_after( $bottom, 1, $top );
3507                my $top_commit = $gs->find_rev_before( $top, 1, $bottom );
3508
3509                unless ($top_commit and $bottom_commit) {
3510                        warn "W:unknown path/rev in svn:mergeinfo "
3511                                ."dirprop: $source:$range\n";
3512                        next;
3513                }
3514
3515                if (scalar(command('rev-parse', "$bottom_commit^@"))) {
3516                        push @merged_commit_ranges,
3517                             "$bottom_commit^..$top_commit";
3518                } else {
3519                        push @merged_commit_ranges, "$top_commit";
3520                }
3521
3522                if ( !defined $tip or $top > $tip ) {
3523                        $tip = $top;
3524                        $tip_commit = $top_commit;
3525                }
3526        }
3527        return ($tip_commit, @merged_commit_ranges);
3528}
3529
3530sub _rev_list {
3531        my ($msg_fh, $ctx) = command_output_pipe(
3532                "rev-list", @_,
3533               );
3534        my @rv;
3535        while ( <$msg_fh> ) {
3536                chomp;
3537                push @rv, $_;
3538        }
3539        command_close_pipe($msg_fh, $ctx);
3540        @rv;
3541}
3542
3543sub check_cherry_pick {
3544        my $base = shift;
3545        my $tip = shift;
3546        my $parents = shift;
3547        my @ranges = @_;
3548        my %commits = map { $_ => 1 }
3549                _rev_list("--no-merges", $tip, "--not", $base, @$parents, "--");
3550        for my $range ( @ranges ) {
3551                delete @commits{_rev_list($range, "--")};
3552        }
3553        for my $commit (keys %commits) {
3554                if (has_no_changes($commit)) {
3555                        delete $commits{$commit};
3556                }
3557        }
3558        return (keys %commits);
3559}
3560
3561sub has_no_changes {
3562        my $commit = shift;
3563
3564        my @revs = split / /, command_oneline(
3565                qw(rev-list --parents -1 -m), $commit);
3566
3567        # Commits with no parents, e.g. the start of a partial branch,
3568        # have changes by definition.
3569        return 1 if (@revs < 2);
3570
3571        # Commits with multiple parents, e.g a merge, have no changes
3572        # by definition.
3573        return 0 if (@revs > 2);
3574
3575        return (command_oneline("rev-parse", "$commit^{tree}") eq
3576                command_oneline("rev-parse", "$commit~1^{tree}"));
3577}
3578
3579# The GIT_DIR environment variable is not always set until after the command
3580# line arguments are processed, so we can't memoize in a BEGIN block.
3581{
3582        my $memoized = 0;
3583
3584        sub memoize_svn_mergeinfo_functions {
3585                return if $memoized;
3586                $memoized = 1;
3587
3588                my $cache_path = "$ENV{GIT_DIR}/svn/.caches/";
3589                mkpath([$cache_path]) unless -d $cache_path;
3590
3591                tie my %lookup_svn_merge_cache => 'Memoize::Storable',
3592                    "$cache_path/lookup_svn_merge.db", 'nstore';
3593                memoize 'lookup_svn_merge',
3594                        SCALAR_CACHE => 'FAULT',
3595                        LIST_CACHE => ['HASH' => \%lookup_svn_merge_cache],
3596                ;
3597
3598                tie my %check_cherry_pick_cache => 'Memoize::Storable',
3599                    "$cache_path/check_cherry_pick.db", 'nstore';
3600                memoize 'check_cherry_pick',
3601                        SCALAR_CACHE => 'FAULT',
3602                        LIST_CACHE => ['HASH' => \%check_cherry_pick_cache],
3603                ;
3604
3605                tie my %has_no_changes_cache => 'Memoize::Storable',
3606                    "$cache_path/has_no_changes.db", 'nstore';
3607                memoize 'has_no_changes',
3608                        SCALAR_CACHE => ['HASH' => \%has_no_changes_cache],
3609                        LIST_CACHE => 'FAULT',
3610                ;
3611        }
3612
3613        sub unmemoize_svn_mergeinfo_functions {
3614                return if not $memoized;
3615                $memoized = 0;
3616
3617                Memoize::unmemoize 'lookup_svn_merge';
3618                Memoize::unmemoize 'check_cherry_pick';
3619                Memoize::unmemoize 'has_no_changes';
3620        }
3621
3622        Memoize::memoize 'Git::SVN::repos_root';
3623}
3624
3625END {
3626        # Force cache writeout explicitly instead of waiting for
3627        # global destruction to avoid segfault in Storable:
3628        # http://rt.cpan.org/Public/Bug/Display.html?id=36087
3629        unmemoize_svn_mergeinfo_functions();
3630}
3631
3632sub parents_exclude {
3633        my $parents = shift;
3634        my @commits = @_;
3635        return unless @commits;
3636
3637        my @excluded;
3638        my $excluded;
3639        do {
3640                my @cmd = ('rev-list', "-1", @commits, "--not", @$parents );
3641                $excluded = command_oneline(@cmd);
3642                if ( $excluded ) {
3643                        my @new;
3644                        my $found;
3645                        for my $commit ( @commits ) {
3646                                if ( $commit eq $excluded ) {
3647                                        push @excluded, $commit;
3648                                        $found++;
3649                                        last;
3650                                }
3651                                else {
3652                                        push @new, $commit;
3653                                }
3654                        }
3655                        die "saw commit '$excluded' in rev-list output, "
3656                                ."but we didn't ask for that commit (wanted: @commits --not @$parents)"
3657                                        unless $found;
3658                        @commits = @new;
3659                }
3660        }
3661                while ($excluded and @commits);
3662
3663        return @excluded;
3664}
3665
3666
3667# note: this function should only be called if the various dirprops
3668# have actually changed
3669sub find_extra_svn_parents {
3670        my ($self, $ed, $mergeinfo, $parents) = @_;
3671        # aha!  svk:merge property changed...
3672
3673        memoize_svn_mergeinfo_functions();
3674
3675        # We first search for merged tips which are not in our
3676        # history.  Then, we figure out which git revisions are in
3677        # that tip, but not this revision.  If all of those revisions
3678        # are now marked as merge, we can add the tip as a parent.
3679        my @merges = split "\n", $mergeinfo;
3680        my @merge_tips;
3681        my $url = $self->{url};
3682        my $uuid = $self->ra_uuid;
3683        my %ranges;
3684        for my $merge ( @merges ) {
3685                my ($tip_commit, @ranges) =
3686                        lookup_svn_merge( $uuid, $url, $merge );
3687                unless (!$tip_commit or
3688                                grep { $_ eq $tip_commit } @$parents ) {
3689                        push @merge_tips, $tip_commit;
3690                        $ranges{$tip_commit} = \@ranges;
3691                } else {
3692                        push @merge_tips, undef;
3693                }
3694        }
3695
3696        my %excluded = map { $_ => 1 }
3697                parents_exclude($parents, grep { defined } @merge_tips);
3698
3699        # check merge tips for new parents
3700        my @new_parents;
3701        for my $merge_tip ( @merge_tips ) {
3702                my $spec = shift @merges;
3703                next unless $merge_tip and $excluded{$merge_tip};
3704
3705                my $ranges = $ranges{$merge_tip};
3706
3707                # check out 'new' tips
3708                my $merge_base;
3709                eval {
3710                        $merge_base = command_oneline(
3711                                "merge-base",
3712                                @$parents, $merge_tip,
3713                        );
3714                };
3715                if ($@) {
3716                        die "An error occurred during merge-base"
3717                                unless $@->isa("Git::Error::Command");
3718
3719                        warn "W: Cannot find common ancestor between ".
3720                             "@$parents and $merge_tip. Ignoring merge info.\n";
3721                        next;
3722                }
3723
3724                # double check that there are no missing non-merge commits
3725                my (@incomplete) = check_cherry_pick(
3726                        $merge_base, $merge_tip,
3727                        $parents,
3728                        @$ranges,
3729                       );
3730
3731                if ( @incomplete ) {
3732                        warn "W:svn cherry-pick ignored ($spec) - missing "
3733                                .@incomplete." commit(s) (eg $incomplete[0])\n";
3734                } else {
3735                        warn
3736                                "Found merge parent (svn:mergeinfo prop): ",
3737                                        $merge_tip, "\n";
3738                        push @new_parents, $merge_tip;
3739                }
3740        }
3741
3742        # cater for merges which merge commits from multiple branches
3743        if ( @new_parents > 1 ) {
3744                for ( my $i = 0; $i <= $#new_parents; $i++ ) {
3745                        for ( my $j = 0; $j <= $#new_parents; $j++ ) {
3746                                next if $i == $j;
3747                                next unless $new_parents[$i];
3748                                next unless $new_parents[$j];
3749                                my $revs = command_oneline(
3750                                        "rev-list", "-1",
3751                                        "$new_parents[$i]..$new_parents[$j]",
3752                                       );
3753                                if ( !$revs ) {
3754                                        undef($new_parents[$j]);
3755                                }
3756                        }
3757                }
3758        }
3759        push @$parents, grep { defined } @new_parents;
3760}
3761
3762sub make_log_entry {
3763        my ($self, $rev, $parents, $ed) = @_;
3764        my $untracked = $self->get_untracked($ed);
3765
3766        my @parents = @$parents;
3767        my $ps = $ed->{path_strip} || "";
3768        for my $path ( grep { m/$ps/ } %{$ed->{dir_prop}} ) {
3769                my $props = $ed->{dir_prop}{$path};
3770                if ( $props->{"svk:merge"} ) {
3771                        $self->find_extra_svk_parents
3772                                ($ed, $props->{"svk:merge"}, \@parents);
3773                }
3774                if ( $props->{"svn:mergeinfo"} ) {
3775                        $self->find_extra_svn_parents
3776                                ($ed,
3777                                 $props->{"svn:mergeinfo"},
3778                                 \@parents);
3779                }
3780        }
3781
3782        open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
3783        print $un "r$rev\n" or croak $!;
3784        print $un $_, "\n" foreach @$untracked;
3785        my %log_entry = ( parents => \@parents, revision => $rev,
3786                          log => '');
3787
3788        my $headrev;
3789        my $logged = delete $self->{logged_rev_props};
3790        if (!$logged || $self->{-want_revprops}) {
3791                my $rp = $self->ra->rev_proplist($rev);
3792                foreach (sort keys %$rp) {
3793                        my $v = $rp->{$_};
3794                        if (/^svn:(author|date|log)$/) {
3795                                $log_entry{$1} = $v;
3796                        } elsif ($_ eq 'svm:headrev') {
3797                                $headrev = $v;
3798                        } else {
3799                                print $un "  rev_prop: ", uri_encode($_), ' ',
3800                                          uri_encode($v), "\n";
3801                        }
3802                }
3803        } else {
3804                map { $log_entry{$_} = $logged->{$_} } keys %$logged;
3805        }
3806        close $un or croak $!;
3807
3808        $log_entry{date} = parse_svn_date($log_entry{date});
3809        $log_entry{log} .= "\n";
3810        my $author = $log_entry{author} = check_author($log_entry{author});
3811        my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
3812                                                       : ($author, undef);
3813
3814        my ($commit_name, $commit_email) = ($name, $email);
3815        if ($_use_log_author) {
3816                my $name_field;
3817                if ($log_entry{log} =~ /From:\s+(.*\S)\s*\n/i) {
3818                        $name_field = $1;
3819                } elsif ($log_entry{log} =~ /Signed-off-by:\s+(.*\S)\s*\n/i) {
3820                        $name_field = $1;
3821                }
3822                if (!defined $name_field) {
3823                        if (!defined $email) {
3824                                $email = $name;
3825                        }
3826                } elsif ($name_field =~ /(.*?)\s+<(.*)>/) {
3827                        ($name, $email) = ($1, $2);
3828                } elsif ($name_field =~ /(.*)@/) {
3829                        ($name, $email) = ($1, $name_field);
3830                } else {
3831                        ($name, $email) = ($name_field, $name_field);
3832                }
3833        }
3834        if (defined $headrev && $self->use_svm_props) {
3835                if ($self->rewrite_root) {
3836                        die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
3837                            "options set!\n";
3838                }
3839                if ($self->rewrite_uuid) {
3840                        die "Can't have both 'useSvmProps' and 'rewriteUUID' ",
3841                            "options set!\n";
3842                }
3843                my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$}i;
3844                # we don't want "SVM: initializing mirror for junk" ...
3845                return undef if $r == 0;
3846                my $svm = $self->svm;
3847                if ($uuid ne $svm->{uuid}) {
3848                        die "UUID mismatch on SVM path:\n",
3849                            "expected: $svm->{uuid}\n",
3850                            "     got: $uuid\n";
3851                }
3852                my $full_url = $self->full_url;
3853                $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
3854                             die "Failed to replace '$svm->{replace}' with ",
3855                                 "'$svm->{source}' in $full_url\n";
3856                # throw away username for storing in records
3857                remove_username($full_url);
3858                $log_entry{metadata} = "$full_url\@$r $uuid";
3859                $log_entry{svm_revision} = $r;
3860                $email ||= "$author\@$uuid";
3861                $commit_email ||= "$author\@$uuid";
3862        } elsif ($self->use_svnsync_props) {
3863                my $full_url = $self->svnsync->{url};
3864                $full_url .= "/$self->{path}" if length $self->{path};
3865                remove_username($full_url);
3866                my $uuid = $self->svnsync->{uuid};
3867                $log_entry{metadata} = "$full_url\@$rev $uuid";
3868                $email ||= "$author\@$uuid";
3869                $commit_email ||= "$author\@$uuid";
3870        } else {
3871                my $url = $self->metadata_url;
3872                remove_username($url);
3873                my $uuid = $self->rewrite_uuid || $self->ra->get_uuid;
3874                $log_entry{metadata} = "$url\@$rev " . $uuid;
3875                $email ||= "$author\@" . $uuid;
3876                $commit_email ||= "$author\@" . $uuid;
3877        }
3878        $log_entry{name} = $name;
3879        $log_entry{email} = $email;
3880        $log_entry{commit_name} = $commit_name;
3881        $log_entry{commit_email} = $commit_email;
3882        \%log_entry;
3883}
3884
3885sub fetch {
3886        my ($self, $min_rev, $max_rev, @parents) = @_;
3887        my ($last_rev, $last_commit) = $self->last_rev_commit;
3888        my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
3889        $self->ra->gs_fetch_loop_common($base, $head, [$self]);
3890}
3891
3892sub set_tree_cb {
3893        my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
3894        $self->{inject_parents} = { $rev => $tree };
3895        $self->fetch(undef, undef);
3896}
3897
3898sub set_tree {
3899        my ($self, $tree) = (shift, shift);
3900        my $log_entry = ::get_commit_entry($tree);
3901        unless ($self->{last_rev}) {
3902                ::fatal("Must have an existing revision to commit");
3903        }
3904        my %ed_opts = ( r => $self->{last_rev},
3905                        log => $log_entry->{log},
3906                        ra => $self->ra,
3907                        tree_a => $self->{last_commit},
3908                        tree_b => $tree,
3909                        editor_cb => sub {
3910                               $self->set_tree_cb($log_entry, $tree, @_) },
3911                        svn_path => $self->{path} );
3912        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
3913                print "No changes\nr$self->{last_rev} = $tree\n";
3914        }
3915}
3916
3917sub rebuild_from_rev_db {
3918        my ($self, $path) = @_;
3919        my $r = -1;
3920        open my $fh, '<', $path or croak "open: $!";
3921        binmode $fh or croak "binmode: $!";
3922        while (<$fh>) {
3923                length($_) == 41 or croak "inconsistent size in ($_) != 41";
3924                chomp($_);
3925                ++$r;
3926                next if $_ eq ('0' x 40);
3927                $self->rev_map_set($r, $_);
3928                print "r$r = $_\n";
3929        }
3930        close $fh or croak "close: $!";
3931        unlink $path or croak "unlink: $!";
3932}
3933
3934sub rebuild {
3935        my ($self) = @_;
3936        my $map_path = $self->map_path;
3937        my $partial = (-e $map_path && ! -z $map_path);
3938        return unless ::verify_ref($self->refname.'^0');
3939        if (!$partial && ($self->use_svm_props || $self->no_metadata)) {
3940                my $rev_db = $self->rev_db_path;
3941                $self->rebuild_from_rev_db($rev_db);
3942                if ($self->use_svm_props) {
3943                        my $svm_rev_db = $self->rev_db_path($self->svm_uuid);
3944                        $self->rebuild_from_rev_db($svm_rev_db);
3945                }
3946                $self->unlink_rev_db_symlink;
3947                return;
3948        }
3949        print "Rebuilding $map_path ...\n" if (!$partial);
3950        my ($base_rev, $head) = ($partial ? $self->rev_map_max_norebuild(1) :
3951                (undef, undef));
3952        my ($log, $ctx) =
3953            command_output_pipe(qw/rev-list --pretty=raw --reverse/,
3954                                ($head ? "$head.." : "") . $self->refname,
3955                                '--');
3956        my $metadata_url = $self->metadata_url;
3957        remove_username($metadata_url);
3958        my $svn_uuid = $self->rewrite_uuid || $self->ra_uuid;
3959        my $c;
3960        while (<$log>) {
3961                if ( m{^commit ($::sha1)$} ) {
3962                        $c = $1;
3963                        next;
3964                }
3965                next unless s{^\s*(git-svn-id:)}{$1};
3966                my ($url, $rev, $uuid) = ::extract_metadata($_);
3967                remove_username($url);
3968
3969                # ignore merges (from set-tree)
3970                next if (!defined $rev || !$uuid);
3971
3972                # if we merged or otherwise started elsewhere, this is
3973                # how we break out of it
3974                if (($uuid ne $svn_uuid) ||
3975                    ($metadata_url && $url && ($url ne $metadata_url))) {
3976                        next;
3977                }
3978                if ($partial && $head) {
3979                        print "Partial-rebuilding $map_path ...\n";
3980                        print "Currently at $base_rev = $head\n";
3981                        $head = undef;
3982                }
3983
3984                $self->rev_map_set($rev, $c);
3985                print "r$rev = $c\n";
3986        }
3987        command_close_pipe($log, $ctx);
3988        print "Done rebuilding $map_path\n" if (!$partial || !$head);
3989        my $rev_db_path = $self->rev_db_path;
3990        if (-f $self->rev_db_path) {
3991                unlink $self->rev_db_path or croak "unlink: $!";
3992        }
3993        $self->unlink_rev_db_symlink;
3994}
3995
3996# rev_map:
3997# Tie::File seems to be prone to offset errors if revisions get sparse,
3998# it's not that fast, either.  Tie::File is also not in Perl 5.6.  So
3999# one of my favorite modules is out :<  Next up would be one of the DBM
4000# modules, but I'm not sure which is most portable...
4001#
4002# This is the replacement for the rev_db format, which was too big
4003# and inefficient for large repositories with a lot of sparse history
4004# (mainly tags)
4005#
4006# The format is this:
4007#   - 24 bytes for every record,
4008#     * 4 bytes for the integer representing an SVN revision number
4009#     * 20 bytes representing the sha1 of a git commit
4010#   - No empty padding records like the old format
4011#     (except the last record, which can be overwritten)
4012#   - new records are written append-only since SVN revision numbers
4013#     increase monotonically
4014#   - lookups on SVN revision number are done via a binary search
4015#   - Piping the file to xxd -c24 is a good way of dumping it for
4016#     viewing or editing (piped back through xxd -r), should the need
4017#     ever arise.
4018#   - The last record can be padding revision with an all-zero sha1
4019#     This is used to optimize fetch performance when using multiple
4020#     "fetch" directives in .git/config
4021#
4022# These files are disposable unless noMetadata or useSvmProps is set
4023
4024sub _rev_map_set {
4025        my ($fh, $rev, $commit) = @_;
4026
4027        binmode $fh or croak "binmode: $!";
4028        my $size = (stat($fh))[7];
4029        ($size % 24) == 0 or croak "inconsistent size: $size";
4030
4031        my $wr_offset = 0;
4032        if ($size > 0) {
4033                sysseek($fh, -24, SEEK_END) or croak "seek: $!";
4034                my $read = sysread($fh, my $buf, 24) or croak "read: $!";
4035                $read == 24 or croak "read only $read bytes (!= 24)";
4036                my ($last_rev, $last_commit) = unpack(rev_map_fmt, $buf);
4037                if ($last_commit eq ('0' x40)) {
4038                        if ($size >= 48) {
4039                                sysseek($fh, -48, SEEK_END) or croak "seek: $!";
4040                                $read = sysread($fh, $buf, 24) or
4041                                    croak "read: $!";
4042                                $read == 24 or
4043                                    croak "read only $read bytes (!= 24)";
4044                                ($last_rev, $last_commit) =
4045                                    unpack(rev_map_fmt, $buf);
4046                                if ($last_commit eq ('0' x40)) {
4047                                        croak "inconsistent .rev_map\n";
4048                                }
4049                        }
4050                        if ($last_rev >= $rev) {
4051                                croak "last_rev is higher!: $last_rev >= $rev";
4052                        }
4053                        $wr_offset = -24;
4054                }
4055        }
4056        sysseek($fh, $wr_offset, SEEK_END) or croak "seek: $!";
4057        syswrite($fh, pack(rev_map_fmt, $rev, $commit), 24) == 24 or
4058          croak "write: $!";
4059}
4060
4061sub _rev_map_reset {
4062        my ($fh, $rev, $commit) = @_;
4063        my $c = _rev_map_get($fh, $rev);
4064        $c eq $commit or die "_rev_map_reset(@_) commit $c does not match!\n";
4065        my $offset = sysseek($fh, 0, SEEK_CUR) or croak "seek: $!";
4066        truncate $fh, $offset or croak "truncate: $!";
4067}
4068
4069sub mkfile {
4070        my ($path) = @_;
4071        unless (-e $path) {
4072                my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
4073                mkpath([$dir]) unless -d $dir;
4074                open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
4075                close $fh or die "Couldn't close (create) $path: $!\n";
4076        }
4077}
4078
4079sub rev_map_set {
4080        my ($self, $rev, $commit, $update_ref, $uuid) = @_;
4081        defined $commit or die "missing arg3\n";
4082        length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
4083        my $db = $self->map_path($uuid);
4084        my $db_lock = "$db.lock";
4085        my $sigmask;
4086        $update_ref ||= 0;
4087        if ($update_ref) {
4088                $sigmask = POSIX::SigSet->new();
4089                my $signew = POSIX::SigSet->new(SIGINT, SIGHUP, SIGTERM,
4090                        SIGALRM, SIGUSR1, SIGUSR2);
4091                sigprocmask(SIG_BLOCK, $signew, $sigmask) or
4092                        croak "Can't block signals: $!";
4093        }
4094        mkfile($db);
4095
4096        $LOCKFILES{$db_lock} = 1;
4097        my $sync;
4098        # both of these options make our .rev_db file very, very important
4099        # and we can't afford to lose it because rebuild() won't work
4100        if ($self->use_svm_props || $self->no_metadata) {
4101                $sync = 1;
4102                copy($db, $db_lock) or die "rev_map_set(@_): ",
4103                                           "Failed to copy: ",
4104                                           "$db => $db_lock ($!)\n";
4105        } else {
4106                rename $db, $db_lock or die "rev_map_set(@_): ",
4107                                            "Failed to rename: ",
4108                                            "$db => $db_lock ($!)\n";
4109        }
4110
4111        sysopen(my $fh, $db_lock, O_RDWR | O_CREAT)
4112             or croak "Couldn't open $db_lock: $!\n";
4113        $update_ref eq 'reset' ? _rev_map_reset($fh, $rev, $commit) :
4114                                 _rev_map_set($fh, $rev, $commit);
4115        if ($sync) {
4116                $fh->flush or die "Couldn't flush $db_lock: $!\n";
4117                $fh->sync or die "Couldn't sync $db_lock: $!\n";
4118        }
4119        close $fh or croak $!;
4120        if ($update_ref) {
4121                $_head = $self;
4122                my $note = "";
4123                $note = " ($update_ref)" if ($update_ref !~ /^\d*$/);
4124                command_noisy('update-ref', '-m', "r$rev$note",
4125                              $self->refname, $commit);
4126        }
4127        rename $db_lock, $db or die "rev_map_set(@_): ", "Failed to rename: ",
4128                                    "$db_lock => $db ($!)\n";
4129        delete $LOCKFILES{$db_lock};
4130        if ($update_ref) {
4131                sigprocmask(SIG_SETMASK, $sigmask) or
4132                        croak "Can't restore signal mask: $!";
4133        }
4134}
4135
4136# If want_commit, this will return an array of (rev, commit) where
4137# commit _must_ be a valid commit in the archive.
4138# Otherwise, it'll return the max revision (whether or not the
4139# commit is valid or just a 0x40 placeholder).
4140sub rev_map_max {
4141        my ($self, $want_commit) = @_;
4142        $self->rebuild;
4143        my ($r, $c) = $self->rev_map_max_norebuild($want_commit);
4144        $want_commit ? ($r, $c) : $r;
4145}
4146
4147sub rev_map_max_norebuild {
4148        my ($self, $want_commit) = @_;
4149        my $map_path = $self->map_path;
4150        stat $map_path or return $want_commit ? (0, undef) : 0;
4151        sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
4152        binmode $fh or croak "binmode: $!";
4153        my $size = (stat($fh))[7];
4154        ($size % 24) == 0 or croak "inconsistent size: $size";
4155
4156        if ($size == 0) {
4157                close $fh or croak "close: $!";
4158                return $want_commit ? (0, undef) : 0;
4159        }
4160
4161        sysseek($fh, -24, SEEK_END) or croak "seek: $!";
4162        sysread($fh, my $buf, 24) == 24 or croak "read: $!";
4163        my ($r, $c) = unpack(rev_map_fmt, $buf);
4164        if ($want_commit && $c eq ('0' x40)) {
4165                if ($size < 48) {
4166                        return $want_commit ? (0, undef) : 0;
4167                }
4168                sysseek($fh, -48, SEEK_END) or croak "seek: $!";
4169                sysread($fh, $buf, 24) == 24 or croak "read: $!";
4170                ($r, $c) = unpack(rev_map_fmt, $buf);
4171                if ($c eq ('0'x40)) {
4172                        croak "Penultimate record is all-zeroes in $map_path";
4173                }
4174        }
4175        close $fh or croak "close: $!";
4176        $want_commit ? ($r, $c) : $r;
4177}
4178
4179sub rev_map_get {
4180        my ($self, $rev, $uuid) = @_;
4181        my $map_path = $self->map_path($uuid);
4182        return undef unless -e $map_path;
4183
4184        sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
4185        my $c = _rev_map_get($fh, $rev);
4186        close($fh) or croak "close: $!";
4187        $c
4188}
4189
4190sub _rev_map_get {
4191        my ($fh, $rev) = @_;
4192
4193        binmode $fh or croak "binmode: $!";
4194        my $size = (stat($fh))[7];
4195        ($size % 24) == 0 or croak "inconsistent size: $size";
4196
4197        if ($size == 0) {
4198                return undef;
4199        }
4200
4201        my ($l, $u) = (0, $size - 24);
4202        my ($r, $c, $buf);
4203
4204        while ($l <= $u) {
4205                my $i = int(($l/24 + $u/24) / 2) * 24;
4206                sysseek($fh, $i, SEEK_SET) or croak "seek: $!";
4207                sysread($fh, my $buf, 24) == 24 or croak "read: $!";
4208                my ($r, $c) = unpack(rev_map_fmt, $buf);
4209
4210                if ($r < $rev) {
4211                        $l = $i + 24;
4212                } elsif ($r > $rev) {
4213                        $u = $i - 24;
4214                } else { # $r == $rev
4215                        return $c eq ('0' x 40) ? undef : $c;
4216                }
4217        }
4218        undef;
4219}
4220
4221# Finds the first svn revision that exists on (if $eq_ok is true) or
4222# before $rev for the current branch.  It will not search any lower
4223# than $min_rev.  Returns the git commit hash and svn revision number
4224# if found, else (undef, undef).
4225sub find_rev_before {
4226        my ($self, $rev, $eq_ok, $min_rev) = @_;
4227        --$rev unless $eq_ok;
4228        $min_rev ||= 1;
4229        my $max_rev = $self->rev_map_max;
4230        $rev = $max_rev if ($rev > $max_rev);
4231        while ($rev >= $min_rev) {
4232                if (my $c = $self->rev_map_get($rev)) {
4233                        return ($rev, $c);
4234                }
4235                --$rev;
4236        }
4237        return (undef, undef);
4238}
4239
4240# Finds the first svn revision that exists on (if $eq_ok is true) or
4241# after $rev for the current branch.  It will not search any higher
4242# than $max_rev.  Returns the git commit hash and svn revision number
4243# if found, else (undef, undef).
4244sub find_rev_after {
4245        my ($self, $rev, $eq_ok, $max_rev) = @_;
4246        ++$rev unless $eq_ok;
4247        $max_rev ||= $self->rev_map_max;
4248        while ($rev <= $max_rev) {
4249                if (my $c = $self->rev_map_get($rev)) {
4250                        return ($rev, $c);
4251                }
4252                ++$rev;
4253        }
4254        return (undef, undef);
4255}
4256
4257sub _new {
4258        my ($class, $repo_id, $ref_id, $path) = @_;
4259        unless (defined $repo_id && length $repo_id) {
4260                $repo_id = $Git::SVN::default_repo_id;
4261        }
4262        unless (defined $ref_id && length $ref_id) {
4263                $_prefix = '' unless defined($_prefix);
4264                $_[2] = $ref_id =
4265                             "refs/remotes/$_prefix$Git::SVN::default_ref_id";
4266        }
4267        $_[1] = $repo_id;
4268        my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
4269
4270        # Older repos imported by us used $GIT_DIR/svn/foo instead of
4271        # $GIT_DIR/svn/refs/remotes/foo when tracking refs/remotes/foo
4272        if ($ref_id =~ m{^refs/remotes/(.*)}) {
4273                my $old_dir = "$ENV{GIT_DIR}/svn/$1";
4274                if (-d $old_dir && ! -d $dir) {
4275                        $dir = $old_dir;
4276                }
4277        }
4278
4279        $_[3] = $path = '' unless (defined $path);
4280        mkpath([$dir]);
4281        bless {
4282                ref_id => $ref_id, dir => $dir, index => "$dir/index",
4283                path => $path, config => "$ENV{GIT_DIR}/svn/config",
4284                map_root => "$dir/.rev_map", repo_id => $repo_id }, $class;
4285}
4286
4287# for read-only access of old .rev_db formats
4288sub unlink_rev_db_symlink {
4289        my ($self) = @_;
4290        my $link = $self->rev_db_path;
4291        $link =~ s/\.[\w-]+$// or croak "missing UUID at the end of $link";
4292        if (-l $link) {
4293                unlink $link or croak "unlink: $link failed!";
4294        }
4295}
4296
4297sub rev_db_path {
4298        my ($self, $uuid) = @_;
4299        my $db_path = $self->map_path($uuid);
4300        $db_path =~ s{/\.rev_map\.}{/\.rev_db\.}
4301            or croak "map_path: $db_path does not contain '/.rev_map.' !";
4302        $db_path;
4303}
4304
4305# the new replacement for .rev_db
4306sub map_path {
4307        my ($self, $uuid) = @_;
4308        $uuid ||= $self->ra_uuid;
4309        "$self->{map_root}.$uuid";
4310}
4311
4312sub uri_encode {
4313        my ($f) = @_;
4314        $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
4315        $f
4316}
4317
4318sub uri_decode {
4319        my ($f) = @_;
4320        $f =~ s#%([0-9a-fA-F]{2})#chr(hex($1))#eg;
4321        $f
4322}
4323
4324sub remove_username {
4325        $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
4326}
4327
4328package Git::SVN::Prompt;
4329use strict;
4330use warnings;
4331require SVN::Core;
4332use vars qw/$_no_auth_cache $_username/;
4333
4334sub simple {
4335        my ($cred, $realm, $default_username, $may_save, $pool) = @_;
4336        $may_save = undef if $_no_auth_cache;
4337        $default_username = $_username if defined $_username;
4338        if (defined $default_username && length $default_username) {
4339                if (defined $realm && length $realm) {
4340                        print STDERR "Authentication realm: $realm\n";
4341                        STDERR->flush;
4342                }
4343                $cred->username($default_username);
4344        } else {
4345                username($cred, $realm, $may_save, $pool);
4346        }
4347        $cred->password(_read_password("Password for '" .
4348                                       $cred->username . "': ", $realm));
4349        $cred->may_save($may_save);
4350        $SVN::_Core::SVN_NO_ERROR;
4351}
4352
4353sub ssl_server_trust {
4354        my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
4355        $may_save = undef if $_no_auth_cache;
4356        print STDERR "Error validating server certificate for '$realm':\n";
4357        {
4358                no warnings 'once';
4359                # All variables SVN::Auth::SSL::* are used only once,
4360                # so we're shutting up Perl warnings about this.
4361                if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
4362                        print STDERR " - The certificate is not issued ",
4363                            "by a trusted authority. Use the\n",
4364                            "   fingerprint to validate ",
4365                            "the certificate manually!\n";
4366                }
4367                if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
4368                        print STDERR " - The certificate hostname ",
4369                            "does not match.\n";
4370                }
4371                if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
4372                        print STDERR " - The certificate is not yet valid.\n";
4373                }
4374                if ($failures & $SVN::Auth::SSL::EXPIRED) {
4375                        print STDERR " - The certificate has expired.\n";
4376                }
4377                if ($failures & $SVN::Auth::SSL::OTHER) {
4378                        print STDERR " - The certificate has ",
4379                            "an unknown error.\n";
4380                }
4381        } # no warnings 'once'
4382        printf STDERR
4383                "Certificate information:\n".
4384                " - Hostname: %s\n".
4385                " - Valid: from %s until %s\n".
4386                " - Issuer: %s\n".
4387                " - Fingerprint: %s\n",
4388                map $cert_info->$_, qw(hostname valid_from valid_until
4389                                       issuer_dname fingerprint);
4390        my $choice;
4391prompt:
4392        print STDERR $may_save ?
4393              "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
4394              "(R)eject or accept (t)emporarily? ";
4395        STDERR->flush;
4396        $choice = lc(substr(<STDIN> || 'R', 0, 1));
4397        if ($choice =~ /^t$/i) {
4398                $cred->may_save(undef);
4399        } elsif ($choice =~ /^r$/i) {
4400                return -1;
4401        } elsif ($may_save && $choice =~ /^p$/i) {
4402                $cred->may_save($may_save);
4403        } else {
4404                goto prompt;
4405        }
4406        $cred->accepted_failures($failures);
4407        $SVN::_Core::SVN_NO_ERROR;
4408}
4409
4410sub ssl_client_cert {
4411        my ($cred, $realm, $may_save, $pool) = @_;
4412        $may_save = undef if $_no_auth_cache;
4413        print STDERR "Client certificate filename: ";
4414        STDERR->flush;
4415        chomp(my $filename = <STDIN>);
4416        $cred->cert_file($filename);
4417        $cred->may_save($may_save);
4418        $SVN::_Core::SVN_NO_ERROR;
4419}
4420
4421sub ssl_client_cert_pw {
4422        my ($cred, $realm, $may_save, $pool) = @_;
4423        $may_save = undef if $_no_auth_cache;
4424        $cred->password(_read_password("Password: ", $realm));
4425        $cred->may_save($may_save);
4426        $SVN::_Core::SVN_NO_ERROR;
4427}
4428
4429sub username {
4430        my ($cred, $realm, $may_save, $pool) = @_;
4431        $may_save = undef if $_no_auth_cache;
4432        if (defined $realm && length $realm) {
4433                print STDERR "Authentication realm: $realm\n";
4434        }
4435        my $username;
4436        if (defined $_username) {
4437                $username = $_username;
4438        } else {
4439                print STDERR "Username: ";
4440                STDERR->flush;
4441                chomp($username = <STDIN>);
4442        }
4443        $cred->username($username);
4444        $cred->may_save($may_save);
4445        $SVN::_Core::SVN_NO_ERROR;
4446}
4447
4448sub _read_password {
4449        my ($prompt, $realm) = @_;
4450        my $password = '';
4451        if (exists $ENV{GIT_ASKPASS}) {
4452                open(PH, "-|", $ENV{GIT_ASKPASS}, $prompt);
4453                $password = <PH>;
4454                $password =~ s/[\012\015]//; # \n\r
4455                close(PH);
4456        } else {
4457                print STDERR $prompt;
4458                STDERR->flush;
4459                require Term::ReadKey;
4460                Term::ReadKey::ReadMode('noecho');
4461                while (defined(my $key = Term::ReadKey::ReadKey(0))) {
4462                        last if $key =~ /[\012\015]/; # \n\r
4463                        $password .= $key;
4464                }
4465                Term::ReadKey::ReadMode('restore');
4466                print STDERR "\n";
4467                STDERR->flush;
4468        }
4469        $password;
4470}
4471
4472package SVN::Git::Fetcher;
4473use vars qw/@ISA $_ignore_regex $_preserve_empty_dirs $_placeholder_filename
4474            @deleted_gpath %added_placeholder $repo_id/;
4475use strict;
4476use warnings;
4477use Carp qw/croak/;
4478use File::Basename qw/dirname/;
4479use IO::File qw//;
4480
4481# file baton members: path, mode_a, mode_b, pool, fh, blob, base
4482sub new {
4483        my ($class, $git_svn, $switch_path) = @_;
4484        my $self = SVN::Delta::Editor->new;
4485        bless $self, $class;
4486        if (exists $git_svn->{last_commit}) {
4487                $self->{c} = $git_svn->{last_commit};
4488                $self->{empty_symlinks} =
4489                                  _mark_empty_symlinks($git_svn, $switch_path);
4490        }
4491
4492        # some options are read globally, but can be overridden locally
4493        # per [svn-remote "..."] section.  Command-line options will *NOT*
4494        # override options set in an [svn-remote "..."] section
4495        $repo_id = $git_svn->{repo_id};
4496        my $k = "svn-remote.$repo_id.ignore-paths";
4497        my $v = eval { command_oneline('config', '--get', $k) };
4498        $self->{ignore_regex} = $v;
4499
4500        $k = "svn-remote.$repo_id.preserve-empty-dirs";
4501        $v = eval { command_oneline('config', '--get', '--bool', $k) };
4502        if ($v && $v eq 'true') {
4503                $_preserve_empty_dirs = 1;
4504                $k = "svn-remote.$repo_id.placeholder-filename";
4505                $v = eval { command_oneline('config', '--get', $k) };
4506                $_placeholder_filename = $v;
4507        }
4508
4509        # Load the list of placeholder files added during previous invocations.
4510        $k = "svn-remote.$repo_id.added-placeholder";
4511        $v = eval { command_oneline('config', '--get-all', $k) };
4512        if ($_preserve_empty_dirs && $v) {
4513                # command() prints errors to stderr, so we only call it if
4514                # command_oneline() succeeded.
4515                my @v = command('config', '--get-all', $k);
4516                $added_placeholder{ dirname($_) } = $_ foreach @v;
4517        }
4518
4519        $self->{empty} = {};
4520        $self->{dir_prop} = {};
4521        $self->{file_prop} = {};
4522        $self->{absent_dir} = {};
4523        $self->{absent_file} = {};
4524        $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
4525        $self->{pathnameencoding} = Git::config('svn.pathnameencoding');
4526        $self;
4527}
4528
4529# this uses the Ra object, so it must be called before do_{switch,update},
4530# not inside them (when the Git::SVN::Fetcher object is passed) to
4531# do_{switch,update}
4532sub _mark_empty_symlinks {
4533        my ($git_svn, $switch_path) = @_;
4534        my $bool = Git::config_bool('svn.brokenSymlinkWorkaround');
4535        return {} if (!defined($bool)) || (defined($bool) && ! $bool);
4536
4537        my %ret;
4538        my ($rev, $cmt) = $git_svn->last_rev_commit;
4539        return {} unless ($rev && $cmt);
4540
4541        # allow the warning to be printed for each revision we fetch to
4542        # ensure the user sees it.  The user can also disable the workaround
4543        # on the repository even while git svn is running and the next
4544        # revision fetched will skip this expensive function.
4545        my $printed_warning;
4546        chomp(my $empty_blob = `git hash-object -t blob --stdin < /dev/null`);
4547        my ($ls, $ctx) = command_output_pipe(qw/ls-tree -r -z/, $cmt);
4548        local $/ = "\0";
4549        my $pfx = defined($switch_path) ? $switch_path : $git_svn->{path};
4550        $pfx .= '/' if length($pfx);
4551        while (<$ls>) {
4552                chomp;
4553                s/\A100644 blob $empty_blob\t//o or next;
4554                unless ($printed_warning) {
4555                        print STDERR "Scanning for empty symlinks, ",
4556                                     "this may take a while if you have ",
4557                                     "many empty files\n",
4558                                     "You may disable this with `",
4559                                     "git config svn.brokenSymlinkWorkaround ",
4560                                     "false'.\n",
4561                                     "This may be done in a different ",
4562                                     "terminal without restarting ",
4563                                     "git svn\n";
4564                        $printed_warning = 1;
4565                }
4566                my $path = $_;
4567                my (undef, $props) =
4568                               $git_svn->ra->get_file($pfx.$path, $rev, undef);
4569                if ($props->{'svn:special'}) {
4570                        $ret{$path} = 1;
4571                }
4572        }
4573        command_close_pipe($ls, $ctx);
4574        \%ret;
4575}
4576
4577# returns true if a given path is inside a ".git" directory
4578sub in_dot_git {
4579        $_[0] =~ m{(?:^|/)\.git(?:/|$)};
4580}
4581
4582# return value: 0 -- don't ignore, 1 -- ignore
4583sub is_path_ignored {
4584        my ($self, $path) = @_;
4585        return 1 if in_dot_git($path);
4586        return 1 if defined($self->{ignore_regex}) &&
4587                    $path =~ m!$self->{ignore_regex}!;
4588        return 0 unless defined($_ignore_regex);
4589        return 1 if $path =~ m!$_ignore_regex!o;
4590        return 0;
4591}
4592
4593sub set_path_strip {
4594        my ($self, $path) = @_;
4595        $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
4596}
4597
4598sub open_root {
4599        { path => '' };
4600}
4601
4602sub open_directory {
4603        my ($self, $path, $pb, $rev) = @_;
4604        { path => $path };
4605}
4606
4607sub git_path {
4608        my ($self, $path) = @_;
4609        if (my $enc = $self->{pathnameencoding}) {
4610                require Encode;
4611                Encode::from_to($path, 'UTF-8', $enc);
4612        }
4613        if ($self->{path_strip}) {
4614                $path =~ s!$self->{path_strip}!! or
4615                  die "Failed to strip path '$path' ($self->{path_strip})\n";
4616        }
4617        $path;
4618}
4619
4620sub delete_entry {
4621        my ($self, $path, $rev, $pb) = @_;
4622        return undef if $self->is_path_ignored($path);
4623
4624        my $gpath = $self->git_path($path);
4625        return undef if ($gpath eq '');
4626
4627        # remove entire directories.
4628        my ($tree) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
4629                         =~ /\A040000 tree ([a-f\d]{40})\t\Q$gpath\E\0/);
4630        if ($tree) {
4631                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
4632                                                     -r --name-only -z/,
4633                                                     $tree);
4634                local $/ = "\0";
4635                while (<$ls>) {
4636                        chomp;
4637                        my $rmpath = "$gpath/$_";
4638                        $self->{gii}->remove($rmpath);
4639                        print "\tD\t$rmpath\n" unless $::_q;
4640                }
4641                print "\tD\t$gpath/\n" unless $::_q;
4642                command_close_pipe($ls, $ctx);
4643        } else {
4644                $self->{gii}->remove($gpath);
4645                print "\tD\t$gpath\n" unless $::_q;
4646        }
4647        # Don't add to @deleted_gpath if we're deleting a placeholder file.
4648        push @deleted_gpath, $gpath unless $added_placeholder{dirname($path)};
4649        $self->{empty}->{$path} = 0;
4650        undef;
4651}
4652
4653sub open_file {
4654        my ($self, $path, $pb, $rev) = @_;
4655        my ($mode, $blob);
4656
4657        goto out if $self->is_path_ignored($path);
4658
4659        my $gpath = $self->git_path($path);
4660        ($mode, $blob) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
4661                             =~ /\A(\d{6}) blob ([a-f\d]{40})\t\Q$gpath\E\0/);
4662        unless (defined $mode && defined $blob) {
4663                die "$path was not found in commit $self->{c} (r$rev)\n";
4664        }
4665        if ($mode eq '100644' && $self->{empty_symlinks}->{$path}) {
4666                $mode = '120000';
4667        }
4668out:
4669        { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
4670          pool => SVN::Pool->new, action => 'M' };
4671}
4672
4673sub add_file {
4674        my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
4675        my $mode;
4676
4677        if (!$self->is_path_ignored($path)) {
4678                my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
4679                delete $self->{empty}->{$dir};
4680                $mode = '100644';
4681
4682                if ($added_placeholder{$dir}) {
4683                        # Remove our placeholder file, if we created one.
4684                        delete_entry($self, $added_placeholder{$dir})
4685                                unless $path eq $added_placeholder{$dir};
4686                        delete $added_placeholder{$dir}
4687                }
4688        }
4689
4690        { path => $path, mode_a => $mode, mode_b => $mode,
4691          pool => SVN::Pool->new, action => 'A' };
4692}
4693
4694sub add_directory {
4695        my ($self, $path, $cp_path, $cp_rev) = @_;
4696        goto out if $self->is_path_ignored($path);
4697        my $gpath = $self->git_path($path);
4698        if ($gpath eq '') {
4699                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
4700                                                     -r --name-only -z/,
4701                                                     $self->{c});
4702                local $/ = "\0";
4703                while (<$ls>) {
4704                        chomp;
4705                        $self->{gii}->remove($_);
4706                        print "\tD\t$_\n" unless $::_q;
4707                        push @deleted_gpath, $gpath;
4708                }
4709                command_close_pipe($ls, $ctx);
4710                $self->{empty}->{$path} = 0;
4711        }
4712        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
4713        delete $self->{empty}->{$dir};
4714        $self->{empty}->{$path} = 1;
4715
4716        if ($added_placeholder{$dir}) {
4717                # Remove our placeholder file, if we created one.
4718                delete_entry($self, $added_placeholder{$dir});
4719                delete $added_placeholder{$dir}
4720        }
4721
4722out:
4723        { path => $path };
4724}
4725
4726sub change_dir_prop {
4727        my ($self, $db, $prop, $value) = @_;
4728        return undef if $self->is_path_ignored($db->{path});
4729        $self->{dir_prop}->{$db->{path}} ||= {};
4730        $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
4731        undef;
4732}
4733
4734sub absent_directory {
4735        my ($self, $path, $pb) = @_;
4736        return undef if $self->is_path_ignored($path);
4737        $self->{absent_dir}->{$pb->{path}} ||= [];
4738        push @{$self->{absent_dir}->{$pb->{path}}}, $path;
4739        undef;
4740}
4741
4742sub absent_file {
4743        my ($self, $path, $pb) = @_;
4744        return undef if $self->is_path_ignored($path);
4745        $self->{absent_file}->{$pb->{path}} ||= [];
4746        push @{$self->{absent_file}->{$pb->{path}}}, $path;
4747        undef;
4748}
4749
4750sub change_file_prop {
4751        my ($self, $fb, $prop, $value) = @_;
4752        return undef if $self->is_path_ignored($fb->{path});
4753        if ($prop eq 'svn:executable') {
4754                if ($fb->{mode_b} != 120000) {
4755                        $fb->{mode_b} = defined $value ? 100755 : 100644;
4756                }
4757        } elsif ($prop eq 'svn:special') {
4758                $fb->{mode_b} = defined $value ? 120000 : 100644;
4759        } else {
4760                $self->{file_prop}->{$fb->{path}} ||= {};
4761                $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
4762        }
4763        undef;
4764}
4765
4766sub apply_textdelta {
4767        my ($self, $fb, $exp) = @_;
4768        return undef if $self->is_path_ignored($fb->{path});
4769        my $fh = $::_repository->temp_acquire('svn_delta');
4770        # $fh gets auto-closed() by SVN::TxDelta::apply(),
4771        # (but $base does not,) so dup() it for reading in close_file
4772        open my $dup, '<&', $fh or croak $!;
4773        my $base = $::_repository->temp_acquire('git_blob');
4774
4775        if ($fb->{blob}) {
4776                my ($base_is_link, $size);
4777
4778                if ($fb->{mode_a} eq '120000' &&
4779                    ! $self->{empty_symlinks}->{$fb->{path}}) {
4780                        print $base 'link ' or die "print $!\n";
4781                        $base_is_link = 1;
4782                }
4783        retry:
4784                $size = $::_repository->cat_blob($fb->{blob}, $base);
4785                die "Failed to read object $fb->{blob}" if ($size < 0);
4786
4787                if (defined $exp) {
4788                        seek $base, 0, 0 or croak $!;
4789                        my $got = ::md5sum($base);
4790                        if ($got ne $exp) {
4791                                my $err = "Checksum mismatch: ".
4792                                       "$fb->{path} $fb->{blob}\n" .
4793                                       "expected: $exp\n" .
4794                                       "     got: $got\n";
4795                                if ($base_is_link) {
4796                                        warn $err,
4797                                             "Retrying... (possibly ",
4798                                             "a bad symlink from SVN)\n";
4799                                        $::_repository->temp_reset($base);
4800                                        $base_is_link = 0;
4801                                        goto retry;
4802                                }
4803                                die $err;
4804                        }
4805                }
4806        }
4807        seek $base, 0, 0 or croak $!;
4808        $fb->{fh} = $fh;
4809        $fb->{base} = $base;
4810        [ SVN::TxDelta::apply($base, $dup, undef, $fb->{path}, $fb->{pool}) ];
4811}
4812
4813sub close_file {
4814        my ($self, $fb, $exp) = @_;
4815        return undef if $self->is_path_ignored($fb->{path});
4816
4817        my $hash;
4818        my $path = $self->git_path($fb->{path});
4819        if (my $fh = $fb->{fh}) {
4820                if (defined $exp) {
4821                        seek($fh, 0, 0) or croak $!;
4822                        my $got = ::md5sum($fh);
4823                        if ($got ne $exp) {
4824                                die "Checksum mismatch: $path\n",
4825                                    "expected: $exp\n    got: $got\n";
4826                        }
4827                }
4828                if ($fb->{mode_b} == 120000) {
4829                        sysseek($fh, 0, 0) or croak $!;
4830                        my $rd = sysread($fh, my $buf, 5);
4831
4832                        if (!defined $rd) {
4833                                croak "sysread: $!\n";
4834                        } elsif ($rd == 0) {
4835                                warn "$path has mode 120000",
4836                                     " but it points to nothing\n",
4837                                     "converting to an empty file with mode",
4838                                     " 100644\n";
4839                                $fb->{mode_b} = '100644';
4840                        } elsif ($buf ne 'link ') {
4841                                warn "$path has mode 120000",
4842                                     " but is not a link\n";
4843                        } else {
4844                                my $tmp_fh = $::_repository->temp_acquire(
4845                                        'svn_hash');
4846                                my $res;
4847                                while ($res = sysread($fh, my $str, 1024)) {
4848                                        my $out = syswrite($tmp_fh, $str, $res);
4849                                        defined($out) && $out == $res
4850                                                or croak("write ",
4851                                                        Git::temp_path($tmp_fh),
4852                                                        ": $!\n");
4853                                }
4854                                defined $res or croak $!;
4855
4856                                ($fh, $tmp_fh) = ($tmp_fh, $fh);
4857                                Git::temp_release($tmp_fh, 1);
4858                        }
4859                }
4860
4861                $hash = $::_repository->hash_and_insert_object(
4862                                Git::temp_path($fh));
4863                $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
4864
4865                Git::temp_release($fb->{base}, 1);
4866                Git::temp_release($fh, 1);
4867        } else {
4868                $hash = $fb->{blob} or die "no blob information\n";
4869        }
4870        $fb->{pool}->clear;
4871        $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
4872        print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
4873        undef;
4874}
4875
4876sub abort_edit {
4877        my $self = shift;
4878        $self->{nr} = $self->{gii}->{nr};
4879        delete $self->{gii};
4880        $self->SUPER::abort_edit(@_);
4881}
4882
4883sub close_edit {
4884        my $self = shift;
4885
4886        if ($_preserve_empty_dirs) {
4887                my @empty_dirs;
4888
4889                # Any entry flagged as empty that also has an associated
4890                # dir_prop represents a newly created empty directory.
4891                foreach my $i (keys %{$self->{empty}}) {
4892                        push @empty_dirs, $i if exists $self->{dir_prop}->{$i};
4893                }
4894
4895                # Search for directories that have become empty due subsequent
4896                # file deletes.
4897                push @empty_dirs, $self->find_empty_directories();
4898
4899                # Finally, add a placeholder file to each empty directory.
4900                $self->add_placeholder_file($_) foreach (@empty_dirs);
4901
4902                $self->stash_placeholder_list();
4903        }
4904
4905        $self->{git_commit_ok} = 1;
4906        $self->{nr} = $self->{gii}->{nr};
4907        delete $self->{gii};
4908        $self->SUPER::close_edit(@_);
4909}
4910
4911sub find_empty_directories {
4912        my ($self) = @_;
4913        my @empty_dirs;
4914        my %dirs = map { dirname($_) => 1 } @deleted_gpath;
4915
4916        foreach my $dir (sort keys %dirs) {
4917                next if $dir eq ".";
4918
4919                # If there have been any additions to this directory, there is
4920                # no reason to check if it is empty.
4921                my $skip_added = 0;
4922                foreach my $t (qw/dir_prop file_prop/) {
4923                        foreach my $path (keys %{ $self->{$t} }) {
4924                                if (exists $self->{$t}->{dirname($path)}) {
4925                                        $skip_added = 1;
4926                                        last;
4927                                }
4928                        }
4929                        last if $skip_added;
4930                }
4931                next if $skip_added;
4932
4933                # Use `git ls-tree` to get the filenames of this directory
4934                # that existed prior to this particular commit.
4935                my $ls = command('ls-tree', '-z', '--name-only',
4936                                 $self->{c}, "$dir/");
4937                my %files = map { $_ => 1 } split(/\0/, $ls);
4938
4939                # Remove the filenames that were deleted during this commit.
4940                delete $files{$_} foreach (@deleted_gpath);
4941
4942                # Report the directory if there are no filenames left.
4943                push @empty_dirs, $dir unless (scalar %files);
4944        }
4945        @empty_dirs;
4946}
4947
4948sub add_placeholder_file {
4949        my ($self, $dir) = @_;
4950        my $path = "$dir/$_placeholder_filename";
4951        my $gpath = $self->git_path($path);
4952
4953        my $fh = $::_repository->temp_acquire($gpath);
4954        my $hash = $::_repository->hash_and_insert_object(Git::temp_path($fh));
4955        Git::temp_release($fh, 1);
4956        $self->{gii}->update('100644', $hash, $gpath) or croak $!;
4957
4958        # The directory should no longer be considered empty.
4959        delete $self->{empty}->{$dir} if exists $self->{empty}->{$dir};
4960
4961        # Keep track of any placeholder files we create.
4962        $added_placeholder{$dir} = $path;
4963}
4964
4965sub stash_placeholder_list {
4966        my ($self) = @_;
4967        my $k = "svn-remote.$repo_id.added-placeholder";
4968        my $v = eval { command_oneline('config', '--get-all', $k) };
4969        command_noisy('config', '--unset-all', $k) if $v;
4970        foreach (values %added_placeholder) {
4971                command_noisy('config', '--add', $k, $_);
4972        }
4973}
4974
4975package SVN::Git::Editor;
4976use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
4977use strict;
4978use warnings;
4979use Carp qw/croak/;
4980use IO::File;
4981
4982sub new {
4983        my ($class, $opts) = @_;
4984        foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
4985                die "$_ required!\n" unless (defined $opts->{$_});
4986        }
4987
4988        my $pool = SVN::Pool->new;
4989        my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
4990        my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
4991                                     $opts->{r}, $mods);
4992
4993        # $opts->{ra} functions should not be used after this:
4994        my @ce  = $opts->{ra}->get_commit_editor($opts->{log},
4995                                                $opts->{editor_cb}, $pool);
4996        my $self = SVN::Delta::Editor->new(@ce, $pool);
4997        bless $self, $class;
4998        foreach (qw/svn_path r tree_a tree_b/) {
4999                $self->{$_} = $opts->{$_};
5000        }
5001        $self->{url} = $opts->{ra}->{url};
5002        $self->{mods} = $mods;
5003        $self->{types} = $types;
5004        $self->{pool} = $pool;
5005        $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
5006        $self->{rm} = { };
5007        $self->{path_prefix} = length $self->{svn_path} ?
5008                               "$self->{svn_path}/" : '';
5009        $self->{config} = $opts->{config};
5010        $self->{mergeinfo} = $opts->{mergeinfo};
5011        return $self;
5012}
5013
5014sub generate_diff {
5015        my ($tree_a, $tree_b) = @_;
5016        my @diff_tree = qw(diff-tree -z -r);
5017        if ($_cp_similarity) {
5018                push @diff_tree, "-C$_cp_similarity";
5019        } else {
5020                push @diff_tree, '-C';
5021        }
5022        push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
5023        push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
5024        push @diff_tree, $tree_a, $tree_b;
5025        my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
5026        local $/ = "\0";
5027        my $state = 'meta';
5028        my @mods;
5029        while (<$diff_fh>) {
5030                chomp $_; # this gets rid of the trailing "\0"
5031                if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
5032                                        ($::sha1)\s($::sha1)\s
5033                                        ([MTCRAD])\d*$/xo) {
5034                        push @mods, {   mode_a => $1, mode_b => $2,
5035                                        sha1_a => $3, sha1_b => $4,
5036                                        chg => $5 };
5037                        if ($5 =~ /^(?:C|R)$/) {
5038                                $state = 'file_a';
5039                        } else {
5040                                $state = 'file_b';
5041                        }
5042                } elsif ($state eq 'file_a') {
5043                        my $x = $mods[$#mods] or croak "Empty array\n";
5044                        if ($x->{chg} !~ /^(?:C|R)$/) {
5045                                croak "Error parsing $_, $x->{chg}\n";
5046                        }
5047                        $x->{file_a} = $_;
5048                        $state = 'file_b';
5049                } elsif ($state eq 'file_b') {
5050                        my $x = $mods[$#mods] or croak "Empty array\n";
5051                        if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
5052                                croak "Error parsing $_, $x->{chg}\n";
5053                        }
5054                        if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
5055                                croak "Error parsing $_, $x->{chg}\n";
5056                        }
5057                        $x->{file_b} = $_;
5058                        $state = 'meta';
5059                } else {
5060                        croak "Error parsing $_\n";
5061                }
5062        }
5063        command_close_pipe($diff_fh, $ctx);
5064        \@mods;
5065}
5066
5067sub check_diff_paths {
5068        my ($ra, $pfx, $rev, $mods) = @_;
5069        my %types;
5070        $pfx .= '/' if length $pfx;
5071
5072        sub type_diff_paths {
5073                my ($ra, $types, $path, $rev) = @_;
5074                my @p = split m#/+#, $path;
5075                my $c = shift @p;
5076                unless (defined $types->{$c}) {
5077                        $types->{$c} = $ra->check_path($c, $rev);
5078                }
5079                while (@p) {
5080                        $c .= '/' . shift @p;
5081                        next if defined $types->{$c};
5082                        $types->{$c} = $ra->check_path($c, $rev);
5083                }
5084        }
5085
5086        foreach my $m (@$mods) {
5087                foreach my $f (qw/file_a file_b/) {
5088                        next unless defined $m->{$f};
5089                        my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
5090                        if (length $pfx.$dir && ! defined $types{$dir}) {
5091                                type_diff_paths($ra, \%types, $pfx.$dir, $rev);
5092                        }
5093                }
5094        }
5095        \%types;
5096}
5097
5098sub split_path {
5099        return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
5100}
5101
5102sub repo_path {
5103        my ($self, $path) = @_;
5104        if (my $enc = $self->{pathnameencoding}) {
5105                require Encode;
5106                Encode::from_to($path, $enc, 'UTF-8');
5107        }
5108        $self->{path_prefix}.(defined $path ? $path : '');
5109}
5110
5111sub url_path {
5112        my ($self, $path) = @_;
5113        if ($self->{url} =~ m#^https?://#) {
5114                $path =~ s!([^~a-zA-Z0-9_./-])!uc sprintf("%%%02x",ord($1))!eg;
5115        }
5116        $self->{url} . '/' . $self->repo_path($path);
5117}
5118
5119sub rmdirs {
5120        my ($self) = @_;
5121        my $rm = $self->{rm};
5122        delete $rm->{''}; # we never delete the url we're tracking
5123        return unless %$rm;
5124
5125        foreach (keys %$rm) {
5126                my @d = split m#/#, $_;
5127                my $c = shift @d;
5128                $rm->{$c} = 1;
5129                while (@d) {
5130                        $c .= '/' . shift @d;
5131                        $rm->{$c} = 1;
5132                }
5133        }
5134        delete $rm->{$self->{svn_path}};
5135        delete $rm->{''}; # we never delete the url we're tracking
5136        return unless %$rm;
5137
5138        my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
5139                                             $self->{tree_b});
5140        local $/ = "\0";
5141        while (<$fh>) {
5142                chomp;
5143                my @dn = split m#/#, $_;
5144                while (pop @dn) {
5145                        delete $rm->{join '/', @dn};
5146                }
5147                unless (%$rm) {
5148                        close $fh;
5149                        return;
5150                }
5151        }
5152        command_close_pipe($fh, $ctx);
5153
5154        my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
5155        foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
5156                $self->close_directory($bat->{$d}, $p);
5157                my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
5158                print "\tD+\t$d/\n" unless $::_q;
5159                $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
5160                delete $bat->{$d};
5161        }
5162}
5163
5164sub open_or_add_dir {
5165        my ($self, $full_path, $baton, $deletions) = @_;
5166        my $t = $self->{types}->{$full_path};
5167        if (!defined $t) {
5168                die "$full_path not known in r$self->{r} or we have a bug!\n";
5169        }
5170        {
5171                no warnings 'once';
5172                # SVN::Node::none and SVN::Node::file are used only once,
5173                # so we're shutting up Perl's warnings about them.
5174                if ($t == $SVN::Node::none || defined($deletions->{$full_path})) {
5175                        return $self->add_directory($full_path, $baton,
5176                            undef, -1, $self->{pool});
5177                } elsif ($t == $SVN::Node::dir) {
5178                        return $self->open_directory($full_path, $baton,
5179                            $self->{r}, $self->{pool});
5180                } # no warnings 'once'
5181                print STDERR "$full_path already exists in repository at ",
5182                    "r$self->{r} and it is not a directory (",
5183                    ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
5184        } # no warnings 'once'
5185        exit 1;
5186}
5187
5188sub ensure_path {
5189        my ($self, $path, $deletions) = @_;
5190        my $bat = $self->{bat};
5191        my $repo_path = $self->repo_path($path);
5192        return $bat->{''} unless (length $repo_path);
5193
5194        my @p = split m#/+#, $repo_path;
5195        my $c = shift @p;
5196        $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''}, $deletions);
5197        while (@p) {
5198                my $c0 = $c;
5199                $c .= '/' . shift @p;
5200                $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0}, $deletions);
5201        }
5202        return $bat->{$c};
5203}
5204
5205# Subroutine to convert a globbing pattern to a regular expression.
5206# From perl cookbook.
5207sub glob2pat {
5208        my $globstr = shift;
5209        my %patmap = ('*' => '.*', '?' => '.', '[' => '[', ']' => ']');
5210        $globstr =~ s{(.)} { $patmap{$1} || "\Q$1" }ge;
5211        return '^' . $globstr . '$';
5212}
5213
5214sub check_autoprop {
5215        my ($self, $pattern, $properties, $file, $fbat) = @_;
5216        # Convert the globbing pattern to a regular expression.
5217        my $regex = glob2pat($pattern);
5218        # Check if the pattern matches the file name.
5219        if($file =~ m/($regex)/) {
5220                # Parse the list of properties to set.
5221                my @props = split(/;/, $properties);
5222                foreach my $prop (@props) {
5223                        # Parse 'name=value' syntax and set the property.
5224                        if ($prop =~ /([^=]+)=(.*)/) {
5225                                my ($n,$v) = ($1,$2);
5226                                for ($n, $v) {
5227                                        s/^\s+//; s/\s+$//;
5228                                }
5229                                $self->change_file_prop($fbat, $n, $v);
5230                        }
5231                }
5232        }
5233}
5234
5235sub apply_autoprops {
5236        my ($self, $file, $fbat) = @_;
5237        my $conf_t = ${$self->{config}}{'config'};
5238        no warnings 'once';
5239        # Check [miscellany]/enable-auto-props in svn configuration.
5240        if (SVN::_Core::svn_config_get_bool(
5241                $conf_t,
5242                $SVN::_Core::SVN_CONFIG_SECTION_MISCELLANY,
5243                $SVN::_Core::SVN_CONFIG_OPTION_ENABLE_AUTO_PROPS,
5244                0)) {
5245                # Auto-props are enabled.  Enumerate them to look for matches.
5246                my $callback = sub {
5247                        $self->check_autoprop($_[0], $_[1], $file, $fbat);
5248                };
5249                SVN::_Core::svn_config_enumerate(
5250                        $conf_t,
5251                        $SVN::_Core::SVN_CONFIG_SECTION_AUTO_PROPS,
5252                        $callback);
5253        }
5254}
5255
5256sub A {
5257        my ($self, $m, $deletions) = @_;
5258        my ($dir, $file) = split_path($m->{file_b});
5259        my $pbat = $self->ensure_path($dir, $deletions);
5260        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
5261                                        undef, -1);
5262        print "\tA\t$m->{file_b}\n" unless $::_q;
5263        $self->apply_autoprops($file, $fbat);
5264        $self->chg_file($fbat, $m);
5265        $self->close_file($fbat,undef,$self->{pool});
5266}
5267
5268sub C {
5269        my ($self, $m, $deletions) = @_;
5270        my ($dir, $file) = split_path($m->{file_b});
5271        my $pbat = $self->ensure_path($dir, $deletions);
5272        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
5273                                $self->url_path($m->{file_a}), $self->{r});
5274        print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
5275        $self->chg_file($fbat, $m);
5276        $self->close_file($fbat,undef,$self->{pool});
5277}
5278
5279sub delete_entry {
5280        my ($self, $path, $pbat) = @_;
5281        my $rpath = $self->repo_path($path);
5282        my ($dir, $file) = split_path($rpath);
5283        $self->{rm}->{$dir} = 1;
5284        $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
5285}
5286
5287sub R {
5288        my ($self, $m, $deletions) = @_;
5289        my ($dir, $file) = split_path($m->{file_b});
5290        my $pbat = $self->ensure_path($dir, $deletions);
5291        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
5292                                $self->url_path($m->{file_a}), $self->{r});
5293        print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
5294        $self->apply_autoprops($file, $fbat);
5295        $self->chg_file($fbat, $m);
5296        $self->close_file($fbat,undef,$self->{pool});
5297
5298        ($dir, $file) = split_path($m->{file_a});
5299        $pbat = $self->ensure_path($dir, $deletions);
5300        $self->delete_entry($m->{file_a}, $pbat);
5301}
5302
5303sub M {
5304        my ($self, $m, $deletions) = @_;
5305        my ($dir, $file) = split_path($m->{file_b});
5306        my $pbat = $self->ensure_path($dir, $deletions);
5307        my $fbat = $self->open_file($self->repo_path($m->{file_b}),
5308                                $pbat,$self->{r},$self->{pool});
5309        print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
5310        $self->chg_file($fbat, $m);
5311        $self->close_file($fbat,undef,$self->{pool});
5312}
5313
5314sub T { shift->M(@_) }
5315
5316sub change_file_prop {
5317        my ($self, $fbat, $pname, $pval) = @_;
5318        $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
5319}
5320
5321sub change_dir_prop {
5322        my ($self, $pbat, $pname, $pval) = @_;
5323        $self->SUPER::change_dir_prop($pbat, $pname, $pval, $self->{pool});
5324}
5325
5326sub _chg_file_get_blob ($$$$) {
5327        my ($self, $fbat, $m, $which) = @_;
5328        my $fh = $::_repository->temp_acquire("git_blob_$which");
5329        if ($m->{"mode_$which"} =~ /^120/) {
5330                print $fh 'link ' or croak $!;
5331                $self->change_file_prop($fbat,'svn:special','*');
5332        } elsif ($m->{mode_a} =~ /^120/ && $m->{"mode_$which"} !~ /^120/) {
5333                $self->change_file_prop($fbat,'svn:special',undef);
5334        }
5335        my $blob = $m->{"sha1_$which"};
5336        return ($fh,) if ($blob =~ /^0{40}$/);
5337        my $size = $::_repository->cat_blob($blob, $fh);
5338        croak "Failed to read object $blob" if ($size < 0);
5339        $fh->flush == 0 or croak $!;
5340        seek $fh, 0, 0 or croak $!;
5341
5342        my $exp = ::md5sum($fh);
5343        seek $fh, 0, 0 or croak $!;
5344        return ($fh, $exp);
5345}
5346
5347sub chg_file {
5348        my ($self, $fbat, $m) = @_;
5349        if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
5350                $self->change_file_prop($fbat,'svn:executable','*');
5351        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
5352                $self->change_file_prop($fbat,'svn:executable',undef);
5353        }
5354        my ($fh_a, $exp_a) = _chg_file_get_blob $self, $fbat, $m, 'a';
5355        my ($fh_b, $exp_b) = _chg_file_get_blob $self, $fbat, $m, 'b';
5356        my $pool = SVN::Pool->new;
5357        my $atd = $self->apply_textdelta($fbat, $exp_a, $pool);
5358        if (-s $fh_a) {
5359                my $txstream = SVN::TxDelta::new ($fh_a, $fh_b, $pool);
5360                my $res = SVN::TxDelta::send_txstream($txstream, @$atd, $pool);
5361                if (defined $res) {
5362                        die "Unexpected result from send_txstream: $res\n",
5363                            "(SVN::Core::VERSION: $SVN::Core::VERSION)\n";
5364                }
5365        } else {
5366                my $got = SVN::TxDelta::send_stream($fh_b, @$atd, $pool);
5367                die "Checksum mismatch\nexpected: $exp_b\ngot: $got\n"
5368                    if ($got ne $exp_b);
5369        }
5370        Git::temp_release($fh_b, 1);
5371        Git::temp_release($fh_a, 1);
5372        $pool->clear;
5373}
5374
5375sub D {
5376        my ($self, $m, $deletions) = @_;
5377        my ($dir, $file) = split_path($m->{file_b});
5378        my $pbat = $self->ensure_path($dir, $deletions);
5379        print "\tD\t$m->{file_b}\n" unless $::_q;
5380        $self->delete_entry($m->{file_b}, $pbat);
5381}
5382
5383sub close_edit {
5384        my ($self) = @_;
5385        my ($p,$bat) = ($self->{pool}, $self->{bat});
5386        foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
5387                next if $_ eq '';
5388                $self->close_directory($bat->{$_}, $p);
5389        }
5390        $self->close_directory($bat->{''}, $p);
5391        $self->SUPER::close_edit($p);
5392        $p->clear;
5393}
5394
5395sub abort_edit {
5396        my ($self) = @_;
5397        $self->SUPER::abort_edit($self->{pool});
5398}
5399
5400sub DESTROY {
5401        my $self = shift;
5402        $self->SUPER::DESTROY(@_);
5403        $self->{pool}->clear;
5404}
5405
5406# this drives the editor
5407sub apply_diff {
5408        my ($self) = @_;
5409        my $mods = $self->{mods};
5410        my %o = ( D => 0, C => 1, R => 2, A => 3, M => 4, T => 5 );
5411        my %deletions;
5412
5413        foreach my $m (@$mods) {
5414                if ($m->{chg} eq "D") {
5415                        $deletions{$m->{file_b}} = 1;
5416                }
5417        }
5418
5419        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
5420                my $f = $m->{chg};
5421                if (defined $o{$f}) {
5422                        $self->$f($m, \%deletions);
5423                } else {
5424                        fatal("Invalid change type: $f");
5425                }
5426        }
5427
5428        if (defined($self->{mergeinfo})) {
5429                $self->change_dir_prop($self->{bat}{''}, "svn:mergeinfo",
5430                                       $self->{mergeinfo});
5431        }
5432        $self->rmdirs if $_rmdir;
5433        if (@$mods == 0 && !defined($self->{mergeinfo})) {
5434                $self->abort_edit;
5435        } else {
5436                $self->close_edit;
5437        }
5438        return scalar @$mods;
5439}
5440
5441package Git::SVN::Ra;
5442use vars qw/@ISA $config_dir $_ignore_refs_regex $_log_window_size/;
5443use strict;
5444use warnings;
5445my ($ra_invalid, $can_do_switch, %ignored_err, $RA);
5446
5447BEGIN {
5448        # enforce temporary pool usage for some simple functions
5449        no strict 'refs';
5450        for my $f (qw/rev_proplist get_latest_revnum get_uuid get_repos_root
5451                      get_file/) {
5452                my $SUPER = "SUPER::$f";
5453                *$f = sub {
5454                        my $self = shift;
5455                        my $pool = SVN::Pool->new;
5456                        my @ret = $self->$SUPER(@_,$pool);
5457                        $pool->clear;
5458                        wantarray ? @ret : $ret[0];
5459                };
5460        }
5461}
5462
5463sub _auth_providers () {
5464        my @rv = (
5465          SVN::Client::get_simple_provider(),
5466          SVN::Client::get_ssl_server_trust_file_provider(),
5467          SVN::Client::get_simple_prompt_provider(
5468            \&Git::SVN::Prompt::simple, 2),
5469          SVN::Client::get_ssl_client_cert_file_provider(),
5470          SVN::Client::get_ssl_client_cert_prompt_provider(
5471            \&Git::SVN::Prompt::ssl_client_cert, 2),
5472          SVN::Client::get_ssl_client_cert_pw_file_provider(),
5473          SVN::Client::get_ssl_client_cert_pw_prompt_provider(
5474            \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
5475          SVN::Client::get_username_provider(),
5476          SVN::Client::get_ssl_server_trust_prompt_provider(
5477            \&Git::SVN::Prompt::ssl_server_trust),
5478          SVN::Client::get_username_prompt_provider(
5479            \&Git::SVN::Prompt::username, 2)
5480        );
5481
5482        # earlier 1.6.x versions would segfault, and <= 1.5.x didn't have
5483        # this function
5484        if (::compare_svn_version('1.6.12') > 0) {
5485                my $config = SVN::Core::config_get_config($config_dir);
5486                my ($p, @a);
5487                # config_get_config returns all config files from
5488                # ~/.subversion, auth_get_platform_specific_client_providers
5489                # just wants the config "file".
5490                @a = ($config->{'config'}, undef);
5491                $p = SVN::Core::auth_get_platform_specific_client_providers(@a);
5492                # Insert the return value from
5493                # auth_get_platform_specific_providers
5494                unshift @rv, @$p;
5495        }
5496        \@rv;
5497}
5498
5499sub escape_uri_only {
5500        my ($uri) = @_;
5501        my @tmp;
5502        foreach (split m{/}, $uri) {
5503                s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
5504                push @tmp, $_;
5505        }
5506        join('/', @tmp);
5507}
5508
5509sub escape_url {
5510        my ($url) = @_;
5511        if ($url =~ m#^(https?)://([^/]+)(.*)$#) {
5512                my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
5513                $url = "$scheme://$domain$uri";
5514        }
5515        $url;
5516}
5517
5518sub new {
5519        my ($class, $url) = @_;
5520        $url =~ s!/+$!!;
5521        return $RA if ($RA && $RA->{url} eq $url);
5522
5523        ::_req_svn();
5524
5525        SVN::_Core::svn_config_ensure($config_dir, undef);
5526        my ($baton, $callbacks) = SVN::Core::auth_open_helper(_auth_providers);
5527        my $config = SVN::Core::config_get_config($config_dir);
5528        $RA = undef;
5529        my $dont_store_passwords = 1;
5530        my $conf_t = ${$config}{'config'};
5531        {
5532                no warnings 'once';
5533                # The usage of $SVN::_Core::SVN_CONFIG_* variables
5534                # produces warnings that variables are used only once.
5535                # I had not found the better way to shut them up, so
5536                # the warnings of type 'once' are disabled in this block.
5537                if (SVN::_Core::svn_config_get_bool($conf_t,
5538                    $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
5539                    $SVN::_Core::SVN_CONFIG_OPTION_STORE_PASSWORDS,
5540                    1) == 0) {
5541                        SVN::_Core::svn_auth_set_parameter($baton,
5542                            $SVN::_Core::SVN_AUTH_PARAM_DONT_STORE_PASSWORDS,
5543                            bless (\$dont_store_passwords, "_p_void"));
5544                }
5545                if (SVN::_Core::svn_config_get_bool($conf_t,
5546                    $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
5547                    $SVN::_Core::SVN_CONFIG_OPTION_STORE_AUTH_CREDS,
5548                    1) == 0) {
5549                        $Git::SVN::Prompt::_no_auth_cache = 1;
5550                }
5551        } # no warnings 'once'
5552        my $self = SVN::Ra->new(url => escape_url($url), auth => $baton,
5553                              config => $config,
5554                              pool => SVN::Pool->new,
5555                              auth_provider_callbacks => $callbacks);
5556        $self->{url} = $url;
5557        $self->{svn_path} = $url;
5558        $self->{repos_root} = $self->get_repos_root;
5559        $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
5560        $self->{cache} = { check_path => { r => 0, data => {} },
5561                           get_dir => { r => 0, data => {} } };
5562        $RA = bless $self, $class;
5563}
5564
5565sub check_path {
5566        my ($self, $path, $r) = @_;
5567        my $cache = $self->{cache}->{check_path};
5568        if ($r == $cache->{r} && exists $cache->{data}->{$path}) {
5569                return $cache->{data}->{$path};
5570        }
5571        my $pool = SVN::Pool->new;
5572        my $t = $self->SUPER::check_path($path, $r, $pool);
5573        $pool->clear;
5574        if ($r != $cache->{r}) {
5575                %{$cache->{data}} = ();
5576                $cache->{r} = $r;
5577        }
5578        $cache->{data}->{$path} = $t;
5579}
5580
5581sub get_dir {
5582        my ($self, $dir, $r) = @_;
5583        my $cache = $self->{cache}->{get_dir};
5584        if ($r == $cache->{r}) {
5585                if (my $x = $cache->{data}->{$dir}) {
5586                        return wantarray ? @$x : $x->[0];
5587                }
5588        }
5589        my $pool = SVN::Pool->new;
5590        my ($d, undef, $props) = $self->SUPER::get_dir($dir, $r, $pool);
5591        my %dirents = map { $_ => { kind => $d->{$_}->kind } } keys %$d;
5592        $pool->clear;
5593        if ($r != $cache->{r}) {
5594                %{$cache->{data}} = ();
5595                $cache->{r} = $r;
5596        }
5597        $cache->{data}->{$dir} = [ \%dirents, $r, $props ];
5598        wantarray ? (\%dirents, $r, $props) : \%dirents;
5599}
5600
5601sub DESTROY {
5602        # do not call the real DESTROY since we store ourselves in $RA
5603}
5604
5605# get_log(paths, start, end, limit,
5606#         discover_changed_paths, strict_node_history, receiver)
5607sub get_log {
5608        my ($self, @args) = @_;
5609        my $pool = SVN::Pool->new;
5610
5611        # svn_log_changed_path_t objects passed to get_log are likely to be
5612        # overwritten even if only the refs are copied to an external variable,
5613        # so we should dup the structures in their entirety.  Using an
5614        # externally passed pool (instead of our temporary and quickly cleared
5615        # pool in Git::SVN::Ra) does not help matters at all...
5616        my $receiver = pop @args;
5617        my $prefix = "/".$self->{svn_path};
5618        $prefix =~ s#/+($)##;
5619        my $prefix_regex = qr#^\Q$prefix\E#;
5620        push(@args, sub {
5621                my ($paths) = $_[0];
5622                return &$receiver(@_) unless $paths;
5623                $_[0] = ();
5624                foreach my $p (keys %$paths) {
5625                        my $i = $paths->{$p};
5626                        # Make path relative to our url, not repos_root
5627                        $p =~ s/$prefix_regex//;
5628                        my %s = map { $_ => $i->$_; }
5629                                qw/copyfrom_path copyfrom_rev action/;
5630                        if ($s{'copyfrom_path'}) {
5631                                $s{'copyfrom_path'} =~ s/$prefix_regex//;
5632                        }
5633                        $_[0]{$p} = \%s;
5634                }
5635                &$receiver(@_);
5636        });
5637
5638
5639        # the limit parameter was not supported in SVN 1.1.x, so we
5640        # drop it.  Therefore, the receiver callback passed to it
5641        # is made aware of this limitation by being wrapped if
5642        # the limit passed to is being wrapped.
5643        if (::compare_svn_version('1.2.0') <= 0) {
5644                my $limit = splice(@args, 3, 1);
5645                if ($limit > 0) {
5646                        my $receiver = pop @args;
5647                        push(@args, sub { &$receiver(@_) if (--$limit >= 0) });
5648                }
5649        }
5650        my $ret = $self->SUPER::get_log(@args, $pool);
5651        $pool->clear;
5652        $ret;
5653}
5654
5655sub trees_match {
5656        my ($self, $url1, $rev1, $url2, $rev2) = @_;
5657        my $ctx = SVN::Client->new(auth => _auth_providers);
5658        my $out = IO::File->new_tmpfile;
5659
5660        # older SVN (1.1.x) doesn't take $pool as the last parameter for
5661        # $ctx->diff(), so we'll create a default one
5662        my $pool = SVN::Pool->new_default_sub;
5663
5664        $ra_invalid = 1; # this will open a new SVN::Ra connection to $url1
5665        $ctx->diff([], $url1, $rev1, $url2, $rev2, 1, 1, 0, $out, $out);
5666        $out->flush;
5667        my $ret = (($out->stat)[7] == 0);
5668        close $out or croak $!;
5669
5670        $ret;
5671}
5672
5673sub get_commit_editor {
5674        my ($self, $log, $cb, $pool) = @_;
5675
5676        my @lock = (::compare_svn_version('1.2.0') >= 0) ? (undef, 0) : ();
5677        $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
5678}
5679
5680sub gs_do_update {
5681        my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
5682        my $new = ($rev_a == $rev_b);
5683        my $path = $gs->{path};
5684
5685        if ($new && -e $gs->{index}) {
5686                unlink $gs->{index} or die
5687                  "Couldn't unlink index: $gs->{index}: $!\n";
5688        }
5689        my $pool = SVN::Pool->new;
5690        $editor->set_path_strip($path);
5691        my (@pc) = split m#/#, $path;
5692        my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
5693                                        1, $editor, $pool);
5694        my @lock = (::compare_svn_version('1.2.0') >= 0) ? (undef) : ();
5695
5696        # Since we can't rely on svn_ra_reparent being available, we'll
5697        # just have to do some magic with set_path to make it so
5698        # we only want a partial path.
5699        my $sp = '';
5700        my $final = join('/', @pc);
5701        while (@pc) {
5702                $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
5703                $sp .= '/' if length $sp;
5704                $sp .= shift @pc;
5705        }
5706        die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
5707
5708        $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
5709
5710        $reporter->finish_report($pool);
5711        $pool->clear;
5712        $editor->{git_commit_ok};
5713}
5714
5715# this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
5716# svn_ra_reparent didn't work before 1.4)
5717sub gs_do_switch {
5718        my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
5719        my $path = $gs->{path};
5720        my $pool = SVN::Pool->new;
5721
5722        my $full_url = $self->{url};
5723        my $old_url = $full_url;
5724        $full_url .= '/' . $path if length $path;
5725        my ($ra, $reparented);
5726
5727        if ($old_url =~ m#^svn(\+ssh)?://# ||
5728            ($full_url =~ m#^https?://# &&
5729             escape_url($full_url) ne $full_url)) {
5730                $_[0] = undef;
5731                $self = undef;
5732                $RA = undef;
5733                $ra = Git::SVN::Ra->new($full_url);
5734                $ra_invalid = 1;
5735        } elsif ($old_url ne $full_url) {
5736                SVN::_Ra::svn_ra_reparent($self->{session}, $full_url, $pool);
5737                $self->{url} = $full_url;
5738                $reparented = 1;
5739        }
5740
5741        $ra ||= $self;
5742        $url_b = escape_url($url_b);
5743        my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
5744        my @lock = (::compare_svn_version('1.2.0') >= 0) ? (undef) : ();
5745        $reporter->set_path('', $rev_a, 0, @lock, $pool);
5746        $reporter->finish_report($pool);
5747
5748        if ($reparented) {
5749                SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
5750                $self->{url} = $old_url;
5751        }
5752
5753        $pool->clear;
5754        $editor->{git_commit_ok};
5755}
5756
5757sub longest_common_path {
5758        my ($gsv, $globs) = @_;
5759        my %common;
5760        my $common_max = scalar @$gsv;
5761
5762        foreach my $gs (@$gsv) {
5763                my @tmp = split m#/#, $gs->{path};
5764                my $p = '';
5765                foreach (@tmp) {
5766                        $p .= length($p) ? "/$_" : $_;
5767                        $common{$p} ||= 0;
5768                        $common{$p}++;
5769                }
5770        }
5771        $globs ||= [];
5772        $common_max += scalar @$globs;
5773        foreach my $glob (@$globs) {
5774                my @tmp = split m#/#, $glob->{path}->{left};
5775                my $p = '';
5776                foreach (@tmp) {
5777                        $p .= length($p) ? "/$_" : $_;
5778                        $common{$p} ||= 0;
5779                        $common{$p}++;
5780                }
5781        }
5782
5783        my $longest_path = '';
5784        foreach (sort {length $b <=> length $a} keys %common) {
5785                if ($common{$_} == $common_max) {
5786                        $longest_path = $_;
5787                        last;
5788                }
5789        }
5790        $longest_path;
5791}
5792
5793sub gs_fetch_loop_common {
5794        my ($self, $base, $head, $gsv, $globs) = @_;
5795        return if ($base > $head);
5796        my $inc = $_log_window_size;
5797        my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
5798        my $longest_path = longest_common_path($gsv, $globs);
5799        my $ra_url = $self->{url};
5800        my $find_trailing_edge;
5801        while (1) {
5802                my %revs;
5803                my $err;
5804                my $err_handler = $SVN::Error::handler;
5805                $SVN::Error::handler = sub {
5806                        ($err) = @_;
5807                        skip_unknown_revs($err);
5808                };
5809                sub _cb {
5810                        my ($paths, $r, $author, $date, $log) = @_;
5811                        [ $paths,
5812                          { author => $author, date => $date, log => $log } ];
5813                }
5814                $self->get_log([$longest_path], $min, $max, 0, 1, 1,
5815                               sub { $revs{$_[1]} = _cb(@_) });
5816                if ($err) {
5817                        print "Checked through r$max\r";
5818                } else {
5819                        $find_trailing_edge = 1;
5820                }
5821                if ($err and $find_trailing_edge) {
5822                        print STDERR "Path '$longest_path' ",
5823                                     "was probably deleted:\n",
5824                                     $err->expanded_message,
5825                                     "\nWill attempt to follow ",
5826                                     "revisions r$min .. r$max ",
5827                                     "committed before the deletion\n";
5828                        my $hi = $max;
5829                        while (--$hi >= $min) {
5830                                my $ok;
5831                                $self->get_log([$longest_path], $min, $hi,
5832                                               0, 1, 1, sub {
5833                                               $ok = $_[1];
5834                                               $revs{$_[1]} = _cb(@_) });
5835                                if ($ok) {
5836                                        print STDERR "r$min .. r$ok OK\n";
5837                                        last;
5838                                }
5839                        }
5840                        $find_trailing_edge = 0;
5841                }
5842                $SVN::Error::handler = $err_handler;
5843
5844                my %exists = map { $_->{path} => $_ } @$gsv;
5845                foreach my $r (sort {$a <=> $b} keys %revs) {
5846                        my ($paths, $logged) = @{$revs{$r}};
5847
5848                        foreach my $gs ($self->match_globs(\%exists, $paths,
5849                                                           $globs, $r)) {
5850                                if ($gs->rev_map_max >= $r) {
5851                                        next;
5852                                }
5853                                next unless $gs->match_paths($paths, $r);
5854                                $gs->{logged_rev_props} = $logged;
5855                                if (my $last_commit = $gs->last_commit) {
5856                                        $gs->assert_index_clean($last_commit);
5857                                }
5858                                my $log_entry = $gs->do_fetch($paths, $r);
5859                                if ($log_entry) {
5860                                        $gs->do_git_commit($log_entry);
5861                                }
5862                                $INDEX_FILES{$gs->{index}} = 1;
5863                        }
5864                        foreach my $g (@$globs) {
5865                                my $k = "svn-remote.$g->{remote}." .
5866                                        "$g->{t}-maxRev";
5867                                Git::SVN::tmp_config($k, $r);
5868                        }
5869                        if ($ra_invalid) {
5870                                $_[0] = undef;
5871                                $self = undef;
5872                                $RA = undef;
5873                                $self = Git::SVN::Ra->new($ra_url);
5874                                $ra_invalid = undef;
5875                        }
5876                }
5877                # pre-fill the .rev_db since it'll eventually get filled in
5878                # with '0' x40 if something new gets committed
5879                foreach my $gs (@$gsv) {
5880                        next if $gs->rev_map_max >= $max;
5881                        next if defined $gs->rev_map_get($max);
5882                        $gs->rev_map_set($max, 0 x40);
5883                }
5884                foreach my $g (@$globs) {
5885                        my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
5886                        Git::SVN::tmp_config($k, $max);
5887                }
5888                last if $max >= $head;
5889                $min = $max + 1;
5890                $max += $inc;
5891                $max = $head if ($max > $head);
5892        }
5893        Git::SVN::gc();
5894}
5895
5896sub get_dir_globbed {
5897        my ($self, $left, $depth, $r) = @_;
5898
5899        my @x = eval { $self->get_dir($left, $r) };
5900        return unless scalar @x == 3;
5901        my $dirents = $x[0];
5902        my @finalents;
5903        foreach my $de (keys %$dirents) {
5904                next if $dirents->{$de}->{kind} != $SVN::Node::dir;
5905                if ($depth > 1) {
5906                        my @args = ("$left/$de", $depth - 1, $r);
5907                        foreach my $dir ($self->get_dir_globbed(@args)) {
5908                                push @finalents, "$de/$dir";
5909                        }
5910                } else {
5911                        push @finalents, $de;
5912                }
5913        }
5914        @finalents;
5915}
5916
5917# return value: 0 -- don't ignore, 1 -- ignore
5918sub is_ref_ignored {
5919        my ($g, $p) = @_;
5920        my $refname = $g->{ref}->full_path($p);
5921        return 1 if defined($g->{ignore_refs_regex}) &&
5922                    $refname =~ m!$g->{ignore_refs_regex}!;
5923        return 0 unless defined($_ignore_refs_regex);
5924        return 1 if $refname =~ m!$_ignore_refs_regex!o;
5925        return 0;
5926}
5927
5928sub match_globs {
5929        my ($self, $exists, $paths, $globs, $r) = @_;
5930
5931        sub get_dir_check {
5932                my ($self, $exists, $g, $r) = @_;
5933
5934                my @dirs = $self->get_dir_globbed($g->{path}->{left},
5935                                                  $g->{path}->{depth},
5936                                                  $r);
5937
5938                foreach my $de (@dirs) {
5939                        my $p = $g->{path}->full_path($de);
5940                        next if $exists->{$p};
5941                        next if (length $g->{path}->{right} &&
5942                                 ($self->check_path($p, $r) !=
5943                                  $SVN::Node::dir));
5944                        next unless $p =~ /$g->{path}->{regex}/;
5945                        $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
5946                                         $g->{ref}->full_path($de), 1);
5947                }
5948        }
5949        foreach my $g (@$globs) {
5950                if (my $path = $paths->{"/$g->{path}->{left}"}) {
5951                        if ($path->{action} =~ /^[AR]$/) {
5952                                get_dir_check($self, $exists, $g, $r);
5953                        }
5954                }
5955                foreach (keys %$paths) {
5956                        if (/$g->{path}->{left_regex}/ &&
5957                            !/$g->{path}->{regex}/) {
5958                                next if $paths->{$_}->{action} !~ /^[AR]$/;
5959                                get_dir_check($self, $exists, $g, $r);
5960                        }
5961                        next unless /$g->{path}->{regex}/;
5962                        my $p = $1;
5963                        my $pathname = $g->{path}->full_path($p);
5964                        next if is_ref_ignored($g, $p);
5965                        next if $exists->{$pathname};
5966                        next if ($self->check_path($pathname, $r) !=
5967                                 $SVN::Node::dir);
5968                        $exists->{$pathname} = Git::SVN->init(
5969                                              $self->{url}, $pathname, undef,
5970                                              $g->{ref}->full_path($p), 1);
5971                }
5972                my $c = '';
5973                foreach (split m#/#, $g->{path}->{left}) {
5974                        $c .= "/$_";
5975                        next unless ($paths->{$c} &&
5976                                     ($paths->{$c}->{action} =~ /^[AR]$/));
5977                        get_dir_check($self, $exists, $g, $r);
5978                }
5979        }
5980        values %$exists;
5981}
5982
5983sub minimize_url {
5984        my ($self) = @_;
5985        return $self->{url} if ($self->{url} eq $self->{repos_root});
5986        my $url = $self->{repos_root};
5987        my @components = split(m!/!, $self->{svn_path});
5988        my $c = '';
5989        do {
5990                $url .= "/$c" if length $c;
5991                eval {
5992                        my $ra = (ref $self)->new($url);
5993                        my $latest = $ra->get_latest_revnum;
5994                        $ra->get_log("", $latest, 0, 1, 0, 1, sub {});
5995                };
5996        } while ($@ && ($c = shift @components));
5997        $url;
5998}
5999
6000sub can_do_switch {
6001        my $self = shift;
6002        unless (defined $can_do_switch) {
6003                my $pool = SVN::Pool->new;
6004                my $rep = eval {
6005                        $self->do_switch(1, '', 0, $self->{url},
6006                                         SVN::Delta::Editor->new, $pool);
6007                };
6008                if ($@) {
6009                        $can_do_switch = 0;
6010                } else {
6011                        $rep->abort_report($pool);
6012                        $can_do_switch = 1;
6013                }
6014                $pool->clear;
6015        }
6016        $can_do_switch;
6017}
6018
6019sub skip_unknown_revs {
6020        my ($err) = @_;
6021        my $errno = $err->apr_err();
6022        # Maybe the branch we're tracking didn't
6023        # exist when the repo started, so it's
6024        # not an error if it doesn't, just continue
6025        #
6026        # Wonderfully consistent library, eh?
6027        # 160013 - svn:// and file://
6028        # 175002 - http(s)://
6029        # 175007 - http(s):// (this repo required authorization, too...)
6030        #   More codes may be discovered later...
6031        if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
6032                my $err_key = $err->expanded_message;
6033                # revision numbers change every time, filter them out
6034                $err_key =~ s/\d+/\0/g;
6035                $err_key = "$errno\0$err_key";
6036                unless ($ignored_err{$err_key}) {
6037                        warn "W: Ignoring error from SVN, path probably ",
6038                             "does not exist: ($errno): ",
6039                             $err->expanded_message,"\n";
6040                        warn "W: Do not be alarmed at the above message ",
6041                             "git-svn is just searching aggressively for ",
6042                             "old history.\n",
6043                             "This may take a while on large repositories\n";
6044                        $ignored_err{$err_key} = 1;
6045                }
6046                return;
6047        }
6048        die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
6049}
6050
6051package Git::SVN::Log;
6052use strict;
6053use warnings;
6054use POSIX qw/strftime/;
6055use constant commit_log_separator => ('-' x 72) . "\n";
6056use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
6057            %rusers $show_commit $incremental/;
6058my $l_fmt;
6059
6060sub cmt_showable {
6061        my ($c) = @_;
6062        return 1 if defined $c->{r};
6063
6064        # big commit message got truncated by the 16k pretty buffer in rev-list
6065        if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
6066                                $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
6067                @{$c->{l}} = ();
6068                my @log = command(qw/cat-file commit/, $c->{c});
6069
6070                # shift off the headers
6071                shift @log while ($log[0] ne '');
6072                shift @log;
6073
6074                # TODO: make $c->{l} not have a trailing newline in the future
6075                @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
6076
6077                (undef, $c->{r}, undef) = ::extract_metadata(
6078                                (grep(/^git-svn-id: /, @log))[-1]);
6079        }
6080        return defined $c->{r};
6081}
6082
6083sub log_use_color {
6084        return $color || Git->repository->get_colorbool('color.diff');
6085}
6086
6087sub git_svn_log_cmd {
6088        my ($r_min, $r_max, @args) = @_;
6089        my $head = 'HEAD';
6090        my (@files, @log_opts);
6091        foreach my $x (@args) {
6092                if ($x eq '--' || @files) {
6093                        push @files, $x;
6094                } else {
6095                        if (::verify_ref("$x^0")) {
6096                                $head = $x;
6097                        } else {
6098                                push @log_opts, $x;
6099                        }
6100                }
6101        }
6102
6103        my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
6104        $gs ||= Git::SVN->_new;
6105        my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
6106                   $gs->refname);
6107        push @cmd, '-r' unless $non_recursive;
6108        push @cmd, qw/--raw --name-status/ if $verbose;
6109        push @cmd, '--color' if log_use_color();
6110        push @cmd, @log_opts;
6111        if (defined $r_max && $r_max == $r_min) {
6112                push @cmd, '--max-count=1';
6113                if (my $c = $gs->rev_map_get($r_max)) {
6114                        push @cmd, $c;
6115                }
6116        } elsif (defined $r_max) {
6117                if ($r_max < $r_min) {
6118                        ($r_min, $r_max) = ($r_max, $r_min);
6119                }
6120                my (undef, $c_max) = $gs->find_rev_before($r_max, 1, $r_min);
6121                my (undef, $c_min) = $gs->find_rev_after($r_min, 1, $r_max);
6122                # If there are no commits in the range, both $c_max and $c_min
6123                # will be undefined.  If there is at least 1 commit in the
6124                # range, both will be defined.
6125                return () if !defined $c_min || !defined $c_max;
6126                if ($c_min eq $c_max) {
6127                        push @cmd, '--max-count=1', $c_min;
6128                } else {
6129                        push @cmd, '--boundary', "$c_min..$c_max";
6130                }
6131        }
6132        return (@cmd, @files);
6133}
6134
6135# adapted from pager.c
6136sub config_pager {
6137        if (! -t *STDOUT) {
6138                $ENV{GIT_PAGER_IN_USE} = 'false';
6139                $pager = undef;
6140                return;
6141        }
6142        chomp($pager = command_oneline(qw(var GIT_PAGER)));
6143        if ($pager eq 'cat') {
6144                $pager = undef;
6145        }
6146        $ENV{GIT_PAGER_IN_USE} = defined($pager);
6147}
6148
6149sub run_pager {
6150        return unless defined $pager;
6151        pipe my ($rfd, $wfd) or return;
6152        defined(my $pid = fork) or ::fatal "Can't fork: $!";
6153        if (!$pid) {
6154                open STDOUT, '>&', $wfd or
6155                                     ::fatal "Can't redirect to stdout: $!";
6156                return;
6157        }
6158        open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!";
6159        $ENV{LESS} ||= 'FRSX';
6160        exec $pager or ::fatal "Can't run pager: $! ($pager)";
6161}
6162
6163sub format_svn_date {
6164        my $t = shift || time;
6165        my $gmoff = Git::SVN::get_tz($t);
6166        return strftime("%Y-%m-%d %H:%M:%S $gmoff (%a, %d %b %Y)", localtime($t));
6167}
6168
6169sub parse_git_date {
6170        my ($t, $tz) = @_;
6171        # Date::Parse isn't in the standard Perl distro :(
6172        if ($tz =~ s/^\+//) {
6173                $t += tz_to_s_offset($tz);
6174        } elsif ($tz =~ s/^\-//) {
6175                $t -= tz_to_s_offset($tz);
6176        }
6177        return $t;
6178}
6179
6180sub set_local_timezone {
6181        if (defined $TZ) {
6182                $ENV{TZ} = $TZ;
6183        } else {
6184                delete $ENV{TZ};
6185        }
6186}
6187
6188sub tz_to_s_offset {
6189        my ($tz) = @_;
6190        $tz =~ s/(\d\d)$//;
6191        return ($1 * 60) + ($tz * 3600);
6192}
6193
6194sub get_author_info {
6195        my ($dest, $author, $t, $tz) = @_;
6196        $author =~ s/(?:^\s*|\s*$)//g;
6197        $dest->{a_raw} = $author;
6198        my $au;
6199        if ($::_authors) {
6200                $au = $rusers{$author} || undef;
6201        }
6202        if (!$au) {
6203                ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
6204        }
6205        $dest->{t} = $t;
6206        $dest->{tz} = $tz;
6207        $dest->{a} = $au;
6208        $dest->{t_utc} = parse_git_date($t, $tz);
6209}
6210
6211sub process_commit {
6212        my ($c, $r_min, $r_max, $defer) = @_;
6213        if (defined $r_min && defined $r_max) {
6214                if ($r_min == $c->{r} && $r_min == $r_max) {
6215                        show_commit($c);
6216                        return 0;
6217                }
6218                return 1 if $r_min == $r_max;
6219                if ($r_min < $r_max) {
6220                        # we need to reverse the print order
6221                        return 0 if (defined $limit && --$limit < 0);
6222                        push @$defer, $c;
6223                        return 1;
6224                }
6225                if ($r_min != $r_max) {
6226                        return 1 if ($r_min < $c->{r});
6227                        return 1 if ($r_max > $c->{r});
6228                }
6229        }
6230        return 0 if (defined $limit && --$limit < 0);
6231        show_commit($c);
6232        return 1;
6233}
6234
6235sub show_commit {
6236        my $c = shift;
6237        if ($oneline) {
6238                my $x = "\n";
6239                if (my $l = $c->{l}) {
6240                        while ($l->[0] =~ /^\s*$/) { shift @$l }
6241                        $x = $l->[0];
6242                }
6243                $l_fmt ||= 'A' . length($c->{r});
6244                print 'r',pack($l_fmt, $c->{r}),' | ';
6245                print "$c->{c} | " if $show_commit;
6246                print $x;
6247        } else {
6248                show_commit_normal($c);
6249        }
6250}
6251
6252sub show_commit_changed_paths {
6253        my ($c) = @_;
6254        return unless $c->{changed};
6255        print "Changed paths:\n", @{$c->{changed}};
6256}
6257
6258sub show_commit_normal {
6259        my ($c) = @_;
6260        print commit_log_separator, "r$c->{r} | ";
6261        print "$c->{c} | " if $show_commit;
6262        print "$c->{a} | ", format_svn_date($c->{t_utc}), ' | ';
6263        my $nr_line = 0;
6264
6265        if (my $l = $c->{l}) {
6266                while ($l->[$#$l] eq "\n" && $#$l > 0
6267                                          && $l->[($#$l - 1)] eq "\n") {
6268                        pop @$l;
6269                }
6270                $nr_line = scalar @$l;
6271                if (!$nr_line) {
6272                        print "1 line\n\n\n";
6273                } else {
6274                        if ($nr_line == 1) {
6275                                $nr_line = '1 line';
6276                        } else {
6277                                $nr_line .= ' lines';
6278                        }
6279                        print $nr_line, "\n";
6280                        show_commit_changed_paths($c);
6281                        print "\n";
6282                        print $_ foreach @$l;
6283                }
6284        } else {
6285                print "1 line\n";
6286                show_commit_changed_paths($c);
6287                print "\n";
6288
6289        }
6290        foreach my $x (qw/raw stat diff/) {
6291                if ($c->{$x}) {
6292                        print "\n";
6293                        print $_ foreach @{$c->{$x}}
6294                }
6295        }
6296}
6297
6298sub cmd_show_log {
6299        my (@args) = @_;
6300        my ($r_min, $r_max);
6301        my $r_last = -1; # prevent dupes
6302        set_local_timezone();
6303        if (defined $::_revision) {
6304                if ($::_revision =~ /^(\d+):(\d+)$/) {
6305                        ($r_min, $r_max) = ($1, $2);
6306                } elsif ($::_revision =~ /^\d+$/) {
6307                        $r_min = $r_max = $::_revision;
6308                } else {
6309                        ::fatal "-r$::_revision is not supported, use ",
6310                                "standard 'git log' arguments instead";
6311                }
6312        }
6313
6314        config_pager();
6315        @args = git_svn_log_cmd($r_min, $r_max, @args);
6316        if (!@args) {
6317                print commit_log_separator unless $incremental || $oneline;
6318                return;
6319        }
6320        my $log = command_output_pipe(@args);
6321        run_pager();
6322        my (@k, $c, $d, $stat);
6323        my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
6324        while (<$log>) {
6325                if (/^${esc_color}commit (?:- )?($::sha1_short)/o) {
6326                        my $cmt = $1;
6327                        if ($c && cmt_showable($c) && $c->{r} != $r_last) {
6328                                $r_last = $c->{r};
6329                                process_commit($c, $r_min, $r_max, \@k) or
6330                                                                goto out;
6331                        }
6332                        $d = undef;
6333                        $c = { c => $cmt };
6334                } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
6335                        get_author_info($c, $1, $2, $3);
6336                } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
6337                        # ignore
6338                } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
6339                        push @{$c->{raw}}, $_;
6340                } elsif (/^${esc_color}[ACRMDT]\t/) {
6341                        # we could add $SVN->{svn_path} here, but that requires
6342                        # remote access at the moment (repo_path_split)...
6343                        s#^(${esc_color})([ACRMDT])\t#$1   $2 #o;
6344                        push @{$c->{changed}}, $_;
6345                } elsif (/^${esc_color}diff /o) {
6346                        $d = 1;
6347                        push @{$c->{diff}}, $_;
6348                } elsif ($d) {
6349                        push @{$c->{diff}}, $_;
6350                } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
6351                          $esc_color*[\+\-]*$esc_color$/x) {
6352                        $stat = 1;
6353                        push @{$c->{stat}}, $_;
6354                } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
6355                        push @{$c->{stat}}, $_;
6356                        $stat = undef;
6357                } elsif (/^${esc_color}    (git-svn-id:.+)$/o) {
6358                        ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
6359                } elsif (s/^${esc_color}    //o) {
6360                        push @{$c->{l}}, $_;
6361                }
6362        }
6363        if ($c && defined $c->{r} && $c->{r} != $r_last) {
6364                $r_last = $c->{r};
6365                process_commit($c, $r_min, $r_max, \@k);
6366        }
6367        if (@k) {
6368                ($r_min, $r_max) = ($r_max, $r_min);
6369                process_commit($_, $r_min, $r_max) foreach reverse @k;
6370        }
6371out:
6372        close $log;
6373        print commit_log_separator unless $incremental || $oneline;
6374}
6375
6376sub cmd_blame {
6377        my $path = pop;
6378
6379        config_pager();
6380        run_pager();
6381
6382        my ($fh, $ctx, $rev);
6383
6384        if ($_git_format) {
6385                ($fh, $ctx) = command_output_pipe('blame', @_, $path);
6386                while (my $line = <$fh>) {
6387                        if ($line =~ /^\^?([[:xdigit:]]+)\s/) {
6388                                # Uncommitted edits show up as a rev ID of
6389                                # all zeros, which we can't look up with
6390                                # cmt_metadata
6391                                if ($1 !~ /^0+$/) {
6392                                        (undef, $rev, undef) =
6393                                                ::cmt_metadata($1);
6394                                        $rev = '0' if (!$rev);
6395                                } else {
6396                                        $rev = '0';
6397                                }
6398                                $rev = sprintf('%-10s', $rev);
6399                                $line =~ s/^\^?[[:xdigit:]]+(\s)/$rev$1/;
6400                        }
6401                        print $line;
6402                }
6403        } else {
6404                ($fh, $ctx) = command_output_pipe('blame', '-p', @_, 'HEAD',
6405                                                  '--', $path);
6406                my ($sha1);
6407                my %authors;
6408                my @buffer;
6409                my %dsha; #distinct sha keys
6410
6411                while (my $line = <$fh>) {
6412                        push @buffer, $line;
6413                        if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
6414                                $dsha{$1} = 1;
6415                        }
6416                }
6417
6418                my $s2r = ::cmt_sha2rev_batch([keys %dsha]);
6419
6420                foreach my $line (@buffer) {
6421                        if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
6422                                $rev = $s2r->{$1};
6423                                $rev = '0' if (!$rev)
6424                        }
6425                        elsif ($line =~ /^author (.*)/) {
6426                                $authors{$rev} = $1;
6427                                $authors{$rev} =~ s/\s/_/g;
6428                        }
6429                        elsif ($line =~ /^\t(.*)$/) {
6430                                printf("%6s %10s %s\n", $rev, $authors{$rev}, $1);
6431                        }
6432                }
6433        }
6434        command_close_pipe($fh, $ctx);
6435}
6436
6437package Git::SVN::Migration;
6438# these version numbers do NOT correspond to actual version numbers
6439# of git nor git-svn.  They are just relative.
6440#
6441# v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
6442#
6443# v1 layout: .git/$id/info/url, refs/remotes/$id
6444#
6445# v2 layout: .git/svn/$id/info/url, refs/remotes/$id
6446#
6447# v3 layout: .git/svn/$id, refs/remotes/$id
6448#            - info/url may remain for backwards compatibility
6449#            - this is what we migrate up to this layout automatically,
6450#            - this will be used by git svn init on single branches
6451# v3.1 layout (auto migrated):
6452#            - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
6453#              for backwards compatibility
6454#
6455# v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
6456#            - this is only created for newly multi-init-ed
6457#              repositories.  Similar in spirit to the
6458#              --use-separate-remotes option in git-clone (now default)
6459#            - we do not automatically migrate to this (following
6460#              the example set by core git)
6461#
6462# v5 layout: .rev_db.$UUID => .rev_map.$UUID
6463#            - newer, more-efficient format that uses 24-bytes per record
6464#              with no filler space.
6465#            - use xxd -c24 < .rev_map.$UUID to view and debug
6466#            - This is a one-way migration, repositories updated to the
6467#              new format will not be able to use old git-svn without
6468#              rebuilding the .rev_db.  Rebuilding the rev_db is not
6469#              possible if noMetadata or useSvmProps are set; but should
6470#              be no problem for users that use the (sensible) defaults.
6471use strict;
6472use warnings;
6473use Carp qw/croak/;
6474use File::Path qw/mkpath/;
6475use File::Basename qw/dirname basename/;
6476use vars qw/$_minimize/;
6477
6478sub migrate_from_v0 {
6479        my $git_dir = $ENV{GIT_DIR};
6480        return undef unless -d $git_dir;
6481        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
6482        my $migrated = 0;
6483        while (<$fh>) {
6484                chomp;
6485                my ($id, $orig_ref) = ($_, $_);
6486                next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
6487                next unless -f "$git_dir/$id/info/url";
6488                my $new_ref = "refs/remotes/$id";
6489                if (::verify_ref("$new_ref^0")) {
6490                        print STDERR "W: $orig_ref is probably an old ",
6491                                     "branch used by an ancient version of ",
6492                                     "git-svn.\n",
6493                                     "However, $new_ref also exists.\n",
6494                                     "We will not be able ",
6495                                     "to use this branch until this ",
6496                                     "ambiguity is resolved.\n";
6497                        next;
6498                }
6499                print STDERR "Migrating from v0 layout...\n" if !$migrated;
6500                print STDERR "Renaming ref: $orig_ref => $new_ref\n";
6501                command_noisy('update-ref', $new_ref, $orig_ref);
6502                command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
6503                $migrated++;
6504        }
6505        command_close_pipe($fh, $ctx);
6506        print STDERR "Done migrating from v0 layout...\n" if $migrated;
6507        $migrated;
6508}
6509
6510sub migrate_from_v1 {
6511        my $git_dir = $ENV{GIT_DIR};
6512        my $migrated = 0;
6513        return $migrated unless -d $git_dir;
6514        my $svn_dir = "$git_dir/svn";
6515
6516        # just in case somebody used 'svn' as their $id at some point...
6517        return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
6518
6519        print STDERR "Migrating from a git-svn v1 layout...\n";
6520        mkpath([$svn_dir]);
6521        print STDERR "Data from a previous version of git-svn exists, but\n\t",
6522                     "$svn_dir\n\t(required for this version ",
6523                     "($::VERSION) of git-svn) does not exist.\n";
6524        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
6525        while (<$fh>) {
6526                my $x = $_;
6527                next unless $x =~ s#^refs/remotes/##;
6528                chomp $x;
6529                next unless -f "$git_dir/$x/info/url";
6530                my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
6531                next unless $u;
6532                my $dn = dirname("$git_dir/svn/$x");
6533                mkpath([$dn]) unless -d $dn;
6534                if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
6535                        mkpath(["$git_dir/svn/svn"]);
6536                        print STDERR " - $git_dir/$x/info => ",
6537                                        "$git_dir/svn/$x/info\n";
6538                        rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
6539                               croak "$!: $x";
6540                        # don't worry too much about these, they probably
6541                        # don't exist with repos this old (save for index,
6542                        # and we can easily regenerate that)
6543                        foreach my $f (qw/unhandled.log index .rev_db/) {
6544                                rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
6545                        }
6546                } else {
6547                        print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
6548                        rename "$git_dir/$x", "$git_dir/svn/$x" or
6549                               croak "$!: $x";
6550                }
6551                $migrated++;
6552        }
6553        command_close_pipe($fh, $ctx);
6554        print STDERR "Done migrating from a git-svn v1 layout\n";
6555        $migrated;
6556}
6557
6558sub read_old_urls {
6559        my ($l_map, $pfx, $path) = @_;
6560        my @dir;
6561        foreach (<$path/*>) {
6562                if (-r "$_/info/url") {
6563                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
6564                        my $ref_id = $pfx . basename $_;
6565                        my $url = ::file_to_s("$_/info/url");
6566                        $l_map->{$ref_id} = $url;
6567                } elsif (-d $_) {
6568                        push @dir, $_;
6569                }
6570        }
6571        foreach (@dir) {
6572                my $x = $_;
6573                $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
6574                read_old_urls($l_map, $x, $_);
6575        }
6576}
6577
6578sub migrate_from_v2 {
6579        my @cfg = command(qw/config -l/);
6580        return if grep /^svn-remote\..+\.url=/, @cfg;
6581        my %l_map;
6582        read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
6583        my $migrated = 0;
6584
6585        foreach my $ref_id (sort keys %l_map) {
6586                eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
6587                if ($@) {
6588                        Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
6589                }
6590                $migrated++;
6591        }
6592        $migrated;
6593}
6594
6595sub minimize_connections {
6596        my $r = Git::SVN::read_all_remotes();
6597        my $new_urls = {};
6598        my $root_repos = {};
6599        foreach my $repo_id (keys %$r) {
6600                my $url = $r->{$repo_id}->{url} or next;
6601                my $fetch = $r->{$repo_id}->{fetch} or next;
6602                my $ra = Git::SVN::Ra->new($url);
6603
6604                # skip existing cases where we already connect to the root
6605                if (($ra->{url} eq $ra->{repos_root}) ||
6606                    ($ra->{repos_root} eq $repo_id)) {
6607                        $root_repos->{$ra->{url}} = $repo_id;
6608                        next;
6609                }
6610
6611                my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
6612                my $root_path = $ra->{url};
6613                $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
6614                foreach my $path (keys %$fetch) {
6615                        my $ref_id = $fetch->{$path};
6616                        my $gs = Git::SVN->new($ref_id, $repo_id, $path);
6617
6618                        # make sure we can read when connecting to
6619                        # a higher level of a repository
6620                        my ($last_rev, undef) = $gs->last_rev_commit;
6621                        if (!defined $last_rev) {
6622                                $last_rev = eval {
6623                                        $root_ra->get_latest_revnum;
6624                                };
6625                                next if $@;
6626                        }
6627                        my $new = $root_path;
6628                        $new .= length $path ? "/$path" : '';
6629                        eval {
6630                                $root_ra->get_log([$new], $last_rev, $last_rev,
6631                                                  0, 0, 1, sub { });
6632                        };
6633                        next if $@;
6634                        $new_urls->{$ra->{repos_root}}->{$new} =
6635                                { ref_id => $ref_id,
6636                                  old_repo_id => $repo_id,
6637                                  old_path => $path };
6638                }
6639        }
6640
6641        my @emptied;
6642        foreach my $url (keys %$new_urls) {
6643                # see if we can re-use an existing [svn-remote "repo_id"]
6644                # instead of creating a(n ugly) new section:
6645                my $repo_id = $root_repos->{$url} || $url;
6646
6647                my $fetch = $new_urls->{$url};
6648                foreach my $path (keys %$fetch) {
6649                        my $x = $fetch->{$path};
6650                        Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
6651                        my $pfx = "svn-remote.$x->{old_repo_id}";
6652
6653                        my $old_fetch = quotemeta("$x->{old_path}:".
6654                                                  "$x->{ref_id}");
6655                        command_noisy(qw/config --unset/,
6656                                      "$pfx.fetch", '^'. $old_fetch . '$');
6657                        delete $r->{$x->{old_repo_id}}->
6658                               {fetch}->{$x->{old_path}};
6659                        if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
6660                                command_noisy(qw/config --unset/,
6661                                              "$pfx.url");
6662                                push @emptied, $x->{old_repo_id}
6663                        }
6664                }
6665        }
6666        if (@emptied) {
6667                my $file = $ENV{GIT_CONFIG} || "$ENV{GIT_DIR}/config";
6668                print STDERR <<EOF;
6669The following [svn-remote] sections in your config file ($file) are empty
6670and can be safely removed:
6671EOF
6672                print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
6673        }
6674}
6675
6676sub migration_check {
6677        migrate_from_v0();
6678        migrate_from_v1();
6679        migrate_from_v2();
6680        minimize_connections() if $_minimize;
6681}
6682
6683package Git::IndexInfo;
6684use strict;
6685use warnings;
6686use Git qw/command_input_pipe command_close_pipe/;
6687
6688sub new {
6689        my ($class) = @_;
6690        my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
6691        bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
6692}
6693
6694sub remove {
6695        my ($self, $path) = @_;
6696        if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
6697                return ++$self->{nr};
6698        }
6699        undef;
6700}
6701
6702sub update {
6703        my ($self, $mode, $hash, $path) = @_;
6704        if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
6705                return ++$self->{nr};
6706        }
6707        undef;
6708}
6709
6710sub DESTROY {
6711        my ($self) = @_;
6712        command_close_pipe($self->{gui}, $self->{ctx});
6713}
6714
6715package Git::SVN::GlobSpec;
6716use strict;
6717use warnings;
6718
6719sub new {
6720        my ($class, $glob, $pattern_ok) = @_;
6721        my $re = $glob;
6722        $re =~ s!/+$!!g; # no need for trailing slashes
6723        my (@left, @right, @patterns);
6724        my $state = "left";
6725        my $die_msg = "Only one set of wildcard directories " .
6726                                "(e.g. '*' or '*/*/*') is supported: '$glob'\n";
6727        for my $part (split(m|/|, $glob)) {
6728                if ($part =~ /\*/ && $part ne "*") {
6729                        die "Invalid pattern in '$glob': $part\n";
6730                } elsif ($pattern_ok && $part =~ /[{}]/ &&
6731                         $part !~ /^\{[^{}]+\}/) {
6732                        die "Invalid pattern in '$glob': $part\n";
6733                }
6734                if ($part eq "*") {
6735                        die $die_msg if $state eq "right";
6736                        $state = "pattern";
6737                        push(@patterns, "[^/]*");
6738                } elsif ($pattern_ok && $part =~ /^\{(.*)\}$/) {
6739                        die $die_msg if $state eq "right";
6740                        $state = "pattern";
6741                        my $p = quotemeta($1);
6742                        $p =~ s/\\,/|/g;
6743                        push(@patterns, "(?:$p)");
6744                } else {
6745                        if ($state eq "left") {
6746                                push(@left, $part);
6747                        } else {
6748                                push(@right, $part);
6749                                $state = "right";
6750                        }
6751                }
6752        }
6753        my $depth = @patterns;
6754        if ($depth == 0) {
6755                die "One '*' is needed in glob: '$glob'\n";
6756        }
6757        my $left = join('/', @left);
6758        my $right = join('/', @right);
6759        $re = join('/', @patterns);
6760        $re = join('\/',
6761                   grep(length, quotemeta($left), "($re)", quotemeta($right)));
6762        my $left_re = qr/^\/\Q$left\E(\/|$)/;
6763        bless { left => $left, right => $right, left_regex => $left_re,
6764                regex => qr/$re/, glob => $glob, depth => $depth }, $class;
6765}
6766
6767sub full_path {
6768        my ($self, $path) = @_;
6769        return (length $self->{left} ? "$self->{left}/" : '') .
6770               $path . (length $self->{right} ? "/$self->{right}" : '');
6771}
6772
6773__END__
6774
6775Data structures:
6776
6777
6778$remotes = { # returned by read_all_remotes()
6779        'svn' => {
6780                # svn-remote.svn.url=https://svn.musicpd.org
6781                url => 'https://svn.musicpd.org',
6782                # svn-remote.svn.fetch=mpd/trunk:trunk
6783                fetch => {
6784                        'mpd/trunk' => 'trunk',
6785                },
6786                # svn-remote.svn.tags=mpd/tags/*:tags/*
6787                tags => {
6788                        path => {
6789                                left => 'mpd/tags',
6790                                right => '',
6791                                regex => qr!mpd/tags/([^/]+)$!,
6792                                glob => 'tags/*',
6793                        },
6794                        ref => {
6795                                left => 'tags',
6796                                right => '',
6797                                regex => qr!tags/([^/]+)$!,
6798                                glob => 'tags/*',
6799                        },
6800                }
6801        }
6802};
6803
6804$log_entry hashref as returned by libsvn_log_entry()
6805{
6806        log => 'whitespace-formatted log entry
6807',                                              # trailing newline is preserved
6808        revision => '8',                        # integer
6809        date => '2004-02-24T17:01:44.108345Z',  # commit date
6810        author => 'committer name'
6811};
6812
6813
6814# this is generated by generate_diff();
6815@mods = array of diff-index line hashes, each element represents one line
6816        of diff-index output
6817
6818diff-index line ($m hash)
6819{
6820        mode_a => first column of diff-index output, no leading ':',
6821        mode_b => second column of diff-index output,
6822        sha1_b => sha1sum of the final blob,
6823        chg => change type [MCRADT],
6824        file_a => original file name of a file (iff chg is 'C' or 'R')
6825        file_b => new/current file name of a file (any chg)
6826}
6827;
6828
6829# retval of read_url_paths{,_all}();
6830$l_map = {
6831        # repository root url
6832        'https://svn.musicpd.org' => {
6833                # repository path               # GIT_SVN_ID
6834                'mpd/trunk'             =>      'trunk',
6835                'mpd/tags/0.11.5'       =>      'tags/0.11.5',
6836        },
6837}
6838
6839Notes:
6840        I don't trust the each() function on unless I created %hash myself
6841        because the internal iterator may not have started at base.