git-svn.perlon commit Merge branch 'jc/unseekable-bundle' (b76c561)
   1#!/usr/bin/env 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$| = 1; # unbuffer STDOUT
  37
  38sub fatal (@) { print STDERR "@_\n"; exit 1 }
  39sub _req_svn {
  40        require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
  41        require SVN::Ra;
  42        require SVN::Delta;
  43        if ($SVN::Core::VERSION lt '1.1.0') {
  44                fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)";
  45        }
  46}
  47my $can_compress = eval { require Compress::Zlib; 1};
  48push @Git::SVN::Ra::ISA, 'SVN::Ra';
  49push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
  50push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
  51use Carp qw/croak/;
  52use Digest::MD5;
  53use IO::File qw//;
  54use File::Basename qw/dirname basename/;
  55use File::Path qw/mkpath/;
  56use File::Spec;
  57use File::Find;
  58use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev/;
  59use IPC::Open3;
  60use Git;
  61use Memoize;  # core since 5.8.0, Jul 2002
  62
  63BEGIN {
  64        # import functions from Git into our packages, en masse
  65        no strict 'refs';
  66        foreach (qw/command command_oneline command_noisy command_output_pipe
  67                    command_input_pipe command_close_pipe
  68                    command_bidi_pipe command_close_bidi_pipe/) {
  69                for my $package ( qw(SVN::Git::Editor SVN::Git::Fetcher
  70                        Git::SVN::Migration Git::SVN::Log Git::SVN),
  71                        __PACKAGE__) {
  72                        *{"${package}::$_"} = \&{"Git::$_"};
  73                }
  74        }
  75        Memoize::memoize 'Git::config';
  76        Memoize::memoize 'Git::config_bool';
  77}
  78
  79my ($SVN);
  80
  81$sha1 = qr/[a-f\d]{40}/;
  82$sha1_short = qr/[a-f\d]{4,40}/;
  83my ($_stdin, $_help, $_edit,
  84        $_message, $_file, $_branch_dest,
  85        $_template, $_shared,
  86        $_version, $_fetch_all, $_no_rebase, $_fetch_parent,
  87        $_merge, $_strategy, $_dry_run, $_local,
  88        $_prefix, $_no_checkout, $_url, $_verbose,
  89        $_git_format, $_commit_url, $_tag, $_merge_info, $_interactive);
  90$Git::SVN::_follow_parent = 1;
  91$SVN::Git::Fetcher::_placeholder_filename = ".gitignore";
  92$_q ||= 0;
  93my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
  94                    'config-dir=s' => \$Git::SVN::Ra::config_dir,
  95                    'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache,
  96                    'ignore-paths=s' => \$SVN::Git::Fetcher::_ignore_regex,
  97                    'ignore-refs=s' => \$Git::SVN::Ra::_ignore_refs_regex );
  98my %fc_opts = ( 'follow-parent|follow!' => \$Git::SVN::_follow_parent,
  99                'authors-file|A=s' => \$_authors,
 100                'authors-prog=s' => \$_authors_prog,
 101                'repack:i' => \$Git::SVN::_repack,
 102                'noMetadata' => \$Git::SVN::_no_metadata,
 103                'useSvmProps' => \$Git::SVN::_use_svm_props,
 104                'useSvnsyncProps' => \$Git::SVN::_use_svnsync_props,
 105                'log-window-size=i' => \$Git::SVN::Ra::_log_window_size,
 106                'no-checkout' => \$_no_checkout,
 107                'quiet|q+' => \$_q,
 108                'repack-flags|repack-args|repack-opts=s' =>
 109                   \$Git::SVN::_repack_flags,
 110                'use-log-author' => \$Git::SVN::_use_log_author,
 111                'add-author-from' => \$Git::SVN::_add_author_from,
 112                'localtime' => \$Git::SVN::_localtime,
 113                %remote_opts );
 114
 115my ($_trunk, @_tags, @_branches, $_stdlayout);
 116my %icv;
 117my %init_opts = ( 'template=s' => \$_template, 'shared:s' => \$_shared,
 118                  'trunk|T=s' => \$_trunk, 'tags|t=s@' => \@_tags,
 119                  'branches|b=s@' => \@_branches, 'prefix=s' => \$_prefix,
 120                  'stdlayout|s' => \$_stdlayout,
 121                  'minimize-url|m!' => \$Git::SVN::_minimize_url,
 122                  'no-metadata' => sub { $icv{noMetadata} = 1 },
 123                  'use-svm-props' => sub { $icv{useSvmProps} = 1 },
 124                  'use-svnsync-props' => sub { $icv{useSvnsyncProps} = 1 },
 125                  'rewrite-root=s' => sub { $icv{rewriteRoot} = $_[1] },
 126                  'rewrite-uuid=s' => sub { $icv{rewriteUUID} = $_[1] },
 127                  %remote_opts );
 128my %cmt_opts = ( 'edit|e' => \$_edit,
 129                'rmdir' => \$SVN::Git::Editor::_rmdir,
 130                'find-copies-harder' => \$SVN::Git::Editor::_find_copies_harder,
 131                'l=i' => \$SVN::Git::Editor::_rename_limit,
 132                'copy-similarity|C=i'=> \$SVN::Git::Editor::_cp_similarity
 133);
 134
 135my %cmd = (
 136        fetch => [ \&cmd_fetch, "Download new revisions from SVN",
 137                        { 'revision|r=s' => \$_revision,
 138                          'fetch-all|all' => \$_fetch_all,
 139                          'parent|p' => \$_fetch_parent,
 140                           %fc_opts } ],
 141        clone => [ \&cmd_clone, "Initialize and fetch revisions",
 142                        { 'revision|r=s' => \$_revision,
 143                          'preserve-empty-dirs' =>
 144                                \$SVN::Git::Fetcher::_preserve_empty_dirs,
 145                          'placeholder-filename=s' =>
 146                                \$SVN::Git::Fetcher::_placeholder_filename,
 147                           %fc_opts, %init_opts } ],
 148        init => [ \&cmd_init, "Initialize a repo for tracking" .
 149                          " (requires URL argument)",
 150                          \%init_opts ],
 151        'multi-init' => [ \&cmd_multi_init,
 152                          "Deprecated alias for ".
 153                          "'$0 init -T<trunk> -b<branches> -t<tags>'",
 154                          \%init_opts ],
 155        dcommit => [ \&cmd_dcommit,
 156                     'Commit several diffs to merge with upstream',
 157                        { 'merge|m|M' => \$_merge,
 158                          'strategy|s=s' => \$_strategy,
 159                          'verbose|v' => \$_verbose,
 160                          'dry-run|n' => \$_dry_run,
 161                          'fetch-all|all' => \$_fetch_all,
 162                          'commit-url=s' => \$_commit_url,
 163                          'revision|r=i' => \$_revision,
 164                          'no-rebase' => \$_no_rebase,
 165                          'mergeinfo=s' => \$_merge_info,
 166                          'interactive|i' => \$_interactive,
 167                        %cmt_opts, %fc_opts } ],
 168        branch => [ \&cmd_branch,
 169                    'Create a branch in the SVN repository',
 170                    { 'message|m=s' => \$_message,
 171                      'destination|d=s' => \$_branch_dest,
 172                      'dry-run|n' => \$_dry_run,
 173                      'tag|t' => \$_tag,
 174                      'username=s' => \$Git::SVN::Prompt::_username,
 175                      'commit-url=s' => \$_commit_url } ],
 176        tag => [ sub { $_tag = 1; cmd_branch(@_) },
 177                 'Create a tag in the SVN repository',
 178                 { 'message|m=s' => \$_message,
 179                   'destination|d=s' => \$_branch_dest,
 180                   'dry-run|n' => \$_dry_run,
 181                   'username=s' => \$Git::SVN::Prompt::_username,
 182                   'commit-url=s' => \$_commit_url } ],
 183        'set-tree' => [ \&cmd_set_tree,
 184                        "Set an SVN repository to a git tree-ish",
 185                        { 'stdin' => \$_stdin, %cmt_opts, %fc_opts, } ],
 186        'create-ignore' => [ \&cmd_create_ignore,
 187                             'Create a .gitignore per svn:ignore',
 188                             { 'revision|r=i' => \$_revision
 189                             } ],
 190        'mkdirs' => [ \&cmd_mkdirs ,
 191                      "recreate empty directories after a checkout",
 192                      { 'revision|r=i' => \$_revision } ],
 193        'propget' => [ \&cmd_propget,
 194                       'Print the value of a property on a file or directory',
 195                       { 'revision|r=i' => \$_revision } ],
 196        'proplist' => [ \&cmd_proplist,
 197                       'List all properties of a file or directory',
 198                       { 'revision|r=i' => \$_revision } ],
 199        'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
 200                        { 'revision|r=i' => \$_revision
 201                        } ],
 202        'show-externals' => [ \&cmd_show_externals, "Show svn:externals listings",
 203                        { 'revision|r=i' => \$_revision
 204                        } ],
 205        'multi-fetch' => [ \&cmd_multi_fetch,
 206                           "Deprecated alias for $0 fetch --all",
 207                           { 'revision|r=s' => \$_revision, %fc_opts } ],
 208        'migrate' => [ sub { },
 209                       # no-op, we automatically run this anyways,
 210                       'Migrate configuration/metadata/layout from
 211                        previous versions of git-svn',
 212                       { 'minimize' => \$Git::SVN::Migration::_minimize,
 213                         %remote_opts } ],
 214        'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
 215                        { 'limit=i' => \$Git::SVN::Log::limit,
 216                          'revision|r=s' => \$_revision,
 217                          'verbose|v' => \$Git::SVN::Log::verbose,
 218                          'incremental' => \$Git::SVN::Log::incremental,
 219                          'oneline' => \$Git::SVN::Log::oneline,
 220                          'show-commit' => \$Git::SVN::Log::show_commit,
 221                          'non-recursive' => \$Git::SVN::Log::non_recursive,
 222                          'authors-file|A=s' => \$_authors,
 223                          'color' => \$Git::SVN::Log::color,
 224                          'pager=s' => \$Git::SVN::Log::pager
 225                        } ],
 226        'find-rev' => [ \&cmd_find_rev,
 227                        "Translate between SVN revision numbers and tree-ish",
 228                        {} ],
 229        'rebase' => [ \&cmd_rebase, "Fetch and rebase your working directory",
 230                        { 'merge|m|M' => \$_merge,
 231                          'verbose|v' => \$_verbose,
 232                          'strategy|s=s' => \$_strategy,
 233                          'local|l' => \$_local,
 234                          'fetch-all|all' => \$_fetch_all,
 235                          'dry-run|n' => \$_dry_run,
 236                          %fc_opts } ],
 237        'commit-diff' => [ \&cmd_commit_diff,
 238                           'Commit a diff between two trees',
 239                        { 'message|m=s' => \$_message,
 240                          'file|F=s' => \$_file,
 241                          'revision|r=s' => \$_revision,
 242                        %cmt_opts } ],
 243        'info' => [ \&cmd_info,
 244                    "Show info about the latest SVN revision
 245                     on the current branch",
 246                    { 'url' => \$_url, } ],
 247        'blame' => [ \&Git::SVN::Log::cmd_blame,
 248                    "Show what revision and author last modified each line of a file",
 249                    { 'git-format' => \$_git_format } ],
 250        'reset' => [ \&cmd_reset,
 251                     "Undo fetches back to the specified SVN revision",
 252                     { 'revision|r=s' => \$_revision,
 253                       'parent|p' => \$_fetch_parent } ],
 254        'gc' => [ \&cmd_gc,
 255                  "Compress unhandled.log files in .git/svn and remove " .
 256                  "index files in .git/svn",
 257                {} ],
 258);
 259
 260use Term::ReadLine;
 261package FakeTerm;
 262sub new {
 263        my ($class, $reason) = @_;
 264        return bless \$reason, shift;
 265}
 266sub readline {
 267        my $self = shift;
 268        die "Cannot use readline on FakeTerm: $$self";
 269}
 270package main;
 271
 272my $term = eval {
 273        $ENV{"GIT_SVN_NOTTY"}
 274                ? new Term::ReadLine 'git-svn', \*STDIN, \*STDOUT
 275                : new Term::ReadLine 'git-svn';
 276};
 277if ($@) {
 278        $term = new FakeTerm "$@: going non-interactive";
 279}
 280
 281my $cmd;
 282for (my $i = 0; $i < @ARGV; $i++) {
 283        if (defined $cmd{$ARGV[$i]}) {
 284                $cmd = $ARGV[$i];
 285                splice @ARGV, $i, 1;
 286                last;
 287        } elsif ($ARGV[$i] eq 'help') {
 288                $cmd = $ARGV[$i+1];
 289                usage(0);
 290        }
 291};
 292
 293# make sure we're always running at the top-level working directory
 294unless ($cmd && $cmd =~ /(?:clone|init|multi-init)$/) {
 295        unless (-d $ENV{GIT_DIR}) {
 296                if ($git_dir_user_set) {
 297                        die "GIT_DIR=$ENV{GIT_DIR} explicitly set, ",
 298                            "but it is not a directory\n";
 299                }
 300                my $git_dir = delete $ENV{GIT_DIR};
 301                my $cdup = undef;
 302                git_cmd_try {
 303                        $cdup = command_oneline(qw/rev-parse --show-cdup/);
 304                        $git_dir = '.' unless ($cdup);
 305                        chomp $cdup if ($cdup);
 306                        $cdup = "." unless ($cdup && length $cdup);
 307                } "Already at toplevel, but $git_dir not found\n";
 308                chdir $cdup or die "Unable to chdir up to '$cdup'\n";
 309                unless (-d $git_dir) {
 310                        die "$git_dir still not found after going to ",
 311                            "'$cdup'\n";
 312                }
 313                $ENV{GIT_DIR} = $git_dir;
 314        }
 315        $_repository = Git->repository(Repository => $ENV{GIT_DIR});
 316}
 317
 318my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
 319
 320read_git_config(\%opts);
 321if ($cmd && ($cmd eq 'log' || $cmd eq 'blame')) {
 322        Getopt::Long::Configure('pass_through');
 323}
 324my $rv = GetOptions(%opts, 'h|H' => \$_help, 'version|V' => \$_version,
 325                    'minimize-connections' => \$Git::SVN::Migration::_minimize,
 326                    'id|i=s' => \$Git::SVN::default_ref_id,
 327                    'svn-remote|remote|R=s' => sub {
 328                       $Git::SVN::no_reuse_existing = 1;
 329                       $Git::SVN::default_repo_id = $_[1] });
 330exit 1 if (!$rv && $cmd && $cmd ne 'log');
 331
 332usage(0) if $_help;
 333version() if $_version;
 334usage(1) unless defined $cmd;
 335load_authors() if $_authors;
 336if (defined $_authors_prog) {
 337        $_authors_prog = "'" . File::Spec->rel2abs($_authors_prog) . "'";
 338}
 339
 340unless ($cmd =~ /^(?:clone|init|multi-init|commit-diff)$/) {
 341        Git::SVN::Migration::migration_check();
 342}
 343Git::SVN::init_vars();
 344eval {
 345        Git::SVN::verify_remotes_sanity();
 346        $cmd{$cmd}->[0]->(@ARGV);
 347};
 348fatal $@ if $@;
 349post_fetch_checkout();
 350exit 0;
 351
 352####################### primary functions ######################
 353sub usage {
 354        my $exit = shift || 0;
 355        my $fd = $exit ? \*STDERR : \*STDOUT;
 356        print $fd <<"";
 357git-svn - bidirectional operations between a single Subversion tree and git
 358Usage: git svn <command> [options] [arguments]\n
 359
 360        print $fd "Available commands:\n" unless $cmd;
 361
 362        foreach (sort keys %cmd) {
 363                next if $cmd && $cmd ne $_;
 364                next if /^multi-/; # don't show deprecated commands
 365                print $fd '  ',pack('A17',$_),$cmd{$_}->[1],"\n";
 366                foreach (sort keys %{$cmd{$_}->[2]}) {
 367                        # mixed-case options are for .git/config only
 368                        next if /[A-Z]/ && /^[a-z]+$/i;
 369                        # prints out arguments as they should be passed:
 370                        my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
 371                        print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
 372                                                        "--$_" : "-$_" }
 373                                                split /\|/,$_)," $x\n";
 374                }
 375        }
 376        print $fd <<"";
 377\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
 378arbitrary identifier if you're tracking multiple SVN branches/repositories in
 379one git repository and want to keep them separate.  See git-svn(1) for more
 380information.
 381
 382        exit $exit;
 383}
 384
 385sub version {
 386        ::_req_svn();
 387        print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
 388        exit 0;
 389}
 390
 391sub ask {
 392        my ($prompt, %arg) = @_;
 393        my $valid_re = $arg{valid_re};
 394        my $default = $arg{default};
 395        my $resp;
 396        my $i = 0;
 397
 398        if ( !( defined($term->IN)
 399            && defined( fileno($term->IN) )
 400            && defined( $term->OUT )
 401            && defined( fileno($term->OUT) ) ) ){
 402                return defined($default) ? $default : undef;
 403        }
 404
 405        while ($i++ < 10) {
 406                $resp = $term->readline($prompt);
 407                if (!defined $resp) { # EOF
 408                        print "\n";
 409                        return defined $default ? $default : undef;
 410                }
 411                if ($resp eq '' and defined $default) {
 412                        return $default;
 413                }
 414                if (!defined $valid_re or $resp =~ /$valid_re/) {
 415                        return $resp;
 416                }
 417        }
 418        return undef;
 419}
 420
 421sub do_git_init_db {
 422        unless (-d $ENV{GIT_DIR}) {
 423                my @init_db = ('init');
 424                push @init_db, "--template=$_template" if defined $_template;
 425                if (defined $_shared) {
 426                        if ($_shared =~ /[a-z]/) {
 427                                push @init_db, "--shared=$_shared";
 428                        } else {
 429                                push @init_db, "--shared";
 430                        }
 431                }
 432                command_noisy(@init_db);
 433                $_repository = Git->repository(Repository => ".git");
 434        }
 435        my $set;
 436        my $pfx = "svn-remote.$Git::SVN::default_repo_id";
 437        foreach my $i (keys %icv) {
 438                die "'$set' and '$i' cannot both be set\n" if $set;
 439                next unless defined $icv{$i};
 440                command_noisy('config', "$pfx.$i", $icv{$i});
 441                $set = $i;
 442        }
 443        my $ignore_paths_regex = \$SVN::Git::Fetcher::_ignore_regex;
 444        command_noisy('config', "$pfx.ignore-paths", $$ignore_paths_regex)
 445                if defined $$ignore_paths_regex;
 446        my $ignore_refs_regex = \$Git::SVN::Ra::_ignore_refs_regex;
 447        command_noisy('config', "$pfx.ignore-refs", $$ignore_refs_regex)
 448                if defined $$ignore_refs_regex;
 449
 450        if (defined $SVN::Git::Fetcher::_preserve_empty_dirs) {
 451                my $fname = \$SVN::Git::Fetcher::_placeholder_filename;
 452                command_noisy('config', "$pfx.preserve-empty-dirs", 'true');
 453                command_noisy('config', "$pfx.placeholder-filename", $$fname);
 454        }
 455}
 456
 457sub init_subdir {
 458        my $repo_path = shift or return;
 459        mkpath([$repo_path]) unless -d $repo_path;
 460        chdir $repo_path or die "Couldn't chdir to $repo_path: $!\n";
 461        $ENV{GIT_DIR} = '.git';
 462        $_repository = Git->repository(Repository => $ENV{GIT_DIR});
 463}
 464
 465sub cmd_clone {
 466        my ($url, $path) = @_;
 467        if (!defined $path &&
 468            (defined $_trunk || @_branches || @_tags ||
 469             defined $_stdlayout) &&
 470            $url !~ m#^[a-z\+]+://#) {
 471                $path = $url;
 472        }
 473        $path = basename($url) if !defined $path || !length $path;
 474        my $authors_absolute = $_authors ? File::Spec->rel2abs($_authors) : "";
 475        cmd_init($url, $path);
 476        command_oneline('config', 'svn.authorsfile', $authors_absolute)
 477            if $_authors;
 478        Git::SVN::fetch_all($Git::SVN::default_repo_id);
 479}
 480
 481sub cmd_init {
 482        if (defined $_stdlayout) {
 483                $_trunk = 'trunk' if (!defined $_trunk);
 484                @_tags = 'tags' if (! @_tags);
 485                @_branches = 'branches' if (! @_branches);
 486        }
 487        if (defined $_trunk || @_branches || @_tags) {
 488                return cmd_multi_init(@_);
 489        }
 490        my $url = shift or die "SVN repository location required ",
 491                               "as a command-line argument\n";
 492        $url = canonicalize_url($url);
 493        init_subdir(@_);
 494        do_git_init_db();
 495
 496        if ($Git::SVN::_minimize_url eq 'unset') {
 497                $Git::SVN::_minimize_url = 0;
 498        }
 499
 500        Git::SVN->init($url);
 501}
 502
 503sub cmd_fetch {
 504        if (grep /^\d+=./, @_) {
 505                die "'<rev>=<commit>' fetch arguments are ",
 506                    "no longer supported.\n";
 507        }
 508        my ($remote) = @_;
 509        if (@_ > 1) {
 510                die "Usage: $0 fetch [--all] [--parent] [svn-remote]\n";
 511        }
 512        $Git::SVN::no_reuse_existing = undef;
 513        if ($_fetch_parent) {
 514                my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 515                unless ($gs) {
 516                        die "Unable to determine upstream SVN information from ",
 517                            "working tree history\n";
 518                }
 519                # just fetch, don't checkout.
 520                $_no_checkout = 'true';
 521                $_fetch_all ? $gs->fetch_all : $gs->fetch;
 522        } elsif ($_fetch_all) {
 523                cmd_multi_fetch();
 524        } else {
 525                $remote ||= $Git::SVN::default_repo_id;
 526                Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
 527        }
 528}
 529
 530sub cmd_set_tree {
 531        my (@commits) = @_;
 532        if ($_stdin || !@commits) {
 533                print "Reading from stdin...\n";
 534                @commits = ();
 535                while (<STDIN>) {
 536                        if (/\b($sha1_short)\b/o) {
 537                                unshift @commits, $1;
 538                        }
 539                }
 540        }
 541        my @revs;
 542        foreach my $c (@commits) {
 543                my @tmp = command('rev-parse',$c);
 544                if (scalar @tmp == 1) {
 545                        push @revs, $tmp[0];
 546                } elsif (scalar @tmp > 1) {
 547                        push @revs, reverse(command('rev-list',@tmp));
 548                } else {
 549                        fatal "Failed to rev-parse $c";
 550                }
 551        }
 552        my $gs = Git::SVN->new;
 553        my ($r_last, $cmt_last) = $gs->last_rev_commit;
 554        $gs->fetch;
 555        if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
 556                fatal "There are new revisions that were fetched ",
 557                      "and need to be merged (or acknowledged) ",
 558                      "before committing.\nlast rev: $r_last\n",
 559                      " current: $gs->{last_rev}";
 560        }
 561        $gs->set_tree($_) foreach @revs;
 562        print "Done committing ",scalar @revs," revisions to SVN\n";
 563        unlink $gs->{index};
 564}
 565
 566sub split_merge_info_range {
 567        my ($range) = @_;
 568        if ($range =~ /(\d+)-(\d+)/) {
 569                return (int($1), int($2));
 570        } else {
 571                return (int($range), int($range));
 572        }
 573}
 574
 575sub combine_ranges {
 576        my ($in) = @_;
 577
 578        my @fnums = ();
 579        my @arr = split(/,/, $in);
 580        for my $element (@arr) {
 581                my ($start, $end) = split_merge_info_range($element);
 582                push @fnums, $start;
 583        }
 584
 585        my @sorted = @arr [ sort {
 586                $fnums[$a] <=> $fnums[$b]
 587        } 0..$#arr ];
 588
 589        my @return = ();
 590        my $last = -1;
 591        my $first = -1;
 592        for my $element (@sorted) {
 593                my ($start, $end) = split_merge_info_range($element);
 594
 595                if ($last == -1) {
 596                        $first = $start;
 597                        $last = $end;
 598                        next;
 599                }
 600                if ($start <= $last+1) {
 601                        if ($end > $last) {
 602                                $last = $end;
 603                        }
 604                        next;
 605                }
 606                if ($first == $last) {
 607                        push @return, "$first";
 608                } else {
 609                        push @return, "$first-$last";
 610                }
 611                $first = $start;
 612                $last = $end;
 613        }
 614
 615        if ($first != -1) {
 616                if ($first == $last) {
 617                        push @return, "$first";
 618                } else {
 619                        push @return, "$first-$last";
 620                }
 621        }
 622
 623        return join(',', @return);
 624}
 625
 626sub merge_revs_into_hash {
 627        my ($hash, $minfo) = @_;
 628        my @lines = split(' ', $minfo);
 629
 630        for my $line (@lines) {
 631                my ($branchpath, $revs) = split(/:/, $line);
 632
 633                if (exists($hash->{$branchpath})) {
 634                        # Merge the two revision sets
 635                        my $combined = "$hash->{$branchpath},$revs";
 636                        $hash->{$branchpath} = combine_ranges($combined);
 637                } else {
 638                        # Just do range combining for consolidation
 639                        $hash->{$branchpath} = combine_ranges($revs);
 640                }
 641        }
 642}
 643
 644sub merge_merge_info {
 645        my ($mergeinfo_one, $mergeinfo_two) = @_;
 646        my %result_hash = ();
 647
 648        merge_revs_into_hash(\%result_hash, $mergeinfo_one);
 649        merge_revs_into_hash(\%result_hash, $mergeinfo_two);
 650
 651        my $result = '';
 652        # Sort below is for consistency's sake
 653        for my $branchname (sort keys(%result_hash)) {
 654                my $revlist = $result_hash{$branchname};
 655                $result .= "$branchname:$revlist\n"
 656        }
 657        return $result;
 658}
 659
 660sub populate_merge_info {
 661        my ($d, $gs, $uuid, $linear_refs, $rewritten_parent) = @_;
 662
 663        my %parentshash;
 664        read_commit_parents(\%parentshash, $d);
 665        my @parents = @{$parentshash{$d}};
 666        if ($#parents > 0) {
 667                # Merge commit
 668                my $all_parents_ok = 1;
 669                my $aggregate_mergeinfo = '';
 670                my $rooturl = $gs->repos_root;
 671
 672                if (defined($rewritten_parent)) {
 673                        # Replace first parent with newly-rewritten version
 674                        shift @parents;
 675                        unshift @parents, $rewritten_parent;
 676                }
 677
 678                foreach my $parent (@parents) {
 679                        my ($branchurl, $svnrev, $paruuid) =
 680                                cmt_metadata($parent);
 681
 682                        unless (defined($svnrev)) {
 683                                # Should have been caught be preflight check
 684                                fatal "merge commit $d has ancestor $parent, but that change "
 685                     ."does not have git-svn metadata!";
 686                        }
 687                        unless ($branchurl =~ /^$rooturl(.*)/) {
 688                                fatal "commit $parent git-svn metadata changed mid-run!";
 689                        }
 690                        my $branchpath = $1;
 691
 692                        my $ra = Git::SVN::Ra->new($branchurl);
 693                        my (undef, undef, $props) =
 694                                $ra->get_dir(canonicalize_path("."), $svnrev);
 695                        my $par_mergeinfo = $props->{'svn:mergeinfo'};
 696                        unless (defined $par_mergeinfo) {
 697                                $par_mergeinfo = '';
 698                        }
 699                        # Merge previous mergeinfo values
 700                        $aggregate_mergeinfo =
 701                                merge_merge_info($aggregate_mergeinfo,
 702                                                                 $par_mergeinfo, 0);
 703
 704                        next if $parent eq $parents[0]; # Skip first parent
 705                        # Add new changes being placed in tree by merge
 706                        my @cmd = (qw/rev-list --reverse/,
 707                                           $parent, qw/--not/);
 708                        foreach my $par (@parents) {
 709                                unless ($par eq $parent) {
 710                                        push @cmd, $par;
 711                                }
 712                        }
 713                        my @revsin = ();
 714                        my ($revlist, $ctx) = command_output_pipe(@cmd);
 715                        while (<$revlist>) {
 716                                my $irev = $_;
 717                                chomp $irev;
 718                                my (undef, $csvnrev, undef) =
 719                                        cmt_metadata($irev);
 720                                unless (defined $csvnrev) {
 721                                        # A child is missing SVN annotations...
 722                                        # this might be OK, or might not be.
 723                                        warn "W:child $irev is merged into revision "
 724                                                 ."$d but does not have git-svn metadata. "
 725                                                 ."This means git-svn cannot determine the "
 726                                                 ."svn revision numbers to place into the "
 727                                                 ."svn:mergeinfo property. You must ensure "
 728                                                 ."a branch is entirely committed to "
 729                                                 ."SVN before merging it in order for "
 730                                                 ."svn:mergeinfo population to function "
 731                                                 ."properly";
 732                                }
 733                                push @revsin, $csvnrev;
 734                        }
 735                        command_close_pipe($revlist, $ctx);
 736
 737                        last unless $all_parents_ok;
 738
 739                        # We now have a list of all SVN revnos which are
 740                        # merged by this particular parent. Integrate them.
 741                        next if $#revsin == -1;
 742                        my $newmergeinfo = "$branchpath:" . join(',', @revsin);
 743                        $aggregate_mergeinfo =
 744                                merge_merge_info($aggregate_mergeinfo,
 745                                                                 $newmergeinfo, 1);
 746                }
 747                if ($all_parents_ok and $aggregate_mergeinfo) {
 748                        return $aggregate_mergeinfo;
 749                }
 750        }
 751
 752        return undef;
 753}
 754
 755sub cmd_dcommit {
 756        my $head = shift;
 757        command_noisy(qw/update-index --refresh/);
 758        git_cmd_try { command_oneline(qw/diff-index --quiet HEAD/) }
 759                'Cannot dcommit with a dirty index.  Commit your changes first, '
 760                . "or stash them with `git stash'.\n";
 761        $head ||= 'HEAD';
 762
 763        my $old_head;
 764        if ($head ne 'HEAD') {
 765                $old_head = eval {
 766                        command_oneline([qw/symbolic-ref -q HEAD/])
 767                };
 768                if ($old_head) {
 769                        $old_head =~ s{^refs/heads/}{};
 770                } else {
 771                        $old_head = eval { command_oneline(qw/rev-parse HEAD/) };
 772                }
 773                command(['checkout', $head], STDERR => 0);
 774        }
 775
 776        my @refs;
 777        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD', \@refs);
 778        unless ($gs) {
 779                die "Unable to determine upstream SVN information from ",
 780                    "$head history.\nPerhaps the repository is empty.";
 781        }
 782
 783        if (defined $_commit_url) {
 784                $url = $_commit_url;
 785        } else {
 786                $url = eval { command_oneline('config', '--get',
 787                              "svn-remote.$gs->{repo_id}.commiturl") };
 788                if (!$url) {
 789                        $url = $gs->full_pushurl
 790                }
 791        }
 792
 793        my $last_rev = $_revision if defined $_revision;
 794        if ($url) {
 795                print "Committing to $url ...\n";
 796        }
 797        my ($linear_refs, $parents) = linearize_history($gs, \@refs);
 798        if ($_no_rebase && scalar(@$linear_refs) > 1) {
 799                warn "Attempting to commit more than one change while ",
 800                     "--no-rebase is enabled.\n",
 801                     "If these changes depend on each other, re-running ",
 802                     "without --no-rebase may be required."
 803        }
 804
 805        if (defined $_interactive){
 806                my $ask_default = "y";
 807                foreach my $d (@$linear_refs){
 808                        my ($fh, $ctx) = command_output_pipe(qw(show --summary), "$d");
 809                        while (<$fh>){
 810                                print $_;
 811                        }
 812                        command_close_pipe($fh, $ctx);
 813                        $_ = ask("Commit this patch to SVN? ([y]es (default)|[n]o|[q]uit|[a]ll): ",
 814                                 valid_re => qr/^(?:yes|y|no|n|quit|q|all|a)/i,
 815                                 default => $ask_default);
 816                        die "Commit this patch reply required" unless defined $_;
 817                        if (/^[nq]/i) {
 818                                exit(0);
 819                        } elsif (/^a/i) {
 820                                last;
 821                        }
 822                }
 823        }
 824
 825        my $expect_url = $url;
 826
 827        my $push_merge_info = eval {
 828                command_oneline(qw/config --get svn.pushmergeinfo/)
 829                };
 830        if (not defined($push_merge_info)
 831                        or $push_merge_info eq "false"
 832                        or $push_merge_info eq "no"
 833                        or $push_merge_info eq "never") {
 834                $push_merge_info = 0;
 835        }
 836
 837        unless (defined($_merge_info) || ! $push_merge_info) {
 838                # Preflight check of changes to ensure no issues with mergeinfo
 839                # This includes check for uncommitted-to-SVN parents
 840                # (other than the first parent, which we will handle),
 841                # information from different SVN repos, and paths
 842                # which are not underneath this repository root.
 843                my $rooturl = $gs->repos_root;
 844                foreach my $d (@$linear_refs) {
 845                        my %parentshash;
 846                        read_commit_parents(\%parentshash, $d);
 847                        my @realparents = @{$parentshash{$d}};
 848                        if ($#realparents > 0) {
 849                                # Merge commit
 850                                shift @realparents; # Remove/ignore first parent
 851                                foreach my $parent (@realparents) {
 852                                        my ($branchurl, $svnrev, $paruuid) = cmt_metadata($parent);
 853                                        unless (defined $paruuid) {
 854                                                # A parent is missing SVN annotations...
 855                                                # abort the whole operation.
 856                                                fatal "$parent is merged into revision $d, "
 857                                                         ."but does not have git-svn metadata. "
 858                                                         ."Either dcommit the branch or use a "
 859                                                         ."local cherry-pick, FF merge, or rebase "
 860                                                         ."instead of an explicit merge commit.";
 861                                        }
 862
 863                                        unless ($paruuid eq $uuid) {
 864                                                # Parent has SVN metadata from different repository
 865                                                fatal "merge parent $parent for change $d has "
 866                                                         ."git-svn uuid $paruuid, while current change "
 867                                                         ."has uuid $uuid!";
 868                                        }
 869
 870                                        unless ($branchurl =~ /^$rooturl(.*)/) {
 871                                                # This branch is very strange indeed.
 872                                                fatal "merge parent $parent for $d is on branch "
 873                                                         ."$branchurl, which is not under the "
 874                                                         ."git-svn root $rooturl!";
 875                                        }
 876                                }
 877                        }
 878                }
 879        }
 880
 881        my $rewritten_parent;
 882        Git::SVN::remove_username($expect_url);
 883        if (defined($_merge_info)) {
 884                $_merge_info =~ tr{ }{\n};
 885        }
 886        while (1) {
 887                my $d = shift @$linear_refs or last;
 888                unless (defined $last_rev) {
 889                        (undef, $last_rev, undef) = cmt_metadata("$d~1");
 890                        unless (defined $last_rev) {
 891                                fatal "Unable to extract revision information ",
 892                                      "from commit $d~1";
 893                        }
 894                }
 895                if ($_dry_run) {
 896                        print "diff-tree $d~1 $d\n";
 897                } else {
 898                        my $cmt_rev;
 899
 900                        unless (defined($_merge_info) || ! $push_merge_info) {
 901                                $_merge_info = populate_merge_info($d, $gs,
 902                                                             $uuid,
 903                                                             $linear_refs,
 904                                                             $rewritten_parent);
 905                        }
 906
 907                        my %ed_opts = ( r => $last_rev,
 908                                        log => get_commit_entry($d)->{log},
 909                                        ra => Git::SVN::Ra->new($url),
 910                                        config => SVN::Core::config_get_config(
 911                                                $Git::SVN::Ra::config_dir
 912                                        ),
 913                                        tree_a => "$d~1",
 914                                        tree_b => $d,
 915                                        editor_cb => sub {
 916                                               print "Committed r$_[0]\n";
 917                                               $cmt_rev = $_[0];
 918                                        },
 919                                        mergeinfo => $_merge_info,
 920                                        svn_path => '');
 921                        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
 922                                print "No changes\n$d~1 == $d\n";
 923                        } elsif ($parents->{$d} && @{$parents->{$d}}) {
 924                                $gs->{inject_parents_dcommit}->{$cmt_rev} =
 925                                                               $parents->{$d};
 926                        }
 927                        $_fetch_all ? $gs->fetch_all : $gs->fetch;
 928                        $last_rev = $cmt_rev;
 929                        next if $_no_rebase;
 930
 931                        # we always want to rebase against the current HEAD,
 932                        # not any head that was passed to us
 933                        my @diff = command('diff-tree', $d,
 934                                           $gs->refname, '--');
 935                        my @finish;
 936                        if (@diff) {
 937                                @finish = rebase_cmd();
 938                                print STDERR "W: $d and ", $gs->refname,
 939                                             " differ, using @finish:\n",
 940                                             join("\n", @diff), "\n";
 941                        } else {
 942                                print "No changes between current HEAD and ",
 943                                      $gs->refname,
 944                                      "\nResetting to the latest ",
 945                                      $gs->refname, "\n";
 946                                @finish = qw/reset --mixed/;
 947                        }
 948                        command_noisy(@finish, $gs->refname);
 949
 950                        $rewritten_parent = command_oneline(qw/rev-parse HEAD/);
 951
 952                        if (@diff) {
 953                                @refs = ();
 954                                my ($url_, $rev_, $uuid_, $gs_) =
 955                                              working_head_info('HEAD', \@refs);
 956                                my ($linear_refs_, $parents_) =
 957                                              linearize_history($gs_, \@refs);
 958                                if (scalar(@$linear_refs) !=
 959                                    scalar(@$linear_refs_)) {
 960                                        fatal "# of revisions changed ",
 961                                          "\nbefore:\n",
 962                                          join("\n", @$linear_refs),
 963                                          "\n\nafter:\n",
 964                                          join("\n", @$linear_refs_), "\n",
 965                                          'If you are attempting to commit ',
 966                                          "merges, try running:\n\t",
 967                                          'git rebase --interactive',
 968                                          '--preserve-merges ',
 969                                          $gs->refname,
 970                                          "\nBefore dcommitting";
 971                                }
 972                                if ($url_ ne $expect_url) {
 973                                        if ($url_ eq $gs->metadata_url) {
 974                                                print
 975                                                  "Accepting rewritten URL:",
 976                                                  " $url_\n";
 977                                        } else {
 978                                                fatal
 979                                                  "URL mismatch after rebase:",
 980                                                  " $url_ != $expect_url";
 981                                        }
 982                                }
 983                                if ($uuid_ ne $uuid) {
 984                                        fatal "uuid mismatch after rebase: ",
 985                                              "$uuid_ != $uuid";
 986                                }
 987                                # remap parents
 988                                my (%p, @l, $i);
 989                                for ($i = 0; $i < scalar @$linear_refs; $i++) {
 990                                        my $new = $linear_refs_->[$i] or next;
 991                                        $p{$new} =
 992                                                $parents->{$linear_refs->[$i]};
 993                                        push @l, $new;
 994                                }
 995                                $parents = \%p;
 996                                $linear_refs = \@l;
 997                        }
 998                }
 999        }
1000
1001        if ($old_head) {
1002                my $new_head = command_oneline(qw/rev-parse HEAD/);
1003                my $new_is_symbolic = eval {
1004                        command_oneline(qw/symbolic-ref -q HEAD/);
1005                };
1006                if ($new_is_symbolic) {
1007                        print "dcommitted the branch ", $head, "\n";
1008                } else {
1009                        print "dcommitted on a detached HEAD because you gave ",
1010                              "a revision argument.\n",
1011                              "The rewritten commit is: ", $new_head, "\n";
1012                }
1013                command(['checkout', $old_head], STDERR => 0);
1014        }
1015
1016        unlink $gs->{index};
1017}
1018
1019sub cmd_branch {
1020        my ($branch_name, $head) = @_;
1021
1022        unless (defined $branch_name && length $branch_name) {
1023                die(($_tag ? "tag" : "branch") . " name required\n");
1024        }
1025        $head ||= 'HEAD';
1026
1027        my (undef, $rev, undef, $gs) = working_head_info($head);
1028        my $src = $gs->full_pushurl;
1029
1030        my $remote = Git::SVN::read_all_remotes()->{$gs->{repo_id}};
1031        my $allglobs = $remote->{ $_tag ? 'tags' : 'branches' };
1032        my $glob;
1033        if ($#{$allglobs} == 0) {
1034                $glob = $allglobs->[0];
1035        } else {
1036                unless(defined $_branch_dest) {
1037                        die "Multiple ",
1038                            $_tag ? "tag" : "branch",
1039                            " paths defined for Subversion repository.\n",
1040                            "You must specify where you want to create the ",
1041                            $_tag ? "tag" : "branch",
1042                            " with the --destination argument.\n";
1043                }
1044                foreach my $g (@{$allglobs}) {
1045                        # SVN::Git::Editor could probably be moved to Git.pm..
1046                        my $re = SVN::Git::Editor::glob2pat($g->{path}->{left});
1047                        if ($_branch_dest =~ /$re/) {
1048                                $glob = $g;
1049                                last;
1050                        }
1051                }
1052                unless (defined $glob) {
1053                        my $dest_re = qr/\b\Q$_branch_dest\E\b/;
1054                        foreach my $g (@{$allglobs}) {
1055                                $g->{path}->{left} =~ /$dest_re/ or next;
1056                                if (defined $glob) {
1057                                        die "Ambiguous destination: ",
1058                                            $_branch_dest, "\nmatches both '",
1059                                            $glob->{path}->{left}, "' and '",
1060                                            $g->{path}->{left}, "'\n";
1061                                }
1062                                $glob = $g;
1063                        }
1064                        unless (defined $glob) {
1065                                die "Unknown ",
1066                                    $_tag ? "tag" : "branch",
1067                                    " destination $_branch_dest\n";
1068                        }
1069                }
1070        }
1071        my ($lft, $rgt) = @{ $glob->{path} }{qw/left right/};
1072        my $url;
1073        if (defined $_commit_url) {
1074                $url = $_commit_url;
1075        } else {
1076                $url = eval { command_oneline('config', '--get',
1077                        "svn-remote.$gs->{repo_id}.commiturl") };
1078                if (!$url) {
1079                        $url = $remote->{pushurl} || $remote->{url};
1080                }
1081        }
1082        my $dst = join '/', $url, $lft, $branch_name, ($rgt || ());
1083
1084        if ($dst =~ /^https:/ && $src =~ /^http:/) {
1085                $src=~s/^http:/https:/;
1086        }
1087
1088        ::_req_svn();
1089
1090        my $ctx = SVN::Client->new(
1091                auth    => Git::SVN::Ra::_auth_providers(),
1092                log_msg => sub {
1093                        ${ $_[0] } = defined $_message
1094                                ? $_message
1095                                : 'Create ' . ($_tag ? 'tag ' : 'branch ' )
1096                                . $branch_name;
1097                },
1098        );
1099
1100        eval {
1101                $ctx->ls($dst, 'HEAD', 0);
1102        } and die "branch ${branch_name} already exists\n";
1103
1104        print "Copying ${src} at r${rev} to ${dst}...\n";
1105        $ctx->copy($src, $rev, $dst)
1106                unless $_dry_run;
1107
1108        $gs->fetch_all;
1109}
1110
1111sub cmd_find_rev {
1112        my $revision_or_hash = shift or die "SVN or git revision required ",
1113                                            "as a command-line argument\n";
1114        my $result;
1115        if ($revision_or_hash =~ /^r\d+$/) {
1116                my $head = shift;
1117                $head ||= 'HEAD';
1118                my @refs;
1119                my (undef, undef, $uuid, $gs) = working_head_info($head, \@refs);
1120                unless ($gs) {
1121                        die "Unable to determine upstream SVN information from ",
1122                            "$head history\n";
1123                }
1124                my $desired_revision = substr($revision_or_hash, 1);
1125                $result = $gs->rev_map_get($desired_revision, $uuid);
1126        } else {
1127                my (undef, $rev, undef) = cmt_metadata($revision_or_hash);
1128                $result = $rev;
1129        }
1130        print "$result\n" if $result;
1131}
1132
1133sub auto_create_empty_directories {
1134        my ($gs) = @_;
1135        my $var = eval { command_oneline('config', '--get', '--bool',
1136                                         "svn-remote.$gs->{repo_id}.automkdirs") };
1137        # By default, create empty directories by consulting the unhandled log,
1138        # but allow setting it to 'false' to skip it.
1139        return !($var && $var eq 'false');
1140}
1141
1142sub cmd_rebase {
1143        command_noisy(qw/update-index --refresh/);
1144        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1145        unless ($gs) {
1146                die "Unable to determine upstream SVN information from ",
1147                    "working tree history\n";
1148        }
1149        if ($_dry_run) {
1150                print "Remote Branch: " . $gs->refname . "\n";
1151                print "SVN URL: " . $url . "\n";
1152                return;
1153        }
1154        if (command(qw/diff-index HEAD --/)) {
1155                print STDERR "Cannot rebase with uncommited changes:\n";
1156                command_noisy('status');
1157                exit 1;
1158        }
1159        unless ($_local) {
1160                # rebase will checkout for us, so no need to do it explicitly
1161                $_no_checkout = 'true';
1162                $_fetch_all ? $gs->fetch_all : $gs->fetch;
1163        }
1164        command_noisy(rebase_cmd(), $gs->refname);
1165        if (auto_create_empty_directories($gs)) {
1166                $gs->mkemptydirs;
1167        }
1168}
1169
1170sub cmd_show_ignore {
1171        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1172        $gs ||= Git::SVN->new;
1173        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
1174        $gs->prop_walk($gs->{path}, $r, sub {
1175                my ($gs, $path, $props) = @_;
1176                print STDOUT "\n# $path\n";
1177                my $s = $props->{'svn:ignore'} or return;
1178                $s =~ s/[\r\n]+/\n/g;
1179                $s =~ s/^\n+//;
1180                chomp $s;
1181                $s =~ s#^#$path#gm;
1182                print STDOUT "$s\n";
1183        });
1184}
1185
1186sub cmd_show_externals {
1187        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1188        $gs ||= Git::SVN->new;
1189        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
1190        $gs->prop_walk($gs->{path}, $r, sub {
1191                my ($gs, $path, $props) = @_;
1192                print STDOUT "\n# $path\n";
1193                my $s = $props->{'svn:externals'} or return;
1194                $s =~ s/[\r\n]+/\n/g;
1195                chomp $s;
1196                $s =~ s#^#$path#gm;
1197                print STDOUT "$s\n";
1198        });
1199}
1200
1201sub cmd_create_ignore {
1202        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1203        $gs ||= Git::SVN->new;
1204        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
1205        $gs->prop_walk($gs->{path}, $r, sub {
1206                my ($gs, $path, $props) = @_;
1207                # $path is of the form /path/to/dir/
1208                $path = '.' . $path;
1209                # SVN can have attributes on empty directories,
1210                # which git won't track
1211                mkpath([$path]) unless -d $path;
1212                my $ignore = $path . '.gitignore';
1213                my $s = $props->{'svn:ignore'} or return;
1214                open(GITIGNORE, '>', $ignore)
1215                  or fatal("Failed to open `$ignore' for writing: $!");
1216                $s =~ s/[\r\n]+/\n/g;
1217                $s =~ s/^\n+//;
1218                chomp $s;
1219                # Prefix all patterns so that the ignore doesn't apply
1220                # to sub-directories.
1221                $s =~ s#^#/#gm;
1222                print GITIGNORE "$s\n";
1223                close(GITIGNORE)
1224                  or fatal("Failed to close `$ignore': $!");
1225                command_noisy('add', '-f', $ignore);
1226        });
1227}
1228
1229sub cmd_mkdirs {
1230        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1231        $gs ||= Git::SVN->new;
1232        $gs->mkemptydirs($_revision);
1233}
1234
1235sub canonicalize_path {
1236        my ($path) = @_;
1237        my $dot_slash_added = 0;
1238        if (substr($path, 0, 1) ne "/") {
1239                $path = "./" . $path;
1240                $dot_slash_added = 1;
1241        }
1242        # File::Spec->canonpath doesn't collapse x/../y into y (for a
1243        # good reason), so let's do this manually.
1244        $path =~ s#/+#/#g;
1245        $path =~ s#/\.(?:/|$)#/#g;
1246        $path =~ s#/[^/]+/\.\.##g;
1247        $path =~ s#/$##g;
1248        $path =~ s#^\./## if $dot_slash_added;
1249        $path =~ s#^/##;
1250        $path =~ s#^\.$##;
1251        return $path;
1252}
1253
1254sub canonicalize_url {
1255        my ($url) = @_;
1256        $url =~ s#^([^:]+://[^/]*/)(.*)$#$1 . canonicalize_path($2)#e;
1257        return $url;
1258}
1259
1260# get_svnprops(PATH)
1261# ------------------
1262# Helper for cmd_propget and cmd_proplist below.
1263sub get_svnprops {
1264        my $path = shift;
1265        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1266        $gs ||= Git::SVN->new;
1267
1268        # prefix THE PATH by the sub-directory from which the user
1269        # invoked us.
1270        $path = $cmd_dir_prefix . $path;
1271        fatal("No such file or directory: $path") unless -e $path;
1272        my $is_dir = -d $path ? 1 : 0;
1273        $path = $gs->{path} . '/' . $path;
1274
1275        # canonicalize the path (otherwise libsvn will abort or fail to
1276        # find the file)
1277        $path = canonicalize_path($path);
1278
1279        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
1280        my $props;
1281        if ($is_dir) {
1282                (undef, undef, $props) = $gs->ra->get_dir($path, $r);
1283        }
1284        else {
1285                (undef, $props) = $gs->ra->get_file($path, $r, undef);
1286        }
1287        return $props;
1288}
1289
1290# cmd_propget (PROP, PATH)
1291# ------------------------
1292# Print the SVN property PROP for PATH.
1293sub cmd_propget {
1294        my ($prop, $path) = @_;
1295        $path = '.' if not defined $path;
1296        usage(1) if not defined $prop;
1297        my $props = get_svnprops($path);
1298        if (not defined $props->{$prop}) {
1299                fatal("`$path' does not have a `$prop' SVN property.");
1300        }
1301        print $props->{$prop} . "\n";
1302}
1303
1304# cmd_proplist (PATH)
1305# -------------------
1306# Print the list of SVN properties for PATH.
1307sub cmd_proplist {
1308        my $path = shift;
1309        $path = '.' if not defined $path;
1310        my $props = get_svnprops($path);
1311        print "Properties on '$path':\n";
1312        foreach (sort keys %{$props}) {
1313                print "  $_\n";
1314        }
1315}
1316
1317sub cmd_multi_init {
1318        my $url = shift;
1319        unless (defined $_trunk || @_branches || @_tags) {
1320                usage(1);
1321        }
1322
1323        $_prefix = '' unless defined $_prefix;
1324        if (defined $url) {
1325                $url = canonicalize_url($url);
1326                init_subdir(@_);
1327        }
1328        do_git_init_db();
1329        if (defined $_trunk) {
1330                $_trunk =~ s#^/+##;
1331                my $trunk_ref = 'refs/remotes/' . $_prefix . 'trunk';
1332                # try both old-style and new-style lookups:
1333                my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
1334                unless ($gs_trunk) {
1335                        my ($trunk_url, $trunk_path) =
1336                                              complete_svn_url($url, $_trunk);
1337                        $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
1338                                                   undef, $trunk_ref);
1339                }
1340        }
1341        return unless @_branches || @_tags;
1342        my $ra = $url ? Git::SVN::Ra->new($url) : undef;
1343        foreach my $path (@_branches) {
1344                complete_url_ls_init($ra, $path, '--branches/-b', $_prefix);
1345        }
1346        foreach my $path (@_tags) {
1347                complete_url_ls_init($ra, $path, '--tags/-t', $_prefix.'tags/');
1348        }
1349}
1350
1351sub cmd_multi_fetch {
1352        $Git::SVN::no_reuse_existing = undef;
1353        my $remotes = Git::SVN::read_all_remotes();
1354        foreach my $repo_id (sort keys %$remotes) {
1355                if ($remotes->{$repo_id}->{url}) {
1356                        Git::SVN::fetch_all($repo_id, $remotes);
1357                }
1358        }
1359}
1360
1361# this command is special because it requires no metadata
1362sub cmd_commit_diff {
1363        my ($ta, $tb, $url) = @_;
1364        my $usage = "Usage: $0 commit-diff -r<revision> ".
1365                    "<tree-ish> <tree-ish> [<URL>]";
1366        fatal($usage) if (!defined $ta || !defined $tb);
1367        my $svn_path = '';
1368        if (!defined $url) {
1369                my $gs = eval { Git::SVN->new };
1370                if (!$gs) {
1371                        fatal("Needed URL or usable git-svn --id in ",
1372                              "the command-line\n", $usage);
1373                }
1374                $url = $gs->{url};
1375                $svn_path = $gs->{path};
1376        }
1377        unless (defined $_revision) {
1378                fatal("-r|--revision is a required argument\n", $usage);
1379        }
1380        if (defined $_message && defined $_file) {
1381                fatal("Both --message/-m and --file/-F specified ",
1382                      "for the commit message.\n",
1383                      "I have no idea what you mean");
1384        }
1385        if (defined $_file) {
1386                $_message = file_to_s($_file);
1387        } else {
1388                $_message ||= get_commit_entry($tb)->{log};
1389        }
1390        my $ra ||= Git::SVN::Ra->new($url);
1391        my $r = $_revision;
1392        if ($r eq 'HEAD') {
1393                $r = $ra->get_latest_revnum;
1394        } elsif ($r !~ /^\d+$/) {
1395                die "revision argument: $r not understood by git-svn\n";
1396        }
1397        my %ed_opts = ( r => $r,
1398                        log => $_message,
1399                        ra => $ra,
1400                        tree_a => $ta,
1401                        tree_b => $tb,
1402                        editor_cb => sub { print "Committed r$_[0]\n" },
1403                        svn_path => $svn_path );
1404        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
1405                print "No changes\n$ta == $tb\n";
1406        }
1407}
1408
1409sub escape_uri_only {
1410        my ($uri) = @_;
1411        my @tmp;
1412        foreach (split m{/}, $uri) {
1413                s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
1414                push @tmp, $_;
1415        }
1416        join('/', @tmp);
1417}
1418
1419sub escape_url {
1420        my ($url) = @_;
1421        if ($url =~ m#^([^:]+)://([^/]*)(.*)$#) {
1422                my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
1423                $url = "$scheme://$domain$uri";
1424        }
1425        $url;
1426}
1427
1428sub cmd_info {
1429        my $path = canonicalize_path(defined($_[0]) ? $_[0] : ".");
1430        my $fullpath = canonicalize_path($cmd_dir_prefix . $path);
1431        if (exists $_[1]) {
1432                die "Too many arguments specified\n";
1433        }
1434
1435        my ($file_type, $diff_status) = find_file_type_and_diff_status($path);
1436
1437        if (!$file_type && !$diff_status) {
1438                print STDERR "svn: '$path' is not under version control\n";
1439                exit 1;
1440        }
1441
1442        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1443        unless ($gs) {
1444                die "Unable to determine upstream SVN information from ",
1445                    "working tree history\n";
1446        }
1447
1448        # canonicalize_path() will return "" to make libsvn 1.5.x happy,
1449        $path = "." if $path eq "";
1450
1451        my $full_url = $url . ($fullpath eq "" ? "" : "/$fullpath");
1452
1453        if ($_url) {
1454                print escape_url($full_url), "\n";
1455                return;
1456        }
1457
1458        my $result = "Path: $path\n";
1459        $result .= "Name: " . basename($path) . "\n" if $file_type ne "dir";
1460        $result .= "URL: " . escape_url($full_url) . "\n";
1461
1462        eval {
1463                my $repos_root = $gs->repos_root;
1464                Git::SVN::remove_username($repos_root);
1465                $result .= "Repository Root: " . escape_url($repos_root) . "\n";
1466        };
1467        if ($@) {
1468                $result .= "Repository Root: (offline)\n";
1469        }
1470        ::_req_svn();
1471        $result .= "Repository UUID: $uuid\n" unless $diff_status eq "A" &&
1472                ($SVN::Core::VERSION le '1.5.4' || $file_type ne "dir");
1473        $result .= "Revision: " . ($diff_status eq "A" ? 0 : $rev) . "\n";
1474
1475        $result .= "Node Kind: " .
1476                   ($file_type eq "dir" ? "directory" : "file") . "\n";
1477
1478        my $schedule = $diff_status eq "A"
1479                       ? "add"
1480                       : ($diff_status eq "D" ? "delete" : "normal");
1481        $result .= "Schedule: $schedule\n";
1482
1483        if ($diff_status eq "A") {
1484                print $result, "\n";
1485                return;
1486        }
1487
1488        my ($lc_author, $lc_rev, $lc_date_utc);
1489        my @args = Git::SVN::Log::git_svn_log_cmd($rev, $rev, "--", $fullpath);
1490        my $log = command_output_pipe(@args);
1491        my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
1492        while (<$log>) {
1493                if (/^${esc_color}author (.+) <[^>]+> (\d+) ([\-\+]?\d+)$/o) {
1494                        $lc_author = $1;
1495                        $lc_date_utc = Git::SVN::Log::parse_git_date($2, $3);
1496                } elsif (/^${esc_color}    (git-svn-id:.+)$/o) {
1497                        (undef, $lc_rev, undef) = ::extract_metadata($1);
1498                }
1499        }
1500        close $log;
1501
1502        Git::SVN::Log::set_local_timezone();
1503
1504        $result .= "Last Changed Author: $lc_author\n";
1505        $result .= "Last Changed Rev: $lc_rev\n";
1506        $result .= "Last Changed Date: " .
1507                   Git::SVN::Log::format_svn_date($lc_date_utc) . "\n";
1508
1509        if ($file_type ne "dir") {
1510                my $text_last_updated_date =
1511                    ($diff_status eq "D" ? $lc_date_utc : (stat $path)[9]);
1512                $result .=
1513                    "Text Last Updated: " .
1514                    Git::SVN::Log::format_svn_date($text_last_updated_date) .
1515                    "\n";
1516                my $checksum;
1517                if ($diff_status eq "D") {
1518                        my ($fh, $ctx) =
1519                            command_output_pipe(qw(cat-file blob), "HEAD:$path");
1520                        if ($file_type eq "link") {
1521                                my $file_name = <$fh>;
1522                                $checksum = md5sum("link $file_name");
1523                        } else {
1524                                $checksum = md5sum($fh);
1525                        }
1526                        command_close_pipe($fh, $ctx);
1527                } elsif ($file_type eq "link") {
1528                        my $file_name =
1529                            command(qw(cat-file blob), "HEAD:$path");
1530                        $checksum =
1531                            md5sum("link " . $file_name);
1532                } else {
1533                        open FILE, "<", $path or die $!;
1534                        $checksum = md5sum(\*FILE);
1535                        close FILE or die $!;
1536                }
1537                $result .= "Checksum: " . $checksum . "\n";
1538        }
1539
1540        print $result, "\n";
1541}
1542
1543sub cmd_reset {
1544        my $target = shift || $_revision or die "SVN revision required\n";
1545        $target = $1 if $target =~ /^r(\d+)$/;
1546        $target =~ /^\d+$/ or die "Numeric SVN revision expected\n";
1547        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1548        unless ($gs) {
1549                die "Unable to determine upstream SVN information from ".
1550                    "history\n";
1551        }
1552        my ($r, $c) = $gs->find_rev_before($target, not $_fetch_parent);
1553        die "Cannot find SVN revision $target\n" unless defined($c);
1554        $gs->rev_map_set($r, $c, 'reset', $uuid);
1555        print "r$r = $c ($gs->{ref_id})\n";
1556}
1557
1558sub cmd_gc {
1559        if (!$can_compress) {
1560                warn "Compress::Zlib could not be found; unhandled.log " .
1561                     "files will not be compressed.\n";
1562        }
1563        find({ wanted => \&gc_directory, no_chdir => 1}, "$ENV{GIT_DIR}/svn");
1564}
1565
1566########################### utility functions #########################
1567
1568sub rebase_cmd {
1569        my @cmd = qw/rebase/;
1570        push @cmd, '-v' if $_verbose;
1571        push @cmd, qw/--merge/ if $_merge;
1572        push @cmd, "--strategy=$_strategy" if $_strategy;
1573        @cmd;
1574}
1575
1576sub post_fetch_checkout {
1577        return if $_no_checkout;
1578        my $gs = $Git::SVN::_head or return;
1579        return if verify_ref('refs/heads/master^0');
1580
1581        # look for "trunk" ref if it exists
1582        my $remote = Git::SVN::read_all_remotes()->{$gs->{repo_id}};
1583        my $fetch = $remote->{fetch};
1584        if ($fetch) {
1585                foreach my $p (keys %$fetch) {
1586                        basename($fetch->{$p}) eq 'trunk' or next;
1587                        $gs = Git::SVN->new($fetch->{$p}, $gs->{repo_id}, $p);
1588                        last;
1589                }
1590        }
1591
1592        my $valid_head = verify_ref('HEAD^0');
1593        command_noisy(qw(update-ref refs/heads/master), $gs->refname);
1594        return if ($valid_head || !verify_ref('HEAD^0'));
1595
1596        return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
1597        my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
1598        return if -f $index;
1599
1600        return if command_oneline(qw/rev-parse --is-inside-work-tree/) eq 'false';
1601        return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
1602        command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
1603        print STDERR "Checked out HEAD:\n  ",
1604                     $gs->full_url, " r", $gs->last_rev, "\n";
1605        if (auto_create_empty_directories($gs)) {
1606                $gs->mkemptydirs($gs->last_rev);
1607        }
1608}
1609
1610sub complete_svn_url {
1611        my ($url, $path) = @_;
1612        $path =~ s#/+$##;
1613        if ($path !~ m#^[a-z\+]+://#) {
1614                if (!defined $url || $url !~ m#^[a-z\+]+://#) {
1615                        fatal("E: '$path' is not a complete URL ",
1616                              "and a separate URL is not specified");
1617                }
1618                return ($url, $path);
1619        }
1620        return ($path, '');
1621}
1622
1623sub complete_url_ls_init {
1624        my ($ra, $repo_path, $switch, $pfx) = @_;
1625        unless ($repo_path) {
1626                print STDERR "W: $switch not specified\n";
1627                return;
1628        }
1629        $repo_path =~ s#/+$##;
1630        if ($repo_path =~ m#^[a-z\+]+://#) {
1631                $ra = Git::SVN::Ra->new($repo_path);
1632                $repo_path = '';
1633        } else {
1634                $repo_path =~ s#^/+##;
1635                unless ($ra) {
1636                        fatal("E: '$repo_path' is not a complete URL ",
1637                              "and a separate URL is not specified");
1638                }
1639        }
1640        my $url = $ra->{url};
1641        my $gs = Git::SVN->init($url, undef, undef, undef, 1);
1642        my $k = "svn-remote.$gs->{repo_id}.url";
1643        my $orig_url = eval { command_oneline(qw/config --get/, $k) };
1644        if ($orig_url && ($orig_url ne $gs->{url})) {
1645                die "$k already set: $orig_url\n",
1646                    "wanted to set to: $gs->{url}\n";
1647        }
1648        command_oneline('config', $k, $gs->{url}) unless $orig_url;
1649        my $remote_path = "$gs->{path}/$repo_path";
1650        $remote_path =~ s{%([0-9A-F]{2})}{chr hex($1)}ieg;
1651        $remote_path =~ s#/+#/#g;
1652        $remote_path =~ s#^/##g;
1653        $remote_path .= "/*" if $remote_path !~ /\*/;
1654        my ($n) = ($switch =~ /^--(\w+)/);
1655        if (length $pfx && $pfx !~ m#/$#) {
1656                die "--prefix='$pfx' must have a trailing slash '/'\n";
1657        }
1658        command_noisy('config',
1659                      '--add',
1660                      "svn-remote.$gs->{repo_id}.$n",
1661                      "$remote_path:refs/remotes/$pfx*" .
1662                        ('/*' x (($remote_path =~ tr/*/*/) - 1)) );
1663}
1664
1665sub verify_ref {
1666        my ($ref) = @_;
1667        eval { command_oneline([ 'rev-parse', '--verify', $ref ],
1668                               { STDERR => 0 }); };
1669}
1670
1671sub get_tree_from_treeish {
1672        my ($treeish) = @_;
1673        # $treeish can be a symbolic ref, too:
1674        my $type = command_oneline(qw/cat-file -t/, $treeish);
1675        my $expected;
1676        while ($type eq 'tag') {
1677                ($treeish, $type) = command(qw/cat-file tag/, $treeish);
1678        }
1679        if ($type eq 'commit') {
1680                $expected = (grep /^tree /, command(qw/cat-file commit/,
1681                                                    $treeish))[0];
1682                ($expected) = ($expected =~ /^tree ($sha1)$/o);
1683                die "Unable to get tree from $treeish\n" unless $expected;
1684        } elsif ($type eq 'tree') {
1685                $expected = $treeish;
1686        } else {
1687                die "$treeish is a $type, expected tree, tag or commit\n";
1688        }
1689        return $expected;
1690}
1691
1692sub get_commit_entry {
1693        my ($treeish) = shift;
1694        my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
1695        my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
1696        my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
1697        open my $log_fh, '>', $commit_editmsg or croak $!;
1698
1699        my $type = command_oneline(qw/cat-file -t/, $treeish);
1700        if ($type eq 'commit' || $type eq 'tag') {
1701                my ($msg_fh, $ctx) = command_output_pipe('cat-file',
1702                                                         $type, $treeish);
1703                my $in_msg = 0;
1704                my $author;
1705                my $saw_from = 0;
1706                my $msgbuf = "";
1707                while (<$msg_fh>) {
1708                        if (!$in_msg) {
1709                                $in_msg = 1 if (/^\s*$/);
1710                                $author = $1 if (/^author (.*>)/);
1711                        } elsif (/^git-svn-id: /) {
1712                                # skip this for now, we regenerate the
1713                                # correct one on re-fetch anyways
1714                                # TODO: set *:merge properties or like...
1715                        } else {
1716                                if (/^From:/ || /^Signed-off-by:/) {
1717                                        $saw_from = 1;
1718                                }
1719                                $msgbuf .= $_;
1720                        }
1721                }
1722                $msgbuf =~ s/\s+$//s;
1723                if ($Git::SVN::_add_author_from && defined($author)
1724                    && !$saw_from) {
1725                        $msgbuf .= "\n\nFrom: $author";
1726                }
1727                print $log_fh $msgbuf or croak $!;
1728                command_close_pipe($msg_fh, $ctx);
1729        }
1730        close $log_fh or croak $!;
1731
1732        if ($_edit || ($type eq 'tree')) {
1733                chomp(my $editor = command_oneline(qw(var GIT_EDITOR)));
1734                system('sh', '-c', $editor.' "$@"', $editor, $commit_editmsg);
1735        }
1736        rename $commit_editmsg, $commit_msg or croak $!;
1737        {
1738                require Encode;
1739                # SVN requires messages to be UTF-8 when entering the repo
1740                local $/;
1741                open $log_fh, '<', $commit_msg or croak $!;
1742                binmode $log_fh;
1743                chomp($log_entry{log} = <$log_fh>);
1744
1745                my $enc = Git::config('i18n.commitencoding') || 'UTF-8';
1746                my $msg = $log_entry{log};
1747
1748                eval { $msg = Encode::decode($enc, $msg, 1) };
1749                if ($@) {
1750                        die "Could not decode as $enc:\n", $msg,
1751                            "\nPerhaps you need to set i18n.commitencoding\n";
1752                }
1753
1754                eval { $msg = Encode::encode('UTF-8', $msg, 1) };
1755                die "Could not encode as UTF-8:\n$msg\n" if $@;
1756
1757                $log_entry{log} = $msg;
1758
1759                close $log_fh or croak $!;
1760        }
1761        unlink $commit_msg;
1762        \%log_entry;
1763}
1764
1765sub s_to_file {
1766        my ($str, $file, $mode) = @_;
1767        open my $fd,'>',$file or croak $!;
1768        print $fd $str,"\n" or croak $!;
1769        close $fd or croak $!;
1770        chmod ($mode &~ umask, $file) if (defined $mode);
1771}
1772
1773sub file_to_s {
1774        my $file = shift;
1775        open my $fd,'<',$file or croak "$!: file: $file\n";
1776        local $/;
1777        my $ret = <$fd>;
1778        close $fd or croak $!;
1779        $ret =~ s/\s*$//s;
1780        return $ret;
1781}
1782
1783# '<svn username> = real-name <email address>' mapping based on git-svnimport:
1784sub load_authors {
1785        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1786        my $log = $cmd eq 'log';
1787        while (<$authors>) {
1788                chomp;
1789                next unless /^(.+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
1790                my ($user, $name, $email) = ($1, $2, $3);
1791                if ($log) {
1792                        $Git::SVN::Log::rusers{"$name <$email>"} = $user;
1793                } else {
1794                        $users{$user} = [$name, $email];
1795                }
1796        }
1797        close $authors or croak $!;
1798}
1799
1800# convert GetOpt::Long specs for use by git-config
1801sub read_git_config {
1802        my $opts = shift;
1803        my @config_only;
1804        foreach my $o (keys %$opts) {
1805                # if we have mixedCase and a long option-only, then
1806                # it's a config-only variable that we don't need for
1807                # the command-line.
1808                push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
1809                my $v = $opts->{$o};
1810                my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
1811                $key =~ s/-//g;
1812                my $arg = 'git config';
1813                $arg .= ' --int' if ($o =~ /[:=]i$/);
1814                $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
1815                if (ref $v eq 'ARRAY') {
1816                        chomp(my @tmp = `$arg --get-all svn.$key`);
1817                        @$v = @tmp if @tmp;
1818                } else {
1819                        chomp(my $tmp = `$arg --get svn.$key`);
1820                        if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
1821                                $$v = $tmp;
1822                        }
1823                }
1824        }
1825        delete @$opts{@config_only} if @config_only;
1826}
1827
1828sub extract_metadata {
1829        my $id = shift or return (undef, undef, undef);
1830        my ($url, $rev, $uuid) = ($id =~ /^\s*git-svn-id:\s+(.*)\@(\d+)
1831                                                        \s([a-f\d\-]+)$/ix);
1832        if (!defined $rev || !$uuid || !$url) {
1833                # some of the original repositories I made had
1834                # identifiers like this:
1835                ($rev, $uuid) = ($id =~/^\s*git-svn-id:\s(\d+)\@([a-f\d\-]+)/i);
1836        }
1837        return ($url, $rev, $uuid);
1838}
1839
1840sub cmt_metadata {
1841        return extract_metadata((grep(/^git-svn-id: /,
1842                command(qw/cat-file commit/, shift)))[-1]);
1843}
1844
1845sub cmt_sha2rev_batch {
1846        my %s2r;
1847        my ($pid, $in, $out, $ctx) = command_bidi_pipe(qw/cat-file --batch/);
1848        my $list = shift;
1849
1850        foreach my $sha (@{$list}) {
1851                my $first = 1;
1852                my $size = 0;
1853                print $out $sha, "\n";
1854
1855                while (my $line = <$in>) {
1856                        if ($first && $line =~ /^[[:xdigit:]]{40}\smissing$/) {
1857                                last;
1858                        } elsif ($first &&
1859                               $line =~ /^[[:xdigit:]]{40}\scommit\s(\d+)$/) {
1860                                $first = 0;
1861                                $size = $1;
1862                                next;
1863                        } elsif ($line =~ /^(git-svn-id: )/) {
1864                                my (undef, $rev, undef) =
1865                                                      extract_metadata($line);
1866                                $s2r{$sha} = $rev;
1867                        }
1868
1869                        $size -= length($line);
1870                        last if ($size == 0);
1871                }
1872        }
1873
1874        command_close_bidi_pipe($pid, $in, $out, $ctx);
1875
1876        return \%s2r;
1877}
1878
1879sub working_head_info {
1880        my ($head, $refs) = @_;
1881        my @args = qw/log --no-color --no-decorate --first-parent
1882                      --pretty=medium/;
1883        my ($fh, $ctx) = command_output_pipe(@args, $head);
1884        my $hash;
1885        my %max;
1886        while (<$fh>) {
1887                if ( m{^commit ($::sha1)$} ) {
1888                        unshift @$refs, $hash if $hash and $refs;
1889                        $hash = $1;
1890                        next;
1891                }
1892                next unless s{^\s*(git-svn-id:)}{$1};
1893                my ($url, $rev, $uuid) = extract_metadata($_);
1894                if (defined $url && defined $rev) {
1895                        next if $max{$url} and $max{$url} < $rev;
1896                        if (my $gs = Git::SVN->find_by_url($url)) {
1897                                my $c = $gs->rev_map_get($rev, $uuid);
1898                                if ($c && $c eq $hash) {
1899                                        close $fh; # break the pipe
1900                                        return ($url, $rev, $uuid, $gs);
1901                                } else {
1902                                        $max{$url} ||= $gs->rev_map_max;
1903                                }
1904                        }
1905                }
1906        }
1907        command_close_pipe($fh, $ctx);
1908        (undef, undef, undef, undef);
1909}
1910
1911sub read_commit_parents {
1912        my ($parents, $c) = @_;
1913        chomp(my $p = command_oneline(qw/rev-list --parents -1/, $c));
1914        $p =~ s/^($c)\s*// or die "rev-list --parents -1 $c failed!\n";
1915        @{$parents->{$c}} = split(/ /, $p);
1916}
1917
1918sub linearize_history {
1919        my ($gs, $refs) = @_;
1920        my %parents;
1921        foreach my $c (@$refs) {
1922                read_commit_parents(\%parents, $c);
1923        }
1924
1925        my @linear_refs;
1926        my %skip = ();
1927        my $last_svn_commit = $gs->last_commit;
1928        foreach my $c (reverse @$refs) {
1929                next if $c eq $last_svn_commit;
1930                last if $skip{$c};
1931
1932                unshift @linear_refs, $c;
1933                $skip{$c} = 1;
1934
1935                # we only want the first parent to diff against for linear
1936                # history, we save the rest to inject when we finalize the
1937                # svn commit
1938                my $fp_a = verify_ref("$c~1");
1939                my $fp_b = shift @{$parents{$c}} if $parents{$c};
1940                if (!$fp_a || !$fp_b) {
1941                        die "Commit $c\n",
1942                            "has no parent commit, and therefore ",
1943                            "nothing to diff against.\n",
1944                            "You should be working from a repository ",
1945                            "originally created by git-svn\n";
1946                }
1947                if ($fp_a ne $fp_b) {
1948                        die "$c~1 = $fp_a, however parsing commit $c ",
1949                            "revealed that:\n$c~1 = $fp_b\nBUG!\n";
1950                }
1951
1952                foreach my $p (@{$parents{$c}}) {
1953                        $skip{$p} = 1;
1954                }
1955        }
1956        (\@linear_refs, \%parents);
1957}
1958
1959sub find_file_type_and_diff_status {
1960        my ($path) = @_;
1961        return ('dir', '') if $path eq '';
1962
1963        my $diff_output =
1964            command_oneline(qw(diff --cached --name-status --), $path) || "";
1965        my $diff_status = (split(' ', $diff_output))[0] || "";
1966
1967        my $ls_tree = command_oneline(qw(ls-tree HEAD), $path) || "";
1968
1969        return (undef, undef) if !$diff_status && !$ls_tree;
1970
1971        if ($diff_status eq "A") {
1972                return ("link", $diff_status) if -l $path;
1973                return ("dir", $diff_status) if -d $path;
1974                return ("file", $diff_status);
1975        }
1976
1977        my $mode = (split(' ', $ls_tree))[0] || "";
1978
1979        return ("link", $diff_status) if $mode eq "120000";
1980        return ("dir", $diff_status) if $mode eq "040000";
1981        return ("file", $diff_status);
1982}
1983
1984sub md5sum {
1985        my $arg = shift;
1986        my $ref = ref $arg;
1987        my $md5 = Digest::MD5->new();
1988        if ($ref eq 'GLOB' || $ref eq 'IO::File' || $ref eq 'File::Temp') {
1989                $md5->addfile($arg) or croak $!;
1990        } elsif ($ref eq 'SCALAR') {
1991                $md5->add($$arg) or croak $!;
1992        } elsif (!$ref) {
1993                $md5->add($arg) or croak $!;
1994        } else {
1995                ::fatal "Can't provide MD5 hash for unknown ref type: '", $ref, "'";
1996        }
1997        return $md5->hexdigest();
1998}
1999
2000sub gc_directory {
2001        if ($can_compress && -f $_ && basename($_) eq "unhandled.log") {
2002                my $out_filename = $_ . ".gz";
2003                open my $in_fh, "<", $_ or die "Unable to open $_: $!\n";
2004                binmode $in_fh;
2005                my $gz = Compress::Zlib::gzopen($out_filename, "ab") or
2006                                die "Unable to open $out_filename: $!\n";
2007
2008                my $res;
2009                while ($res = sysread($in_fh, my $str, 1024)) {
2010                        $gz->gzwrite($str) or
2011                                die "Unable to write: ".$gz->gzerror()."!\n";
2012                }
2013                unlink $_ or die "unlink $File::Find::name: $!\n";
2014        } elsif (-f $_ && basename($_) eq "index") {
2015                unlink $_ or die "unlink $_: $!\n";
2016        }
2017}
2018
2019package Git::SVN;
2020use strict;
2021use warnings;
2022use Fcntl qw/:DEFAULT :seek/;
2023use constant rev_map_fmt => 'NH40';
2024use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
2025            $_repack $_repack_flags $_use_svm_props $_head
2026            $_use_svnsync_props $no_reuse_existing $_minimize_url
2027            $_use_log_author $_add_author_from $_localtime/;
2028use Carp qw/croak/;
2029use File::Path qw/mkpath/;
2030use File::Copy qw/copy/;
2031use IPC::Open3;
2032use Memoize;  # core since 5.8.0, Jul 2002
2033use Memoize::Storable;
2034
2035my ($_gc_nr, $_gc_period);
2036
2037# properties that we do not log:
2038my %SKIP_PROP;
2039BEGIN {
2040        %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
2041                                        svn:special svn:executable
2042                                        svn:entry:committed-rev
2043                                        svn:entry:last-author
2044                                        svn:entry:uuid
2045                                        svn:entry:committed-date/;
2046
2047        # some options are read globally, but can be overridden locally
2048        # per [svn-remote "..."] section.  Command-line options will *NOT*
2049        # override options set in an [svn-remote "..."] section
2050        no strict 'refs';
2051        for my $option (qw/follow_parent no_metadata use_svm_props
2052                           use_svnsync_props/) {
2053                my $key = $option;
2054                $key =~ tr/_//d;
2055                my $prop = "-$option";
2056                *$option = sub {
2057                        my ($self) = @_;
2058                        return $self->{$prop} if exists $self->{$prop};
2059                        my $k = "svn-remote.$self->{repo_id}.$key";
2060                        eval { command_oneline(qw/config --get/, $k) };
2061                        if ($@) {
2062                                $self->{$prop} = ${"Git::SVN::_$option"};
2063                        } else {
2064                                my $v = command_oneline(qw/config --bool/,$k);
2065                                $self->{$prop} = $v eq 'false' ? 0 : 1;
2066                        }
2067                        return $self->{$prop};
2068                }
2069        }
2070}
2071
2072
2073my (%LOCKFILES, %INDEX_FILES);
2074END {
2075        unlink keys %LOCKFILES if %LOCKFILES;
2076        unlink keys %INDEX_FILES if %INDEX_FILES;
2077}
2078
2079sub resolve_local_globs {
2080        my ($url, $fetch, $glob_spec) = @_;
2081        return unless defined $glob_spec;
2082        my $ref = $glob_spec->{ref};
2083        my $path = $glob_spec->{path};
2084        foreach (command(qw#for-each-ref --format=%(refname) refs/#)) {
2085                next unless m#^$ref->{regex}$#;
2086                my $p = $1;
2087                my $pathname = desanitize_refname($path->full_path($p));
2088                my $refname = desanitize_refname($ref->full_path($p));
2089                if (my $existing = $fetch->{$pathname}) {
2090                        if ($existing ne $refname) {
2091                                die "Refspec conflict:\n",
2092                                    "existing: $existing\n",
2093                                    " globbed: $refname\n";
2094                        }
2095                        my $u = (::cmt_metadata("$refname"))[0];
2096                        $u =~ s!^\Q$url\E(/|$)!! or die
2097                          "$refname: '$url' not found in '$u'\n";
2098                        if ($pathname ne $u) {
2099                                warn "W: Refspec glob conflict ",
2100                                     "(ref: $refname):\n",
2101                                     "expected path: $pathname\n",
2102                                     "    real path: $u\n",
2103                                     "Continuing ahead with $u\n";
2104                                next;
2105                        }
2106                } else {
2107                        $fetch->{$pathname} = $refname;
2108                }
2109        }
2110}
2111
2112sub parse_revision_argument {
2113        my ($base, $head) = @_;
2114        if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
2115                return ($base, $head);
2116        }
2117        return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
2118        return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
2119        return ($head, $head) if ($::_revision eq 'HEAD');
2120        return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
2121        return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
2122        die "revision argument: $::_revision not understood by git-svn\n";
2123}
2124
2125sub fetch_all {
2126        my ($repo_id, $remotes) = @_;
2127        if (ref $repo_id) {
2128                my $gs = $repo_id;
2129                $repo_id = undef;
2130                $repo_id = $gs->{repo_id};
2131        }
2132        $remotes ||= read_all_remotes();
2133        my $remote = $remotes->{$repo_id} or
2134                     die "[svn-remote \"$repo_id\"] unknown\n";
2135        my $fetch = $remote->{fetch};
2136        my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
2137        my (@gs, @globs);
2138        my $ra = Git::SVN::Ra->new($url);
2139        my $uuid = $ra->get_uuid;
2140        my $head = $ra->get_latest_revnum;
2141
2142        # ignore errors, $head revision may not even exist anymore
2143        eval { $ra->get_log("", $head, 0, 1, 0, 1, sub { $head = $_[1] }) };
2144        warn "W: $@\n" if $@;
2145
2146        my $base = defined $fetch ? $head : 0;
2147
2148        # read the max revs for wildcard expansion (branches/*, tags/*)
2149        foreach my $t (qw/branches tags/) {
2150                defined $remote->{$t} or next;
2151                push @globs, @{$remote->{$t}};
2152
2153                my $max_rev = eval { tmp_config(qw/--int --get/,
2154                                         "svn-remote.$repo_id.${t}-maxRev") };
2155                if (defined $max_rev && ($max_rev < $base)) {
2156                        $base = $max_rev;
2157                } elsif (!defined $max_rev) {
2158                        $base = 0;
2159                }
2160        }
2161
2162        if ($fetch) {
2163                foreach my $p (sort keys %$fetch) {
2164                        my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
2165                        my $lr = $gs->rev_map_max;
2166                        if (defined $lr) {
2167                                $base = $lr if ($lr < $base);
2168                        }
2169                        push @gs, $gs;
2170                }
2171        }
2172
2173        ($base, $head) = parse_revision_argument($base, $head);
2174        $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
2175}
2176
2177sub read_all_remotes {
2178        my $r = {};
2179        my $use_svm_props = eval { command_oneline(qw/config --bool
2180            svn.useSvmProps/) };
2181        $use_svm_props = $use_svm_props eq 'true' if $use_svm_props;
2182        my $svn_refspec = qr{\s*(.*?)\s*:\s*(.+?)\s*};
2183        foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
2184                if (m!^(.+)\.fetch=$svn_refspec$!) {
2185                        my ($remote, $local_ref, $remote_ref) = ($1, $2, $3);
2186                        die("svn-remote.$remote: remote ref '$remote_ref' "
2187                            . "must start with 'refs/'\n")
2188                                unless $remote_ref =~ m{^refs/};
2189                        $local_ref = uri_decode($local_ref);
2190                        $r->{$remote}->{fetch}->{$local_ref} = $remote_ref;
2191                        $r->{$remote}->{svm} = {} if $use_svm_props;
2192                } elsif (m!^(.+)\.usesvmprops=\s*(.*)\s*$!) {
2193                        $r->{$1}->{svm} = {};
2194                } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
2195                        $r->{$1}->{url} = $2;
2196                } elsif (m!^(.+)\.pushurl=\s*(.*)\s*$!) {
2197                        $r->{$1}->{pushurl} = $2;
2198                } elsif (m!^(.+)\.ignore-refs=\s*(.*)\s*$!) {
2199                        $r->{$1}->{ignore_refs_regex} = $2;
2200                } elsif (m!^(.+)\.(branches|tags)=$svn_refspec$!) {
2201                        my ($remote, $t, $local_ref, $remote_ref) =
2202                                                             ($1, $2, $3, $4);
2203                        die("svn-remote.$remote: remote ref '$remote_ref' ($t) "
2204                            . "must start with 'refs/'\n")
2205                                unless $remote_ref =~ m{^refs/};
2206                        $local_ref = uri_decode($local_ref);
2207                        my $rs = {
2208                            t => $t,
2209                            remote => $remote,
2210                            path => Git::SVN::GlobSpec->new($local_ref, 1),
2211                            ref => Git::SVN::GlobSpec->new($remote_ref, 0) };
2212                        if (length($rs->{ref}->{right}) != 0) {
2213                                die "The '*' glob character must be the last ",
2214                                    "character of '$remote_ref'\n";
2215                        }
2216                        push @{ $r->{$remote}->{$t} }, $rs;
2217                }
2218        }
2219
2220        map {
2221                if (defined $r->{$_}->{svm}) {
2222                        my $svm;
2223                        eval {
2224                                my $section = "svn-remote.$_";
2225                                $svm = {
2226                                        source => tmp_config('--get',
2227                                            "$section.svm-source"),
2228                                        replace => tmp_config('--get',
2229                                            "$section.svm-replace"),
2230                                }
2231                        };
2232                        $r->{$_}->{svm} = $svm;
2233                }
2234        } keys %$r;
2235
2236        foreach my $remote (keys %$r) {
2237                foreach ( grep { defined $_ }
2238                          map { $r->{$remote}->{$_} } qw(branches tags) ) {
2239                        foreach my $rs ( @$_ ) {
2240                                $rs->{ignore_refs_regex} =
2241                                    $r->{$remote}->{ignore_refs_regex};
2242                        }
2243                }
2244        }
2245
2246        $r;
2247}
2248
2249sub init_vars {
2250        $_gc_nr = $_gc_period = 1000;
2251        if (defined $_repack || defined $_repack_flags) {
2252               warn "Repack options are obsolete; they have no effect.\n";
2253        }
2254}
2255
2256sub verify_remotes_sanity {
2257        return unless -d $ENV{GIT_DIR};
2258        my %seen;
2259        foreach (command(qw/config -l/)) {
2260                if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
2261                        if ($seen{$1}) {
2262                                die "Remote ref refs/remote/$1 is tracked by",
2263                                    "\n  \"$_\"\nand\n  \"$seen{$1}\"\n",
2264                                    "Please resolve this ambiguity in ",
2265                                    "your git configuration file before ",
2266                                    "continuing\n";
2267                        }
2268                        $seen{$1} = $_;
2269                }
2270        }
2271}
2272
2273sub find_existing_remote {
2274        my ($url, $remotes) = @_;
2275        return undef if $no_reuse_existing;
2276        my $existing;
2277        foreach my $repo_id (keys %$remotes) {
2278                my $u = $remotes->{$repo_id}->{url} or next;
2279                next if $u ne $url;
2280                $existing = $repo_id;
2281                last;
2282        }
2283        $existing;
2284}
2285
2286sub init_remote_config {
2287        my ($self, $url, $no_write) = @_;
2288        $url =~ s!/+$!!; # strip trailing slash
2289        my $r = read_all_remotes();
2290        my $existing = find_existing_remote($url, $r);
2291        if ($existing) {
2292                unless ($no_write) {
2293                        print STDERR "Using existing ",
2294                                     "[svn-remote \"$existing\"]\n";
2295                }
2296                $self->{repo_id} = $existing;
2297        } elsif ($_minimize_url) {
2298                my $min_url = Git::SVN::Ra->new($url)->minimize_url;
2299                $existing = find_existing_remote($min_url, $r);
2300                if ($existing) {
2301                        unless ($no_write) {
2302                                print STDERR "Using existing ",
2303                                             "[svn-remote \"$existing\"]\n";
2304                        }
2305                        $self->{repo_id} = $existing;
2306                }
2307                if ($min_url ne $url) {
2308                        unless ($no_write) {
2309                                print STDERR "Using higher level of URL: ",
2310                                             "$url => $min_url\n";
2311                        }
2312                        my $old_path = $self->{path};
2313                        $self->{path} = $url;
2314                        $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
2315                        if (length $old_path) {
2316                                $self->{path} .= "/$old_path";
2317                        }
2318                        $url = $min_url;
2319                }
2320        }
2321        my $orig_url;
2322        if (!$existing) {
2323                # verify that we aren't overwriting anything:
2324                $orig_url = eval {
2325                        command_oneline('config', '--get',
2326                                        "svn-remote.$self->{repo_id}.url")
2327                };
2328                if ($orig_url && ($orig_url ne $url)) {
2329                        die "svn-remote.$self->{repo_id}.url already set: ",
2330                            "$orig_url\nwanted to set to: $url\n";
2331                }
2332        }
2333        my ($xrepo_id, $xpath) = find_ref($self->refname);
2334        if (!$no_write && defined $xpath) {
2335                die "svn-remote.$xrepo_id.fetch already set to track ",
2336                    "$xpath:", $self->refname, "\n";
2337        }
2338        unless ($no_write) {
2339                command_noisy('config',
2340                              "svn-remote.$self->{repo_id}.url", $url);
2341                $self->{path} =~ s{^/}{};
2342                $self->{path} =~ s{%([0-9A-F]{2})}{chr hex($1)}ieg;
2343                command_noisy('config', '--add',
2344                              "svn-remote.$self->{repo_id}.fetch",
2345                              "$self->{path}:".$self->refname);
2346        }
2347        $self->{url} = $url;
2348}
2349
2350sub find_by_url { # repos_root and, path are optional
2351        my ($class, $full_url, $repos_root, $path) = @_;
2352
2353        return undef unless defined $full_url;
2354        remove_username($full_url);
2355        remove_username($repos_root) if defined $repos_root;
2356        my $remotes = read_all_remotes();
2357        if (defined $full_url && defined $repos_root && !defined $path) {
2358                $path = $full_url;
2359                $path =~ s#^\Q$repos_root\E(?:/|$)##;
2360        }
2361        foreach my $repo_id (keys %$remotes) {
2362                my $u = $remotes->{$repo_id}->{url} or next;
2363                remove_username($u);
2364                next if defined $repos_root && $repos_root ne $u;
2365
2366                my $fetch = $remotes->{$repo_id}->{fetch} || {};
2367                foreach my $t (qw/branches tags/) {
2368                        foreach my $globspec (@{$remotes->{$repo_id}->{$t}}) {
2369                                resolve_local_globs($u, $fetch, $globspec);
2370                        }
2371                }
2372                my $p = $path;
2373                my $rwr = rewrite_root({repo_id => $repo_id});
2374                my $svm = $remotes->{$repo_id}->{svm}
2375                        if defined $remotes->{$repo_id}->{svm};
2376                unless (defined $p) {
2377                        $p = $full_url;
2378                        my $z = $u;
2379                        my $prefix = '';
2380                        if ($rwr) {
2381                                $z = $rwr;
2382                                remove_username($z);
2383                        } elsif (defined $svm) {
2384                                $z = $svm->{source};
2385                                $prefix = $svm->{replace};
2386                                $prefix =~ s#^\Q$u\E(?:/|$)##;
2387                                $prefix =~ s#/$##;
2388                        }
2389                        $p =~ s#^\Q$z\E(?:/|$)#$prefix# or next;
2390                }
2391                foreach my $f (keys %$fetch) {
2392                        next if $f ne $p;
2393                        return Git::SVN->new($fetch->{$f}, $repo_id, $f);
2394                }
2395        }
2396        undef;
2397}
2398
2399sub init {
2400        my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
2401        my $self = _new($class, $repo_id, $ref_id, $path);
2402        if (defined $url) {
2403                $self->init_remote_config($url, $no_write);
2404        }
2405        $self;
2406}
2407
2408sub find_ref {
2409        my ($ref_id) = @_;
2410        foreach (command(qw/config -l/)) {
2411                next unless m!^svn-remote\.(.+)\.fetch=
2412                              \s*(.*?)\s*:\s*(.+?)\s*$!x;
2413                my ($repo_id, $path, $ref) = ($1, $2, $3);
2414                if ($ref eq $ref_id) {
2415                        $path = '' if ($path =~ m#^\./?#);
2416                        return ($repo_id, $path);
2417                }
2418        }
2419        (undef, undef, undef);
2420}
2421
2422sub new {
2423        my ($class, $ref_id, $repo_id, $path) = @_;
2424        if (defined $ref_id && !defined $repo_id && !defined $path) {
2425                ($repo_id, $path) = find_ref($ref_id);
2426                if (!defined $repo_id) {
2427                        die "Could not find a \"svn-remote.*.fetch\" key ",
2428                            "in the repository configuration matching: ",
2429                            "$ref_id\n";
2430                }
2431        }
2432        my $self = _new($class, $repo_id, $ref_id, $path);
2433        if (!defined $self->{path} || !length $self->{path}) {
2434                my $fetch = command_oneline('config', '--get',
2435                                            "svn-remote.$repo_id.fetch",
2436                                            ":$ref_id\$") or
2437                     die "Failed to read \"svn-remote.$repo_id.fetch\" ",
2438                         "\":$ref_id\$\" in config\n";
2439                ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
2440        }
2441        $self->{path} =~ s{/+}{/}g;
2442        $self->{path} =~ s{\A/}{};
2443        $self->{path} =~ s{/\z}{};
2444        $self->{url} = command_oneline('config', '--get',
2445                                       "svn-remote.$repo_id.url") or
2446                  die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
2447        $self->{pushurl} = eval { command_oneline('config', '--get',
2448                                  "svn-remote.$repo_id.pushurl") };
2449        $self->rebuild;
2450        $self;
2451}
2452
2453sub refname {
2454        my ($refname) = $_[0]->{ref_id} ;
2455
2456        # It cannot end with a slash /, we'll throw up on this because
2457        # SVN can't have directories with a slash in their name, either:
2458        if ($refname =~ m{/$}) {
2459                die "ref: '$refname' ends with a trailing slash, this is ",
2460                    "not permitted by git nor Subversion\n";
2461        }
2462
2463        # It cannot have ASCII control character space, tilde ~, caret ^,
2464        # colon :, question-mark ?, asterisk *, space, or open bracket [
2465        # anywhere.
2466        #
2467        # Additionally, % must be escaped because it is used for escaping
2468        # and we want our escaped refname to be reversible
2469        $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
2470
2471        # no slash-separated component can begin with a dot .
2472        # /.* becomes /%2E*
2473        $refname =~ s{/\.}{/%2E}g;
2474
2475        # It cannot have two consecutive dots .. anywhere
2476        # .. becomes %2E%2E
2477        $refname =~ s{\.\.}{%2E%2E}g;
2478
2479        # trailing dots and .lock are not allowed
2480        # .$ becomes %2E and .lock becomes %2Elock
2481        $refname =~ s{\.(?=$|lock$)}{%2E};
2482
2483        # the sequence @{ is used to access the reflog
2484        # @{ becomes %40{
2485        $refname =~ s{\@\{}{%40\{}g;
2486
2487        return $refname;
2488}
2489
2490sub desanitize_refname {
2491        my ($refname) = @_;
2492        $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
2493        return $refname;
2494}
2495
2496sub svm_uuid {
2497        my ($self) = @_;
2498        return $self->{svm}->{uuid} if $self->svm;
2499        $self->ra;
2500        unless ($self->{svm}) {
2501                die "SVM UUID not cached, and reading remotely failed\n";
2502        }
2503        $self->{svm}->{uuid};
2504}
2505
2506sub svm {
2507        my ($self) = @_;
2508        return $self->{svm} if $self->{svm};
2509        my $svm;
2510        # see if we have it in our config, first:
2511        eval {
2512                my $section = "svn-remote.$self->{repo_id}";
2513                $svm = {
2514                  source => tmp_config('--get', "$section.svm-source"),
2515                  uuid => tmp_config('--get', "$section.svm-uuid"),
2516                  replace => tmp_config('--get', "$section.svm-replace"),
2517                }
2518        };
2519        if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
2520                $self->{svm} = $svm;
2521        }
2522        $self->{svm};
2523}
2524
2525sub _set_svm_vars {
2526        my ($self, $ra) = @_;
2527        return $ra if $self->svm;
2528
2529        my @err = ( "useSvmProps set, but failed to read SVM properties\n",
2530                    "(svm:source, svm:uuid) ",
2531                    "from the following URLs:\n" );
2532        sub read_svm_props {
2533                my ($self, $ra, $path, $r) = @_;
2534                my $props = ($ra->get_dir($path, $r))[2];
2535                my $src = $props->{'svm:source'};
2536                my $uuid = $props->{'svm:uuid'};
2537                return undef if (!$src || !$uuid);
2538
2539                chomp($src, $uuid);
2540
2541                $uuid =~ m{^[0-9a-f\-]{30,}$}i
2542                    or die "doesn't look right - svm:uuid is '$uuid'\n";
2543
2544                # the '!' is used to mark the repos_root!/relative/path
2545                $src =~ s{/?!/?}{/};
2546                $src =~ s{/+$}{}; # no trailing slashes please
2547                # username is of no interest
2548                $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
2549
2550                my $replace = $ra->{url};
2551                $replace .= "/$path" if length $path;
2552
2553                my $section = "svn-remote.$self->{repo_id}";
2554                tmp_config("$section.svm-source", $src);
2555                tmp_config("$section.svm-replace", $replace);
2556                tmp_config("$section.svm-uuid", $uuid);
2557                $self->{svm} = {
2558                        source => $src,
2559                        uuid => $uuid,
2560                        replace => $replace
2561                };
2562        }
2563
2564        my $r = $ra->get_latest_revnum;
2565        my $path = $self->{path};
2566        my %tried;
2567        while (length $path) {
2568                unless ($tried{"$self->{url}/$path"}) {
2569                        return $ra if $self->read_svm_props($ra, $path, $r);
2570                        $tried{"$self->{url}/$path"} = 1;
2571                }
2572                $path =~ s#/?[^/]+$##;
2573        }
2574        die "Path: '$path' should be ''\n" if $path ne '';
2575        return $ra if $self->read_svm_props($ra, $path, $r);
2576        $tried{"$self->{url}/$path"} = 1;
2577
2578        if ($ra->{repos_root} eq $self->{url}) {
2579                die @err, (map { "  $_\n" } keys %tried), "\n";
2580        }
2581
2582        # nope, make sure we're connected to the repository root:
2583        my $ok;
2584        my @tried_b;
2585        $path = $ra->{svn_path};
2586        $ra = Git::SVN::Ra->new($ra->{repos_root});
2587        while (length $path) {
2588                unless ($tried{"$ra->{url}/$path"}) {
2589                        $ok = $self->read_svm_props($ra, $path, $r);
2590                        last if $ok;
2591                        $tried{"$ra->{url}/$path"} = 1;
2592                }
2593                $path =~ s#/?[^/]+$##;
2594        }
2595        die "Path: '$path' should be ''\n" if $path ne '';
2596        $ok ||= $self->read_svm_props($ra, $path, $r);
2597        $tried{"$ra->{url}/$path"} = 1;
2598        if (!$ok) {
2599                die @err, (map { "  $_\n" } keys %tried), "\n";
2600        }
2601        Git::SVN::Ra->new($self->{url});
2602}
2603
2604sub svnsync {
2605        my ($self) = @_;
2606        return $self->{svnsync} if $self->{svnsync};
2607
2608        if ($self->no_metadata) {
2609                die "Can't have both 'noMetadata' and ",
2610                    "'useSvnsyncProps' options set!\n";
2611        }
2612        if ($self->rewrite_root) {
2613                die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
2614                    "options set!\n";
2615        }
2616        if ($self->rewrite_uuid) {
2617                die "Can't have both 'useSvnsyncProps' and 'rewriteUUID' ",
2618                    "options set!\n";
2619        }
2620
2621        my $svnsync;
2622        # see if we have it in our config, first:
2623        eval {
2624                my $section = "svn-remote.$self->{repo_id}";
2625
2626                my $url = tmp_config('--get', "$section.svnsync-url");
2627                ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2628                   die "doesn't look right - svn:sync-from-url is '$url'\n";
2629
2630                my $uuid = tmp_config('--get', "$section.svnsync-uuid");
2631                ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2632                   die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2633
2634                $svnsync = { url => $url, uuid => $uuid }
2635        };
2636        if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
2637                return $self->{svnsync} = $svnsync;
2638        }
2639
2640        my $err = "useSvnsyncProps set, but failed to read " .
2641                  "svnsync property: svn:sync-from-";
2642        my $rp = $self->ra->rev_proplist(0);
2643
2644        my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
2645        ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2646                   die "doesn't look right - svn:sync-from-url is '$url'\n";
2647
2648        my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
2649        ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2650                   die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2651
2652        my $section = "svn-remote.$self->{repo_id}";
2653        tmp_config('--add', "$section.svnsync-uuid", $uuid);
2654        tmp_config('--add', "$section.svnsync-url", $url);
2655        return $self->{svnsync} = { url => $url, uuid => $uuid };
2656}
2657
2658# this allows us to memoize our SVN::Ra UUID locally and avoid a
2659# remote lookup (useful for 'git svn log').
2660sub ra_uuid {
2661        my ($self) = @_;
2662        unless ($self->{ra_uuid}) {
2663                my $key = "svn-remote.$self->{repo_id}.uuid";
2664                my $uuid = eval { tmp_config('--get', $key) };
2665                if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/i) {
2666                        $self->{ra_uuid} = $uuid;
2667                } else {
2668                        die "ra_uuid called without URL\n" unless $self->{url};
2669                        $self->{ra_uuid} = $self->ra->get_uuid;
2670                        tmp_config('--add', $key, $self->{ra_uuid});
2671                }
2672        }
2673        $self->{ra_uuid};
2674}
2675
2676sub _set_repos_root {
2677        my ($self, $repos_root) = @_;
2678        my $k = "svn-remote.$self->{repo_id}.reposRoot";
2679        $repos_root ||= $self->ra->{repos_root};
2680        tmp_config($k, $repos_root);
2681        $repos_root;
2682}
2683
2684sub repos_root {
2685        my ($self) = @_;
2686        my $k = "svn-remote.$self->{repo_id}.reposRoot";
2687        eval { tmp_config('--get', $k) } || $self->_set_repos_root;
2688}
2689
2690sub ra {
2691        my ($self) = shift;
2692        my $ra = Git::SVN::Ra->new($self->{url});
2693        $self->_set_repos_root($ra->{repos_root});
2694        if ($self->use_svm_props && !$self->{svm}) {
2695                if ($self->no_metadata) {
2696                        die "Can't have both 'noMetadata' and ",
2697                            "'useSvmProps' options set!\n";
2698                } elsif ($self->use_svnsync_props) {
2699                        die "Can't have both 'useSvnsyncProps' and ",
2700                            "'useSvmProps' options set!\n";
2701                }
2702                $ra = $self->_set_svm_vars($ra);
2703                $self->{-want_revprops} = 1;
2704        }
2705        $ra;
2706}
2707
2708# prop_walk(PATH, REV, SUB)
2709# -------------------------
2710# Recursively traverse PATH at revision REV and invoke SUB for each
2711# directory that contains a SVN property.  SUB will be invoked as
2712# follows:  &SUB(gs, path, props);  where `gs' is this instance of
2713# Git::SVN, `path' the path to the directory where the properties
2714# `props' were found.  The `path' will be relative to point of checkout,
2715# that is, if url://repo/trunk is the current Git branch, and that
2716# directory contains a sub-directory `d', SUB will be invoked with `/d/'
2717# as `path' (note the trailing `/').
2718sub prop_walk {
2719        my ($self, $path, $rev, $sub) = @_;
2720
2721        $path =~ s#^/##;
2722        my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev);
2723        $path =~ s#^/*#/#g;
2724        my $p = $path;
2725        # Strip the irrelevant part of the path.
2726        $p =~ s#^/+\Q$self->{path}\E(/|$)#/#;
2727        # Ensure the path is terminated by a `/'.
2728        $p =~ s#/*$#/#;
2729
2730        # The properties contain all the internal SVN stuff nobody
2731        # (usually) cares about.
2732        my $interesting_props = 0;
2733        foreach (keys %{$props}) {
2734                # If it doesn't start with `svn:', it must be a
2735                # user-defined property.
2736                ++$interesting_props and next if $_ !~ /^svn:/;
2737                # FIXME: Fragile, if SVN adds new public properties,
2738                # this needs to be updated.
2739                ++$interesting_props if /^svn:(?:ignore|keywords|executable
2740                                                 |eol-style|mime-type
2741                                                 |externals|needs-lock)$/x;
2742        }
2743        &$sub($self, $p, $props) if $interesting_props;
2744
2745        foreach (sort keys %$dirent) {
2746                next if $dirent->{$_}->{kind} != $SVN::Node::dir;
2747                $self->prop_walk($self->{path} . $p . $_, $rev, $sub);
2748        }
2749}
2750
2751sub last_rev { ($_[0]->last_rev_commit)[0] }
2752sub last_commit { ($_[0]->last_rev_commit)[1] }
2753
2754# returns the newest SVN revision number and newest commit SHA1
2755sub last_rev_commit {
2756        my ($self) = @_;
2757        if (defined $self->{last_rev} && defined $self->{last_commit}) {
2758                return ($self->{last_rev}, $self->{last_commit});
2759        }
2760        my $c = ::verify_ref($self->refname.'^0');
2761        if ($c && !$self->use_svm_props && !$self->no_metadata) {
2762                my $rev = (::cmt_metadata($c))[1];
2763                if (defined $rev) {
2764                        ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
2765                        return ($rev, $c);
2766                }
2767        }
2768        my $map_path = $self->map_path;
2769        unless (-e $map_path) {
2770                ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
2771                return (undef, undef);
2772        }
2773        my ($rev, $commit) = $self->rev_map_max(1);
2774        ($self->{last_rev}, $self->{last_commit}) = ($rev, $commit);
2775        return ($rev, $commit);
2776}
2777
2778sub get_fetch_range {
2779        my ($self, $min, $max) = @_;
2780        $max ||= $self->ra->get_latest_revnum;
2781        $min ||= $self->rev_map_max;
2782        (++$min, $max);
2783}
2784
2785sub tmp_config {
2786        my (@args) = @_;
2787        my $old_def_config = "$ENV{GIT_DIR}/svn/config";
2788        my $config = "$ENV{GIT_DIR}/svn/.metadata";
2789        if (! -f $config && -f $old_def_config) {
2790                rename $old_def_config, $config or
2791                       die "Failed rename $old_def_config => $config: $!\n";
2792        }
2793        my $old_config = $ENV{GIT_CONFIG};
2794        $ENV{GIT_CONFIG} = $config;
2795        $@ = undef;
2796        my @ret = eval {
2797                unless (-f $config) {
2798                        mkfile($config);
2799                        open my $fh, '>', $config or
2800                            die "Can't open $config: $!\n";
2801                        print $fh "; This file is used internally by ",
2802                                  "git-svn\n" or die
2803                                  "Couldn't write to $config: $!\n";
2804                        print $fh "; You should not have to edit it\n" or
2805                              die "Couldn't write to $config: $!\n";
2806                        close $fh or die "Couldn't close $config: $!\n";
2807                }
2808                command('config', @args);
2809        };
2810        my $err = $@;
2811        if (defined $old_config) {
2812                $ENV{GIT_CONFIG} = $old_config;
2813        } else {
2814                delete $ENV{GIT_CONFIG};
2815        }
2816        die $err if $err;
2817        wantarray ? @ret : $ret[0];
2818}
2819
2820sub tmp_index_do {
2821        my ($self, $sub) = @_;
2822        my $old_index = $ENV{GIT_INDEX_FILE};
2823        $ENV{GIT_INDEX_FILE} = $self->{index};
2824        $@ = undef;
2825        my @ret = eval {
2826                my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
2827                mkpath([$dir]) unless -d $dir;
2828                &$sub;
2829        };
2830        my $err = $@;
2831        if (defined $old_index) {
2832                $ENV{GIT_INDEX_FILE} = $old_index;
2833        } else {
2834                delete $ENV{GIT_INDEX_FILE};
2835        }
2836        die $err if $err;
2837        wantarray ? @ret : $ret[0];
2838}
2839
2840sub assert_index_clean {
2841        my ($self, $treeish) = @_;
2842
2843        $self->tmp_index_do(sub {
2844                command_noisy('read-tree', $treeish) unless -e $self->{index};
2845                my $x = command_oneline('write-tree');
2846                my ($y) = (command(qw/cat-file commit/, $treeish) =~
2847                           /^tree ($::sha1)/mo);
2848                return if $y eq $x;
2849
2850                warn "Index mismatch: $y != $x\nrereading $treeish\n";
2851                unlink $self->{index} or die "unlink $self->{index}: $!\n";
2852                command_noisy('read-tree', $treeish);
2853                $x = command_oneline('write-tree');
2854                if ($y ne $x) {
2855                        ::fatal "trees ($treeish) $y != $x\n",
2856                                "Something is seriously wrong...";
2857                }
2858        });
2859}
2860
2861sub get_commit_parents {
2862        my ($self, $log_entry) = @_;
2863        my (%seen, @ret, @tmp);
2864        # legacy support for 'set-tree'; this is only used by set_tree_cb:
2865        if (my $ip = $self->{inject_parents}) {
2866                if (my $commit = delete $ip->{$log_entry->{revision}}) {
2867                        push @tmp, $commit;
2868                }
2869        }
2870        if (my $cur = ::verify_ref($self->refname.'^0')) {
2871                push @tmp, $cur;
2872        }
2873        if (my $ipd = $self->{inject_parents_dcommit}) {
2874                if (my $commit = delete $ipd->{$log_entry->{revision}}) {
2875                        push @tmp, @$commit;
2876                }
2877        }
2878        push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
2879        while (my $p = shift @tmp) {
2880                next if $seen{$p};
2881                $seen{$p} = 1;
2882                push @ret, $p;
2883        }
2884        @ret;
2885}
2886
2887sub rewrite_root {
2888        my ($self) = @_;
2889        return $self->{-rewrite_root} if exists $self->{-rewrite_root};
2890        my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
2891        my $rwr = eval { command_oneline(qw/config --get/, $k) };
2892        if ($rwr) {
2893                $rwr =~ s#/+$##;
2894                if ($rwr !~ m#^[a-z\+]+://#) {
2895                        die "$rwr is not a valid URL (key: $k)\n";
2896                }
2897        }
2898        $self->{-rewrite_root} = $rwr;
2899}
2900
2901sub rewrite_uuid {
2902        my ($self) = @_;
2903        return $self->{-rewrite_uuid} if exists $self->{-rewrite_uuid};
2904        my $k = "svn-remote.$self->{repo_id}.rewriteUUID";
2905        my $rwid = eval { command_oneline(qw/config --get/, $k) };
2906        if ($rwid) {
2907                $rwid =~ s#/+$##;
2908                if ($rwid !~ m#^[a-f0-9]{8}-(?:[a-f0-9]{4}-){3}[a-f0-9]{12}$#) {
2909                        die "$rwid is not a valid UUID (key: $k)\n";
2910                }
2911        }
2912        $self->{-rewrite_uuid} = $rwid;
2913}
2914
2915sub metadata_url {
2916        my ($self) = @_;
2917        ($self->rewrite_root || $self->{url}) .
2918           (length $self->{path} ? '/' . $self->{path} : '');
2919}
2920
2921sub full_url {
2922        my ($self) = @_;
2923        $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
2924}
2925
2926sub full_pushurl {
2927        my ($self) = @_;
2928        if ($self->{pushurl}) {
2929                return $self->{pushurl} . (length $self->{path} ? '/' .
2930                       $self->{path} : '');
2931        } else {
2932                return $self->full_url;
2933        }
2934}
2935
2936sub set_commit_header_env {
2937        my ($log_entry) = @_;
2938        my %env;
2939        foreach my $ned (qw/NAME EMAIL DATE/) {
2940                foreach my $ac (qw/AUTHOR COMMITTER/) {
2941                        $env{"GIT_${ac}_${ned}"} = $ENV{"GIT_${ac}_${ned}"};
2942                }
2943        }
2944
2945        $ENV{GIT_AUTHOR_NAME} = $log_entry->{name};
2946        $ENV{GIT_AUTHOR_EMAIL} = $log_entry->{email};
2947        $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
2948
2949        $ENV{GIT_COMMITTER_NAME} = (defined $log_entry->{commit_name})
2950                                                ? $log_entry->{commit_name}
2951                                                : $log_entry->{name};
2952        $ENV{GIT_COMMITTER_EMAIL} = (defined $log_entry->{commit_email})
2953                                                ? $log_entry->{commit_email}
2954                                                : $log_entry->{email};
2955        \%env;
2956}
2957
2958sub restore_commit_header_env {
2959        my ($env) = @_;
2960        foreach my $ned (qw/NAME EMAIL DATE/) {
2961                foreach my $ac (qw/AUTHOR COMMITTER/) {
2962                        my $k = "GIT_${ac}_${ned}";
2963                        if (defined $env->{$k}) {
2964                                $ENV{$k} = $env->{$k};
2965                        } else {
2966                                delete $ENV{$k};
2967                        }
2968                }
2969        }
2970}
2971
2972sub gc {
2973        command_noisy('gc', '--auto');
2974};
2975
2976sub do_git_commit {
2977        my ($self, $log_entry) = @_;
2978        my $lr = $self->last_rev;
2979        if (defined $lr && $lr >= $log_entry->{revision}) {
2980                die "Last fetched revision of ", $self->refname,
2981                    " was r$lr, but we are about to fetch: ",
2982                    "r$log_entry->{revision}!\n";
2983        }
2984        if (my $c = $self->rev_map_get($log_entry->{revision})) {
2985                croak "$log_entry->{revision} = $c already exists! ",
2986                      "Why are we refetching it?\n";
2987        }
2988        my $old_env = set_commit_header_env($log_entry);
2989        my $tree = $log_entry->{tree};
2990        if (!defined $tree) {
2991                $tree = $self->tmp_index_do(sub {
2992                                            command_oneline('write-tree') });
2993        }
2994        die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
2995
2996        my @exec = ('git', 'commit-tree', $tree);
2997        foreach ($self->get_commit_parents($log_entry)) {
2998                push @exec, '-p', $_;
2999        }
3000        defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
3001                                                                   or croak $!;
3002        binmode $msg_fh;
3003
3004        # we always get UTF-8 from SVN, but we may want our commits in
3005        # a different encoding.
3006        if (my $enc = Git::config('i18n.commitencoding')) {
3007                require Encode;
3008                Encode::from_to($log_entry->{log}, 'UTF-8', $enc);
3009        }
3010        print $msg_fh $log_entry->{log} or croak $!;
3011        restore_commit_header_env($old_env);
3012        unless ($self->no_metadata) {
3013                print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
3014                              or croak $!;
3015        }
3016        $msg_fh->flush == 0 or croak $!;
3017        close $msg_fh or croak $!;
3018        chomp(my $commit = do { local $/; <$out_fh> });
3019        close $out_fh or croak $!;
3020        waitpid $pid, 0;
3021        croak $? if $?;
3022        if ($commit !~ /^$::sha1$/o) {
3023                die "Failed to commit, invalid sha1: $commit\n";
3024        }
3025
3026        $self->rev_map_set($log_entry->{revision}, $commit, 1);
3027
3028        $self->{last_rev} = $log_entry->{revision};
3029        $self->{last_commit} = $commit;
3030        print "r$log_entry->{revision}" unless $::_q > 1;
3031        if (defined $log_entry->{svm_revision}) {
3032                 print " (\@$log_entry->{svm_revision})" unless $::_q > 1;
3033                 $self->rev_map_set($log_entry->{svm_revision}, $commit,
3034                                   0, $self->svm_uuid);
3035        }
3036        print " = $commit ($self->{ref_id})\n" unless $::_q > 1;
3037        if (--$_gc_nr == 0) {
3038                $_gc_nr = $_gc_period;
3039                gc();
3040        }
3041        return $commit;
3042}
3043
3044sub match_paths {
3045        my ($self, $paths, $r) = @_;
3046        return 1 if $self->{path} eq '';
3047        if (my $path = $paths->{"/$self->{path}"}) {
3048                return ($path->{action} eq 'D') ? 0 : 1;
3049        }
3050        $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
3051        if (grep /$self->{path_regex}/, keys %$paths) {
3052                return 1;
3053        }
3054        my $c = '';
3055        foreach (split m#/#, $self->{path}) {
3056                $c .= "/$_";
3057                next unless ($paths->{$c} &&
3058                             ($paths->{$c}->{action} =~ /^[AR]$/));
3059                if ($self->ra->check_path($self->{path}, $r) ==
3060                    $SVN::Node::dir) {
3061                        return 1;
3062                }
3063        }
3064        return 0;
3065}
3066
3067sub find_parent_branch {
3068        my ($self, $paths, $rev) = @_;
3069        return undef unless $self->follow_parent;
3070        unless (defined $paths) {
3071                my $err_handler = $SVN::Error::handler;
3072                $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
3073                $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1,
3074                                   sub { $paths = $_[0] });
3075                $SVN::Error::handler = $err_handler;
3076        }
3077        return undef unless defined $paths;
3078
3079        # look for a parent from another branch:
3080        my @b_path_components = split m#/#, $self->{path};
3081        my @a_path_components;
3082        my $i;
3083        while (@b_path_components) {
3084                $i = $paths->{'/'.join('/', @b_path_components)};
3085                last if $i && defined $i->{copyfrom_path};
3086                unshift(@a_path_components, pop(@b_path_components));
3087        }
3088        return undef unless defined $i && defined $i->{copyfrom_path};
3089        my $branch_from = $i->{copyfrom_path};
3090        if (@a_path_components) {
3091                print STDERR "branch_from: $branch_from => ";
3092                $branch_from .= '/'.join('/', @a_path_components);
3093                print STDERR $branch_from, "\n";
3094        }
3095        my $r = $i->{copyfrom_rev};
3096        my $repos_root = $self->ra->{repos_root};
3097        my $url = $self->ra->{url};
3098        my $new_url = $url . $branch_from;
3099        print STDERR  "Found possible branch point: ",
3100                      "$new_url => ", $self->full_url, ", $r\n"
3101                      unless $::_q > 1;
3102        $branch_from =~ s#^/##;
3103        my $gs = $self->other_gs($new_url, $url,
3104                                 $branch_from, $r, $self->{ref_id});
3105        my ($r0, $parent) = $gs->find_rev_before($r, 1);
3106        {
3107                my ($base, $head);
3108                if (!defined $r0 || !defined $parent) {
3109                        ($base, $head) = parse_revision_argument(0, $r);
3110                } else {
3111                        if ($r0 < $r) {
3112                                $gs->ra->get_log([$gs->{path}], $r0 + 1, $r, 1,
3113                                        0, 1, sub { $base = $_[1] - 1 });
3114                        }
3115                }
3116                if (defined $base && $base <= $r) {
3117                        $gs->fetch($base, $r);
3118                }
3119                ($r0, $parent) = $gs->find_rev_before($r, 1);
3120        }
3121        if (defined $r0 && defined $parent) {
3122                print STDERR "Found branch parent: ($self->{ref_id}) $parent\n"
3123                             unless $::_q > 1;
3124                my $ed;
3125                if ($self->ra->can_do_switch) {
3126                        $self->assert_index_clean($parent);
3127                        print STDERR "Following parent with do_switch\n"
3128                                     unless $::_q > 1;
3129                        # do_switch works with svn/trunk >= r22312, but that
3130                        # is not included with SVN 1.4.3 (the latest version
3131                        # at the moment), so we can't rely on it
3132                        $self->{last_rev} = $r0;
3133                        $self->{last_commit} = $parent;
3134                        $ed = SVN::Git::Fetcher->new($self, $gs->{path});
3135                        $gs->ra->gs_do_switch($r0, $rev, $gs,
3136                                              $self->full_url, $ed)
3137                          or die "SVN connection failed somewhere...\n";
3138                } elsif ($self->ra->trees_match($new_url, $r0,
3139                                                $self->full_url, $rev)) {
3140                        print STDERR "Trees match:\n",
3141                                     "  $new_url\@$r0\n",
3142                                     "  ${\$self->full_url}\@$rev\n",
3143                                     "Following parent with no changes\n"
3144                                     unless $::_q > 1;
3145                        $self->tmp_index_do(sub {
3146                            command_noisy('read-tree', $parent);
3147                        });
3148                        $self->{last_commit} = $parent;
3149                } else {
3150                        print STDERR "Following parent with do_update\n"
3151                                     unless $::_q > 1;
3152                        $ed = SVN::Git::Fetcher->new($self);
3153                        $self->ra->gs_do_update($rev, $rev, $self, $ed)
3154                          or die "SVN connection failed somewhere...\n";
3155                }
3156                print STDERR "Successfully followed parent\n" unless $::_q > 1;
3157                return $self->make_log_entry($rev, [$parent], $ed);
3158        }
3159        return undef;
3160}
3161
3162sub do_fetch {
3163        my ($self, $paths, $rev) = @_;
3164        my $ed;
3165        my ($last_rev, @parents);
3166        if (my $lc = $self->last_commit) {
3167                # we can have a branch that was deleted, then re-added
3168                # under the same name but copied from another path, in
3169                # which case we'll have multiple parents (we don't
3170                # want to break the original ref, nor lose copypath info):
3171                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
3172                        push @{$log_entry->{parents}}, $lc;
3173                        return $log_entry;
3174                }
3175                $ed = SVN::Git::Fetcher->new($self);
3176                $last_rev = $self->{last_rev};
3177                $ed->{c} = $lc;
3178                @parents = ($lc);
3179        } else {
3180                $last_rev = $rev;
3181                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
3182                        return $log_entry;
3183                }
3184                $ed = SVN::Git::Fetcher->new($self);
3185        }
3186        unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
3187                die "SVN connection failed somewhere...\n";
3188        }
3189        $self->make_log_entry($rev, \@parents, $ed);
3190}
3191
3192sub mkemptydirs {
3193        my ($self, $r) = @_;
3194
3195        sub scan {
3196                my ($r, $empty_dirs, $line) = @_;
3197                if (defined $r && $line =~ /^r(\d+)$/) {
3198                        return 0 if $1 > $r;
3199                } elsif ($line =~ /^  \+empty_dir: (.+)$/) {
3200                        $empty_dirs->{$1} = 1;
3201                } elsif ($line =~ /^  \-empty_dir: (.+)$/) {
3202                        my @d = grep {m[^\Q$1\E(/|$)]} (keys %$empty_dirs);
3203                        delete @$empty_dirs{@d};
3204                }
3205                1; # continue
3206        };
3207
3208        my %empty_dirs = ();
3209        my $gz_file = "$self->{dir}/unhandled.log.gz";
3210        if (-f $gz_file) {
3211                if (!$can_compress) {
3212                        warn "Compress::Zlib could not be found; ",
3213                             "empty directories in $gz_file will not be read\n";
3214                } else {
3215                        my $gz = Compress::Zlib::gzopen($gz_file, "rb") or
3216                                die "Unable to open $gz_file: $!\n";
3217                        my $line;
3218                        while ($gz->gzreadline($line) > 0) {
3219                                scan($r, \%empty_dirs, $line) or last;
3220                        }
3221                        $gz->gzclose;
3222                }
3223        }
3224
3225        if (open my $fh, '<', "$self->{dir}/unhandled.log") {
3226                binmode $fh or croak "binmode: $!";
3227                while (<$fh>) {
3228                        scan($r, \%empty_dirs, $_) or last;
3229                }
3230                close $fh;
3231        }
3232
3233        my $strip = qr/\A\Q$self->{path}\E(?:\/|$)/;
3234        foreach my $d (sort keys %empty_dirs) {
3235                $d = uri_decode($d);
3236                $d =~ s/$strip//;
3237                next unless length($d);
3238                next if -d $d;
3239                if (-e $d) {
3240                        warn "$d exists but is not a directory\n";
3241                } else {
3242                        print "creating empty directory: $d\n";
3243                        mkpath([$d]);
3244                }
3245        }
3246}
3247
3248sub get_untracked {
3249        my ($self, $ed) = @_;
3250        my @out;
3251        my $h = $ed->{empty};
3252        foreach (sort keys %$h) {
3253                my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
3254                push @out, "  $act: " . uri_encode($_);
3255                warn "W: $act: $_\n";
3256        }
3257        foreach my $t (qw/dir_prop file_prop/) {
3258                $h = $ed->{$t} or next;
3259                foreach my $path (sort keys %$h) {
3260                        my $ppath = $path eq '' ? '.' : $path;
3261                        foreach my $prop (sort keys %{$h->{$path}}) {
3262                                next if $SKIP_PROP{$prop};
3263                                my $v = $h->{$path}->{$prop};
3264                                my $t_ppath_prop = "$t: " .
3265                                                    uri_encode($ppath) . ' ' .
3266                                                    uri_encode($prop);
3267                                if (defined $v) {
3268                                        push @out, "  +$t_ppath_prop " .
3269                                                   uri_encode($v);
3270                                } else {
3271                                        push @out, "  -$t_ppath_prop";
3272                                }
3273                        }
3274                }
3275        }
3276        foreach my $t (qw/absent_file absent_directory/) {
3277                $h = $ed->{$t} or next;
3278                foreach my $parent (sort keys %$h) {
3279                        foreach my $path (sort @{$h->{$parent}}) {
3280                                push @out, "  $t: " .
3281                                           uri_encode("$parent/$path");
3282                                warn "W: $t: $parent/$path ",
3283                                     "Insufficient permissions?\n";
3284                        }
3285                }
3286        }
3287        \@out;
3288}
3289
3290# parse_svn_date(DATE)
3291# --------------------
3292# Given a date (in UTC) from Subversion, return a string in the format
3293# "<TZ Offset> <local date/time>" that Git will use.
3294#
3295# By default the parsed date will be in UTC; if $Git::SVN::_localtime
3296# is true we'll convert it to the local timezone instead.
3297sub parse_svn_date {
3298        my $date = shift || return '+0000 1970-01-01 00:00:00';
3299        my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
3300                                            (\d\d)\:(\d\d)\:(\d\d)\.\d*Z$/x) or
3301                                         croak "Unable to parse date: $date\n";
3302        my $parsed_date;    # Set next.
3303
3304        if ($Git::SVN::_localtime) {
3305                # Translate the Subversion datetime to an epoch time.
3306                # Begin by switching ourselves to $date's timezone, UTC.
3307                my $old_env_TZ = $ENV{TZ};
3308                $ENV{TZ} = 'UTC';
3309
3310                my $epoch_in_UTC =
3311                    POSIX::strftime('%s', $S, $M, $H, $d, $m - 1, $Y - 1900);
3312
3313                # Determine our local timezone (including DST) at the
3314                # time of $epoch_in_UTC.  $Git::SVN::Log::TZ stored the
3315                # value of TZ, if any, at the time we were run.
3316                if (defined $Git::SVN::Log::TZ) {
3317                        $ENV{TZ} = $Git::SVN::Log::TZ;
3318                } else {
3319                        delete $ENV{TZ};
3320                }
3321
3322                my $our_TZ =
3323                    POSIX::strftime('%Z', $S, $M, $H, $d, $m - 1, $Y - 1900);
3324
3325                # This converts $epoch_in_UTC into our local timezone.
3326                my ($sec, $min, $hour, $mday, $mon, $year,
3327                    $wday, $yday, $isdst) = localtime($epoch_in_UTC);
3328
3329                $parsed_date = sprintf('%s %04d-%02d-%02d %02d:%02d:%02d',
3330                                       $our_TZ, $year + 1900, $mon + 1,
3331                                       $mday, $hour, $min, $sec);
3332
3333                # Reset us to the timezone in effect when we entered
3334                # this routine.
3335                if (defined $old_env_TZ) {
3336                        $ENV{TZ} = $old_env_TZ;
3337                } else {
3338                        delete $ENV{TZ};
3339                }
3340        } else {
3341                $parsed_date = "+0000 $Y-$m-$d $H:$M:$S";
3342        }
3343
3344        return $parsed_date;
3345}
3346
3347sub other_gs {
3348        my ($self, $new_url, $url,
3349            $branch_from, $r, $old_ref_id) = @_;
3350        my $gs = Git::SVN->find_by_url($new_url, $url, $branch_from);
3351        unless ($gs) {
3352                my $ref_id = $old_ref_id;
3353                $ref_id =~ s/\@\d+-*$//;
3354                $ref_id .= "\@$r";
3355                # just grow a tail if we're not unique enough :x
3356                $ref_id .= '-' while find_ref($ref_id);
3357                my ($u, $p, $repo_id) = ($new_url, '', $ref_id);
3358                if ($u =~ s#^\Q$url\E(/|$)##) {
3359                        $p = $u;
3360                        $u = $url;
3361                        $repo_id = $self->{repo_id};
3362                }
3363                while (1) {
3364                        # It is possible to tag two different subdirectories at
3365                        # the same revision.  If the url for an existing ref
3366                        # does not match, we must either find a ref with a
3367                        # matching url or create a new ref by growing a tail.
3368                        $gs = Git::SVN->init($u, $p, $repo_id, $ref_id, 1);
3369                        my (undef, $max_commit) = $gs->rev_map_max(1);
3370                        last if (!$max_commit);
3371                        my ($url) = ::cmt_metadata($max_commit);
3372                        last if ($url eq $gs->metadata_url);
3373                        $ref_id .= '-';
3374                }
3375                print STDERR "Initializing parent: $ref_id\n" unless $::_q > 1;
3376        }
3377        $gs
3378}
3379
3380sub call_authors_prog {
3381        my ($orig_author) = @_;
3382        $orig_author = command_oneline('rev-parse', '--sq-quote', $orig_author);
3383        my $author = `$::_authors_prog $orig_author`;
3384        if ($? != 0) {
3385                die "$::_authors_prog failed with exit code $?\n"
3386        }
3387        if ($author =~ /^\s*(.+?)\s*<(.*)>\s*$/) {
3388                my ($name, $email) = ($1, $2);
3389                $email = undef if length $2 == 0;
3390                return [$name, $email];
3391        } else {
3392                die "Author: $orig_author: $::_authors_prog returned "
3393                        . "invalid author format: $author\n";
3394        }
3395}
3396
3397sub check_author {
3398        my ($author) = @_;
3399        if (!defined $author || length $author == 0) {
3400                $author = '(no author)';
3401        }
3402        if (!defined $::users{$author}) {
3403                if (defined $::_authors_prog) {
3404                        $::users{$author} = call_authors_prog($author);
3405                } elsif (defined $::_authors) {
3406                        die "Author: $author not defined in $::_authors file\n";
3407                }
3408        }
3409        $author;
3410}
3411
3412sub find_extra_svk_parents {
3413        my ($self, $ed, $tickets, $parents) = @_;
3414        # aha!  svk:merge property changed...
3415        my @tickets = split "\n", $tickets;
3416        my @known_parents;
3417        for my $ticket ( @tickets ) {
3418                my ($uuid, $path, $rev) = split /:/, $ticket;
3419                if ( $uuid eq $self->ra_uuid ) {
3420                        my $url = $self->{url};
3421                        my $repos_root = $url;
3422                        my $branch_from = $path;
3423                        $branch_from =~ s{^/}{};
3424                        my $gs = $self->other_gs($repos_root."/".$branch_from,
3425                                                 $url,
3426                                                 $branch_from,
3427                                                 $rev,
3428                                                 $self->{ref_id});
3429                        if ( my $commit = $gs->rev_map_get($rev, $uuid) ) {
3430                                # wahey!  we found it, but it might be
3431                                # an old one (!)
3432                                push @known_parents, [ $rev, $commit ];
3433                        }
3434                }
3435        }
3436        # Ordering matters; highest-numbered commit merge tickets
3437        # first, as they may account for later merge ticket additions
3438        # or changes.
3439        @known_parents = map {$_->[1]} sort {$b->[0] <=> $a->[0]} @known_parents;
3440        for my $parent ( @known_parents ) {
3441                my @cmd = ('rev-list', $parent, map { "^$_" } @$parents );
3442                my ($msg_fh, $ctx) = command_output_pipe(@cmd);
3443                my $new;
3444                while ( <$msg_fh> ) {
3445                        $new=1;last;
3446                }
3447                command_close_pipe($msg_fh, $ctx);
3448                if ( $new ) {
3449                        print STDERR
3450                            "Found merge parent (svk:merge ticket): $parent\n";
3451                        push @$parents, $parent;
3452                }
3453        }
3454}
3455
3456sub lookup_svn_merge {
3457        my $uuid = shift;
3458        my $url = shift;
3459        my $merge = shift;
3460
3461        my ($source, $revs) = split ":", $merge;
3462        my $path = $source;
3463        $path =~ s{^/}{};
3464        my $gs = Git::SVN->find_by_url($url.$source, $url, $path);
3465        if ( !$gs ) {
3466                warn "Couldn't find revmap for $url$source\n";
3467                return;
3468        }
3469        my @ranges = split ",", $revs;
3470        my ($tip, $tip_commit);
3471        my @merged_commit_ranges;
3472        # find the tip
3473        for my $range ( @ranges ) {
3474                my ($bottom, $top) = split "-", $range;
3475                $top ||= $bottom;
3476                my $bottom_commit = $gs->find_rev_after( $bottom, 1, $top );
3477                my $top_commit = $gs->find_rev_before( $top, 1, $bottom );
3478
3479                unless ($top_commit and $bottom_commit) {
3480                        warn "W:unknown path/rev in svn:mergeinfo "
3481                                ."dirprop: $source:$range\n";
3482                        next;
3483                }
3484
3485                if (scalar(command('rev-parse', "$bottom_commit^@"))) {
3486                        push @merged_commit_ranges,
3487                             "$bottom_commit^..$top_commit";
3488                } else {
3489                        push @merged_commit_ranges, "$top_commit";
3490                }
3491
3492                if ( !defined $tip or $top > $tip ) {
3493                        $tip = $top;
3494                        $tip_commit = $top_commit;
3495                }
3496        }
3497        return ($tip_commit, @merged_commit_ranges);
3498}
3499
3500sub _rev_list {
3501        my ($msg_fh, $ctx) = command_output_pipe(
3502                "rev-list", @_,
3503               );
3504        my @rv;
3505        while ( <$msg_fh> ) {
3506                chomp;
3507                push @rv, $_;
3508        }
3509        command_close_pipe($msg_fh, $ctx);
3510        @rv;
3511}
3512
3513sub check_cherry_pick {
3514        my $base = shift;
3515        my $tip = shift;
3516        my $parents = shift;
3517        my @ranges = @_;
3518        my %commits = map { $_ => 1 }
3519                _rev_list("--no-merges", $tip, "--not", $base, @$parents, "--");
3520        for my $range ( @ranges ) {
3521                delete @commits{_rev_list($range, "--")};
3522        }
3523        for my $commit (keys %commits) {
3524                if (has_no_changes($commit)) {
3525                        delete $commits{$commit};
3526                }
3527        }
3528        return (keys %commits);
3529}
3530
3531sub has_no_changes {
3532        my $commit = shift;
3533
3534        my @revs = split / /, command_oneline(
3535                qw(rev-list --parents -1 -m), $commit);
3536
3537        # Commits with no parents, e.g. the start of a partial branch,
3538        # have changes by definition.
3539        return 1 if (@revs < 2);
3540
3541        # Commits with multiple parents, e.g a merge, have no changes
3542        # by definition.
3543        return 0 if (@revs > 2);
3544
3545        return (command_oneline("rev-parse", "$commit^{tree}") eq
3546                command_oneline("rev-parse", "$commit~1^{tree}"));
3547}
3548
3549# The GIT_DIR environment variable is not always set until after the command
3550# line arguments are processed, so we can't memoize in a BEGIN block.
3551{
3552        my $memoized = 0;
3553
3554        sub memoize_svn_mergeinfo_functions {
3555                return if $memoized;
3556                $memoized = 1;
3557
3558                my $cache_path = "$ENV{GIT_DIR}/svn/.caches/";
3559                mkpath([$cache_path]) unless -d $cache_path;
3560
3561                tie my %lookup_svn_merge_cache => 'Memoize::Storable',
3562                    "$cache_path/lookup_svn_merge.db", 'nstore';
3563                memoize 'lookup_svn_merge',
3564                        SCALAR_CACHE => 'FAULT',
3565                        LIST_CACHE => ['HASH' => \%lookup_svn_merge_cache],
3566                ;
3567
3568                tie my %check_cherry_pick_cache => 'Memoize::Storable',
3569                    "$cache_path/check_cherry_pick.db", 'nstore';
3570                memoize 'check_cherry_pick',
3571                        SCALAR_CACHE => 'FAULT',
3572                        LIST_CACHE => ['HASH' => \%check_cherry_pick_cache],
3573                ;
3574
3575                tie my %has_no_changes_cache => 'Memoize::Storable',
3576                    "$cache_path/has_no_changes.db", 'nstore';
3577                memoize 'has_no_changes',
3578                        SCALAR_CACHE => ['HASH' => \%has_no_changes_cache],
3579                        LIST_CACHE => 'FAULT',
3580                ;
3581        }
3582
3583        sub unmemoize_svn_mergeinfo_functions {
3584                return if not $memoized;
3585                $memoized = 0;
3586
3587                Memoize::unmemoize 'lookup_svn_merge';
3588                Memoize::unmemoize 'check_cherry_pick';
3589                Memoize::unmemoize 'has_no_changes';
3590        }
3591
3592        Memoize::memoize 'Git::SVN::repos_root';
3593}
3594
3595END {
3596        # Force cache writeout explicitly instead of waiting for
3597        # global destruction to avoid segfault in Storable:
3598        # http://rt.cpan.org/Public/Bug/Display.html?id=36087
3599        unmemoize_svn_mergeinfo_functions();
3600}
3601
3602sub parents_exclude {
3603        my $parents = shift;
3604        my @commits = @_;
3605        return unless @commits;
3606
3607        my @excluded;
3608        my $excluded;
3609        do {
3610                my @cmd = ('rev-list', "-1", @commits, "--not", @$parents );
3611                $excluded = command_oneline(@cmd);
3612                if ( $excluded ) {
3613                        my @new;
3614                        my $found;
3615                        for my $commit ( @commits ) {
3616                                if ( $commit eq $excluded ) {
3617                                        push @excluded, $commit;
3618                                        $found++;
3619                                        last;
3620                                }
3621                                else {
3622                                        push @new, $commit;
3623                                }
3624                        }
3625                        die "saw commit '$excluded' in rev-list output, "
3626                                ."but we didn't ask for that commit (wanted: @commits --not @$parents)"
3627                                        unless $found;
3628                        @commits = @new;
3629                }
3630        }
3631                while ($excluded and @commits);
3632
3633        return @excluded;
3634}
3635
3636
3637# note: this function should only be called if the various dirprops
3638# have actually changed
3639sub find_extra_svn_parents {
3640        my ($self, $ed, $mergeinfo, $parents) = @_;
3641        # aha!  svk:merge property changed...
3642
3643        memoize_svn_mergeinfo_functions();
3644
3645        # We first search for merged tips which are not in our
3646        # history.  Then, we figure out which git revisions are in
3647        # that tip, but not this revision.  If all of those revisions
3648        # are now marked as merge, we can add the tip as a parent.
3649        my @merges = split "\n", $mergeinfo;
3650        my @merge_tips;
3651        my $url = $self->{url};
3652        my $uuid = $self->ra_uuid;
3653        my %ranges;
3654        for my $merge ( @merges ) {
3655                my ($tip_commit, @ranges) =
3656                        lookup_svn_merge( $uuid, $url, $merge );
3657                unless (!$tip_commit or
3658                                grep { $_ eq $tip_commit } @$parents ) {
3659                        push @merge_tips, $tip_commit;
3660                        $ranges{$tip_commit} = \@ranges;
3661                } else {
3662                        push @merge_tips, undef;
3663                }
3664        }
3665
3666        my %excluded = map { $_ => 1 }
3667                parents_exclude($parents, grep { defined } @merge_tips);
3668
3669        # check merge tips for new parents
3670        my @new_parents;
3671        for my $merge_tip ( @merge_tips ) {
3672                my $spec = shift @merges;
3673                next unless $merge_tip and $excluded{$merge_tip};
3674
3675                my $ranges = $ranges{$merge_tip};
3676
3677                # check out 'new' tips
3678                my $merge_base;
3679                eval {
3680                        $merge_base = command_oneline(
3681                                "merge-base",
3682                                @$parents, $merge_tip,
3683                        );
3684                };
3685                if ($@) {
3686                        die "An error occurred during merge-base"
3687                                unless $@->isa("Git::Error::Command");
3688
3689                        warn "W: Cannot find common ancestor between ".
3690                             "@$parents and $merge_tip. Ignoring merge info.\n";
3691                        next;
3692                }
3693
3694                # double check that there are no missing non-merge commits
3695                my (@incomplete) = check_cherry_pick(
3696                        $merge_base, $merge_tip,
3697                        $parents,
3698                        @$ranges,
3699                       );
3700
3701                if ( @incomplete ) {
3702                        warn "W:svn cherry-pick ignored ($spec) - missing "
3703                                .@incomplete." commit(s) (eg $incomplete[0])\n";
3704                } else {
3705                        warn
3706                                "Found merge parent (svn:mergeinfo prop): ",
3707                                        $merge_tip, "\n";
3708                        push @new_parents, $merge_tip;
3709                }
3710        }
3711
3712        # cater for merges which merge commits from multiple branches
3713        if ( @new_parents > 1 ) {
3714                for ( my $i = 0; $i <= $#new_parents; $i++ ) {
3715                        for ( my $j = 0; $j <= $#new_parents; $j++ ) {
3716                                next if $i == $j;
3717                                next unless $new_parents[$i];
3718                                next unless $new_parents[$j];
3719                                my $revs = command_oneline(
3720                                        "rev-list", "-1",
3721                                        "$new_parents[$i]..$new_parents[$j]",
3722                                       );
3723                                if ( !$revs ) {
3724                                        undef($new_parents[$j]);
3725                                }
3726                        }
3727                }
3728        }
3729        push @$parents, grep { defined } @new_parents;
3730}
3731
3732sub make_log_entry {
3733        my ($self, $rev, $parents, $ed) = @_;
3734        my $untracked = $self->get_untracked($ed);
3735
3736        my @parents = @$parents;
3737        my $ps = $ed->{path_strip} || "";
3738        for my $path ( grep { m/$ps/ } %{$ed->{dir_prop}} ) {
3739                my $props = $ed->{dir_prop}{$path};
3740                if ( $props->{"svk:merge"} ) {
3741                        $self->find_extra_svk_parents
3742                                ($ed, $props->{"svk:merge"}, \@parents);
3743                }
3744                if ( $props->{"svn:mergeinfo"} ) {
3745                        $self->find_extra_svn_parents
3746                                ($ed,
3747                                 $props->{"svn:mergeinfo"},
3748                                 \@parents);
3749                }
3750        }
3751
3752        open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
3753        print $un "r$rev\n" or croak $!;
3754        print $un $_, "\n" foreach @$untracked;
3755        my %log_entry = ( parents => \@parents, revision => $rev,
3756                          log => '');
3757
3758        my $headrev;
3759        my $logged = delete $self->{logged_rev_props};
3760        if (!$logged || $self->{-want_revprops}) {
3761                my $rp = $self->ra->rev_proplist($rev);
3762                foreach (sort keys %$rp) {
3763                        my $v = $rp->{$_};
3764                        if (/^svn:(author|date|log)$/) {
3765                                $log_entry{$1} = $v;
3766                        } elsif ($_ eq 'svm:headrev') {
3767                                $headrev = $v;
3768                        } else {
3769                                print $un "  rev_prop: ", uri_encode($_), ' ',
3770                                          uri_encode($v), "\n";
3771                        }
3772                }
3773        } else {
3774                map { $log_entry{$_} = $logged->{$_} } keys %$logged;
3775        }
3776        close $un or croak $!;
3777
3778        $log_entry{date} = parse_svn_date($log_entry{date});
3779        $log_entry{log} .= "\n";
3780        my $author = $log_entry{author} = check_author($log_entry{author});
3781        my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
3782                                                       : ($author, undef);
3783
3784        my ($commit_name, $commit_email) = ($name, $email);
3785        if ($_use_log_author) {
3786                my $name_field;
3787                if ($log_entry{log} =~ /From:\s+(.*\S)\s*\n/i) {
3788                        $name_field = $1;
3789                } elsif ($log_entry{log} =~ /Signed-off-by:\s+(.*\S)\s*\n/i) {
3790                        $name_field = $1;
3791                }
3792                if (!defined $name_field) {
3793                        if (!defined $email) {
3794                                $email = $name;
3795                        }
3796                } elsif ($name_field =~ /(.*?)\s+<(.*)>/) {
3797                        ($name, $email) = ($1, $2);
3798                } elsif ($name_field =~ /(.*)@/) {
3799                        ($name, $email) = ($1, $name_field);
3800                } else {
3801                        ($name, $email) = ($name_field, $name_field);
3802                }
3803        }
3804        if (defined $headrev && $self->use_svm_props) {
3805                if ($self->rewrite_root) {
3806                        die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
3807                            "options set!\n";
3808                }
3809                if ($self->rewrite_uuid) {
3810                        die "Can't have both 'useSvmProps' and 'rewriteUUID' ",
3811                            "options set!\n";
3812                }
3813                my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$}i;
3814                # we don't want "SVM: initializing mirror for junk" ...
3815                return undef if $r == 0;
3816                my $svm = $self->svm;
3817                if ($uuid ne $svm->{uuid}) {
3818                        die "UUID mismatch on SVM path:\n",
3819                            "expected: $svm->{uuid}\n",
3820                            "     got: $uuid\n";
3821                }
3822                my $full_url = $self->full_url;
3823                $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
3824                             die "Failed to replace '$svm->{replace}' with ",
3825                                 "'$svm->{source}' in $full_url\n";
3826                # throw away username for storing in records
3827                remove_username($full_url);
3828                $log_entry{metadata} = "$full_url\@$r $uuid";
3829                $log_entry{svm_revision} = $r;
3830                $email ||= "$author\@$uuid";
3831                $commit_email ||= "$author\@$uuid";
3832        } elsif ($self->use_svnsync_props) {
3833                my $full_url = $self->svnsync->{url};
3834                $full_url .= "/$self->{path}" if length $self->{path};
3835                remove_username($full_url);
3836                my $uuid = $self->svnsync->{uuid};
3837                $log_entry{metadata} = "$full_url\@$rev $uuid";
3838                $email ||= "$author\@$uuid";
3839                $commit_email ||= "$author\@$uuid";
3840        } else {
3841                my $url = $self->metadata_url;
3842                remove_username($url);
3843                my $uuid = $self->rewrite_uuid || $self->ra->get_uuid;
3844                $log_entry{metadata} = "$url\@$rev " . $uuid;
3845                $email ||= "$author\@" . $uuid;
3846                $commit_email ||= "$author\@" . $uuid;
3847        }
3848        $log_entry{name} = $name;
3849        $log_entry{email} = $email;
3850        $log_entry{commit_name} = $commit_name;
3851        $log_entry{commit_email} = $commit_email;
3852        \%log_entry;
3853}
3854
3855sub fetch {
3856        my ($self, $min_rev, $max_rev, @parents) = @_;
3857        my ($last_rev, $last_commit) = $self->last_rev_commit;
3858        my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
3859        $self->ra->gs_fetch_loop_common($base, $head, [$self]);
3860}
3861
3862sub set_tree_cb {
3863        my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
3864        $self->{inject_parents} = { $rev => $tree };
3865        $self->fetch(undef, undef);
3866}
3867
3868sub set_tree {
3869        my ($self, $tree) = (shift, shift);
3870        my $log_entry = ::get_commit_entry($tree);
3871        unless ($self->{last_rev}) {
3872                ::fatal("Must have an existing revision to commit");
3873        }
3874        my %ed_opts = ( r => $self->{last_rev},
3875                        log => $log_entry->{log},
3876                        ra => $self->ra,
3877                        tree_a => $self->{last_commit},
3878                        tree_b => $tree,
3879                        editor_cb => sub {
3880                               $self->set_tree_cb($log_entry, $tree, @_) },
3881                        svn_path => $self->{path} );
3882        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
3883                print "No changes\nr$self->{last_rev} = $tree\n";
3884        }
3885}
3886
3887sub rebuild_from_rev_db {
3888        my ($self, $path) = @_;
3889        my $r = -1;
3890        open my $fh, '<', $path or croak "open: $!";
3891        binmode $fh or croak "binmode: $!";
3892        while (<$fh>) {
3893                length($_) == 41 or croak "inconsistent size in ($_) != 41";
3894                chomp($_);
3895                ++$r;
3896                next if $_ eq ('0' x 40);
3897                $self->rev_map_set($r, $_);
3898                print "r$r = $_\n";
3899        }
3900        close $fh or croak "close: $!";
3901        unlink $path or croak "unlink: $!";
3902}
3903
3904sub rebuild {
3905        my ($self) = @_;
3906        my $map_path = $self->map_path;
3907        my $partial = (-e $map_path && ! -z $map_path);
3908        return unless ::verify_ref($self->refname.'^0');
3909        if (!$partial && ($self->use_svm_props || $self->no_metadata)) {
3910                my $rev_db = $self->rev_db_path;
3911                $self->rebuild_from_rev_db($rev_db);
3912                if ($self->use_svm_props) {
3913                        my $svm_rev_db = $self->rev_db_path($self->svm_uuid);
3914                        $self->rebuild_from_rev_db($svm_rev_db);
3915                }
3916                $self->unlink_rev_db_symlink;
3917                return;
3918        }
3919        print "Rebuilding $map_path ...\n" if (!$partial);
3920        my ($base_rev, $head) = ($partial ? $self->rev_map_max_norebuild(1) :
3921                (undef, undef));
3922        my ($log, $ctx) =
3923            command_output_pipe(qw/rev-list --pretty=raw --no-color --reverse/,
3924                                ($head ? "$head.." : "") . $self->refname,
3925                                '--');
3926        my $metadata_url = $self->metadata_url;
3927        remove_username($metadata_url);
3928        my $svn_uuid = $self->rewrite_uuid || $self->ra_uuid;
3929        my $c;
3930        while (<$log>) {
3931                if ( m{^commit ($::sha1)$} ) {
3932                        $c = $1;
3933                        next;
3934                }
3935                next unless s{^\s*(git-svn-id:)}{$1};
3936                my ($url, $rev, $uuid) = ::extract_metadata($_);
3937                remove_username($url);
3938
3939                # ignore merges (from set-tree)
3940                next if (!defined $rev || !$uuid);
3941
3942                # if we merged or otherwise started elsewhere, this is
3943                # how we break out of it
3944                if (($uuid ne $svn_uuid) ||
3945                    ($metadata_url && $url && ($url ne $metadata_url))) {
3946                        next;
3947                }
3948                if ($partial && $head) {
3949                        print "Partial-rebuilding $map_path ...\n";
3950                        print "Currently at $base_rev = $head\n";
3951                        $head = undef;
3952                }
3953
3954                $self->rev_map_set($rev, $c);
3955                print "r$rev = $c\n";
3956        }
3957        command_close_pipe($log, $ctx);
3958        print "Done rebuilding $map_path\n" if (!$partial || !$head);
3959        my $rev_db_path = $self->rev_db_path;
3960        if (-f $self->rev_db_path) {
3961                unlink $self->rev_db_path or croak "unlink: $!";
3962        }
3963        $self->unlink_rev_db_symlink;
3964}
3965
3966# rev_map:
3967# Tie::File seems to be prone to offset errors if revisions get sparse,
3968# it's not that fast, either.  Tie::File is also not in Perl 5.6.  So
3969# one of my favorite modules is out :<  Next up would be one of the DBM
3970# modules, but I'm not sure which is most portable...
3971#
3972# This is the replacement for the rev_db format, which was too big
3973# and inefficient for large repositories with a lot of sparse history
3974# (mainly tags)
3975#
3976# The format is this:
3977#   - 24 bytes for every record,
3978#     * 4 bytes for the integer representing an SVN revision number
3979#     * 20 bytes representing the sha1 of a git commit
3980#   - No empty padding records like the old format
3981#     (except the last record, which can be overwritten)
3982#   - new records are written append-only since SVN revision numbers
3983#     increase monotonically
3984#   - lookups on SVN revision number are done via a binary search
3985#   - Piping the file to xxd -c24 is a good way of dumping it for
3986#     viewing or editing (piped back through xxd -r), should the need
3987#     ever arise.
3988#   - The last record can be padding revision with an all-zero sha1
3989#     This is used to optimize fetch performance when using multiple
3990#     "fetch" directives in .git/config
3991#
3992# These files are disposable unless noMetadata or useSvmProps is set
3993
3994sub _rev_map_set {
3995        my ($fh, $rev, $commit) = @_;
3996
3997        binmode $fh or croak "binmode: $!";
3998        my $size = (stat($fh))[7];
3999        ($size % 24) == 0 or croak "inconsistent size: $size";
4000
4001        my $wr_offset = 0;
4002        if ($size > 0) {
4003                sysseek($fh, -24, SEEK_END) or croak "seek: $!";
4004                my $read = sysread($fh, my $buf, 24) or croak "read: $!";
4005                $read == 24 or croak "read only $read bytes (!= 24)";
4006                my ($last_rev, $last_commit) = unpack(rev_map_fmt, $buf);
4007                if ($last_commit eq ('0' x40)) {
4008                        if ($size >= 48) {
4009                                sysseek($fh, -48, SEEK_END) or croak "seek: $!";
4010                                $read = sysread($fh, $buf, 24) or
4011                                    croak "read: $!";
4012                                $read == 24 or
4013                                    croak "read only $read bytes (!= 24)";
4014                                ($last_rev, $last_commit) =
4015                                    unpack(rev_map_fmt, $buf);
4016                                if ($last_commit eq ('0' x40)) {
4017                                        croak "inconsistent .rev_map\n";
4018                                }
4019                        }
4020                        if ($last_rev >= $rev) {
4021                                croak "last_rev is higher!: $last_rev >= $rev";
4022                        }
4023                        $wr_offset = -24;
4024                }
4025        }
4026        sysseek($fh, $wr_offset, SEEK_END) or croak "seek: $!";
4027        syswrite($fh, pack(rev_map_fmt, $rev, $commit), 24) == 24 or
4028          croak "write: $!";
4029}
4030
4031sub _rev_map_reset {
4032        my ($fh, $rev, $commit) = @_;
4033        my $c = _rev_map_get($fh, $rev);
4034        $c eq $commit or die "_rev_map_reset(@_) commit $c does not match!\n";
4035        my $offset = sysseek($fh, 0, SEEK_CUR) or croak "seek: $!";
4036        truncate $fh, $offset or croak "truncate: $!";
4037}
4038
4039sub mkfile {
4040        my ($path) = @_;
4041        unless (-e $path) {
4042                my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
4043                mkpath([$dir]) unless -d $dir;
4044                open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
4045                close $fh or die "Couldn't close (create) $path: $!\n";
4046        }
4047}
4048
4049sub rev_map_set {
4050        my ($self, $rev, $commit, $update_ref, $uuid) = @_;
4051        defined $commit or die "missing arg3\n";
4052        length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
4053        my $db = $self->map_path($uuid);
4054        my $db_lock = "$db.lock";
4055        my $sig;
4056        $update_ref ||= 0;
4057        if ($update_ref) {
4058                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
4059                            $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
4060        }
4061        mkfile($db);
4062
4063        $LOCKFILES{$db_lock} = 1;
4064        my $sync;
4065        # both of these options make our .rev_db file very, very important
4066        # and we can't afford to lose it because rebuild() won't work
4067        if ($self->use_svm_props || $self->no_metadata) {
4068                $sync = 1;
4069                copy($db, $db_lock) or die "rev_map_set(@_): ",
4070                                           "Failed to copy: ",
4071                                           "$db => $db_lock ($!)\n";
4072        } else {
4073                rename $db, $db_lock or die "rev_map_set(@_): ",
4074                                            "Failed to rename: ",
4075                                            "$db => $db_lock ($!)\n";
4076        }
4077
4078        sysopen(my $fh, $db_lock, O_RDWR | O_CREAT)
4079             or croak "Couldn't open $db_lock: $!\n";
4080        $update_ref eq 'reset' ? _rev_map_reset($fh, $rev, $commit) :
4081                                 _rev_map_set($fh, $rev, $commit);
4082        if ($sync) {
4083                $fh->flush or die "Couldn't flush $db_lock: $!\n";
4084                $fh->sync or die "Couldn't sync $db_lock: $!\n";
4085        }
4086        close $fh or croak $!;
4087        if ($update_ref) {
4088                $_head = $self;
4089                my $note = "";
4090                $note = " ($update_ref)" if ($update_ref !~ /^\d*$/);
4091                command_noisy('update-ref', '-m', "r$rev$note",
4092                              $self->refname, $commit);
4093        }
4094        rename $db_lock, $db or die "rev_map_set(@_): ", "Failed to rename: ",
4095                                    "$db_lock => $db ($!)\n";
4096        delete $LOCKFILES{$db_lock};
4097        if ($update_ref) {
4098                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
4099                            $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
4100                kill $sig, $$ if defined $sig;
4101        }
4102}
4103
4104# If want_commit, this will return an array of (rev, commit) where
4105# commit _must_ be a valid commit in the archive.
4106# Otherwise, it'll return the max revision (whether or not the
4107# commit is valid or just a 0x40 placeholder).
4108sub rev_map_max {
4109        my ($self, $want_commit) = @_;
4110        $self->rebuild;
4111        my ($r, $c) = $self->rev_map_max_norebuild($want_commit);
4112        $want_commit ? ($r, $c) : $r;
4113}
4114
4115sub rev_map_max_norebuild {
4116        my ($self, $want_commit) = @_;
4117        my $map_path = $self->map_path;
4118        stat $map_path or return $want_commit ? (0, undef) : 0;
4119        sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
4120        binmode $fh or croak "binmode: $!";
4121        my $size = (stat($fh))[7];
4122        ($size % 24) == 0 or croak "inconsistent size: $size";
4123
4124        if ($size == 0) {
4125                close $fh or croak "close: $!";
4126                return $want_commit ? (0, undef) : 0;
4127        }
4128
4129        sysseek($fh, -24, SEEK_END) or croak "seek: $!";
4130        sysread($fh, my $buf, 24) == 24 or croak "read: $!";
4131        my ($r, $c) = unpack(rev_map_fmt, $buf);
4132        if ($want_commit && $c eq ('0' x40)) {
4133                if ($size < 48) {
4134                        return $want_commit ? (0, undef) : 0;
4135                }
4136                sysseek($fh, -48, SEEK_END) or croak "seek: $!";
4137                sysread($fh, $buf, 24) == 24 or croak "read: $!";
4138                ($r, $c) = unpack(rev_map_fmt, $buf);
4139                if ($c eq ('0'x40)) {
4140                        croak "Penultimate record is all-zeroes in $map_path";
4141                }
4142        }
4143        close $fh or croak "close: $!";
4144        $want_commit ? ($r, $c) : $r;
4145}
4146
4147sub rev_map_get {
4148        my ($self, $rev, $uuid) = @_;
4149        my $map_path = $self->map_path($uuid);
4150        return undef unless -e $map_path;
4151
4152        sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
4153        my $c = _rev_map_get($fh, $rev);
4154        close($fh) or croak "close: $!";
4155        $c
4156}
4157
4158sub _rev_map_get {
4159        my ($fh, $rev) = @_;
4160
4161        binmode $fh or croak "binmode: $!";
4162        my $size = (stat($fh))[7];
4163        ($size % 24) == 0 or croak "inconsistent size: $size";
4164
4165        if ($size == 0) {
4166                return undef;
4167        }
4168
4169        my ($l, $u) = (0, $size - 24);
4170        my ($r, $c, $buf);
4171
4172        while ($l <= $u) {
4173                my $i = int(($l/24 + $u/24) / 2) * 24;
4174                sysseek($fh, $i, SEEK_SET) or croak "seek: $!";
4175                sysread($fh, my $buf, 24) == 24 or croak "read: $!";
4176                my ($r, $c) = unpack(rev_map_fmt, $buf);
4177
4178                if ($r < $rev) {
4179                        $l = $i + 24;
4180                } elsif ($r > $rev) {
4181                        $u = $i - 24;
4182                } else { # $r == $rev
4183                        return $c eq ('0' x 40) ? undef : $c;
4184                }
4185        }
4186        undef;
4187}
4188
4189# Finds the first svn revision that exists on (if $eq_ok is true) or
4190# before $rev for the current branch.  It will not search any lower
4191# than $min_rev.  Returns the git commit hash and svn revision number
4192# if found, else (undef, undef).
4193sub find_rev_before {
4194        my ($self, $rev, $eq_ok, $min_rev) = @_;
4195        --$rev unless $eq_ok;
4196        $min_rev ||= 1;
4197        my $max_rev = $self->rev_map_max;
4198        $rev = $max_rev if ($rev > $max_rev);
4199        while ($rev >= $min_rev) {
4200                if (my $c = $self->rev_map_get($rev)) {
4201                        return ($rev, $c);
4202                }
4203                --$rev;
4204        }
4205        return (undef, undef);
4206}
4207
4208# Finds the first svn revision that exists on (if $eq_ok is true) or
4209# after $rev for the current branch.  It will not search any higher
4210# than $max_rev.  Returns the git commit hash and svn revision number
4211# if found, else (undef, undef).
4212sub find_rev_after {
4213        my ($self, $rev, $eq_ok, $max_rev) = @_;
4214        ++$rev unless $eq_ok;
4215        $max_rev ||= $self->rev_map_max;
4216        while ($rev <= $max_rev) {
4217                if (my $c = $self->rev_map_get($rev)) {
4218                        return ($rev, $c);
4219                }
4220                ++$rev;
4221        }
4222        return (undef, undef);
4223}
4224
4225sub _new {
4226        my ($class, $repo_id, $ref_id, $path) = @_;
4227        unless (defined $repo_id && length $repo_id) {
4228                $repo_id = $Git::SVN::default_repo_id;
4229        }
4230        unless (defined $ref_id && length $ref_id) {
4231                $_prefix = '' unless defined($_prefix);
4232                $_[2] = $ref_id =
4233                             "refs/remotes/$_prefix$Git::SVN::default_ref_id";
4234        }
4235        $_[1] = $repo_id;
4236        my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
4237
4238        # Older repos imported by us used $GIT_DIR/svn/foo instead of
4239        # $GIT_DIR/svn/refs/remotes/foo when tracking refs/remotes/foo
4240        if ($ref_id =~ m{^refs/remotes/(.*)}) {
4241                my $old_dir = "$ENV{GIT_DIR}/svn/$1";
4242                if (-d $old_dir && ! -d $dir) {
4243                        $dir = $old_dir;
4244                }
4245        }
4246
4247        $_[3] = $path = '' unless (defined $path);
4248        mkpath([$dir]);
4249        bless {
4250                ref_id => $ref_id, dir => $dir, index => "$dir/index",
4251                path => $path, config => "$ENV{GIT_DIR}/svn/config",
4252                map_root => "$dir/.rev_map", repo_id => $repo_id }, $class;
4253}
4254
4255# for read-only access of old .rev_db formats
4256sub unlink_rev_db_symlink {
4257        my ($self) = @_;
4258        my $link = $self->rev_db_path;
4259        $link =~ s/\.[\w-]+$// or croak "missing UUID at the end of $link";
4260        if (-l $link) {
4261                unlink $link or croak "unlink: $link failed!";
4262        }
4263}
4264
4265sub rev_db_path {
4266        my ($self, $uuid) = @_;
4267        my $db_path = $self->map_path($uuid);
4268        $db_path =~ s{/\.rev_map\.}{/\.rev_db\.}
4269            or croak "map_path: $db_path does not contain '/.rev_map.' !";
4270        $db_path;
4271}
4272
4273# the new replacement for .rev_db
4274sub map_path {
4275        my ($self, $uuid) = @_;
4276        $uuid ||= $self->ra_uuid;
4277        "$self->{map_root}.$uuid";
4278}
4279
4280sub uri_encode {
4281        my ($f) = @_;
4282        $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
4283        $f
4284}
4285
4286sub uri_decode {
4287        my ($f) = @_;
4288        $f =~ s#%([0-9a-fA-F]{2})#chr(hex($1))#eg;
4289        $f
4290}
4291
4292sub remove_username {
4293        $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
4294}
4295
4296package Git::SVN::Prompt;
4297use strict;
4298use warnings;
4299require SVN::Core;
4300use vars qw/$_no_auth_cache $_username/;
4301
4302sub simple {
4303        my ($cred, $realm, $default_username, $may_save, $pool) = @_;
4304        $may_save = undef if $_no_auth_cache;
4305        $default_username = $_username if defined $_username;
4306        if (defined $default_username && length $default_username) {
4307                if (defined $realm && length $realm) {
4308                        print STDERR "Authentication realm: $realm\n";
4309                        STDERR->flush;
4310                }
4311                $cred->username($default_username);
4312        } else {
4313                username($cred, $realm, $may_save, $pool);
4314        }
4315        $cred->password(_read_password("Password for '" .
4316                                       $cred->username . "': ", $realm));
4317        $cred->may_save($may_save);
4318        $SVN::_Core::SVN_NO_ERROR;
4319}
4320
4321sub ssl_server_trust {
4322        my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
4323        $may_save = undef if $_no_auth_cache;
4324        print STDERR "Error validating server certificate for '$realm':\n";
4325        {
4326                no warnings 'once';
4327                # All variables SVN::Auth::SSL::* are used only once,
4328                # so we're shutting up Perl warnings about this.
4329                if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
4330                        print STDERR " - The certificate is not issued ",
4331                            "by a trusted authority. Use the\n",
4332                            "   fingerprint to validate ",
4333                            "the certificate manually!\n";
4334                }
4335                if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
4336                        print STDERR " - The certificate hostname ",
4337                            "does not match.\n";
4338                }
4339                if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
4340                        print STDERR " - The certificate is not yet valid.\n";
4341                }
4342                if ($failures & $SVN::Auth::SSL::EXPIRED) {
4343                        print STDERR " - The certificate has expired.\n";
4344                }
4345                if ($failures & $SVN::Auth::SSL::OTHER) {
4346                        print STDERR " - The certificate has ",
4347                            "an unknown error.\n";
4348                }
4349        } # no warnings 'once'
4350        printf STDERR
4351                "Certificate information:\n".
4352                " - Hostname: %s\n".
4353                " - Valid: from %s until %s\n".
4354                " - Issuer: %s\n".
4355                " - Fingerprint: %s\n",
4356                map $cert_info->$_, qw(hostname valid_from valid_until
4357                                       issuer_dname fingerprint);
4358        my $choice;
4359prompt:
4360        print STDERR $may_save ?
4361              "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
4362              "(R)eject or accept (t)emporarily? ";
4363        STDERR->flush;
4364        $choice = lc(substr(<STDIN> || 'R', 0, 1));
4365        if ($choice =~ /^t$/i) {
4366                $cred->may_save(undef);
4367        } elsif ($choice =~ /^r$/i) {
4368                return -1;
4369        } elsif ($may_save && $choice =~ /^p$/i) {
4370                $cred->may_save($may_save);
4371        } else {
4372                goto prompt;
4373        }
4374        $cred->accepted_failures($failures);
4375        $SVN::_Core::SVN_NO_ERROR;
4376}
4377
4378sub ssl_client_cert {
4379        my ($cred, $realm, $may_save, $pool) = @_;
4380        $may_save = undef if $_no_auth_cache;
4381        print STDERR "Client certificate filename: ";
4382        STDERR->flush;
4383        chomp(my $filename = <STDIN>);
4384        $cred->cert_file($filename);
4385        $cred->may_save($may_save);
4386        $SVN::_Core::SVN_NO_ERROR;
4387}
4388
4389sub ssl_client_cert_pw {
4390        my ($cred, $realm, $may_save, $pool) = @_;
4391        $may_save = undef if $_no_auth_cache;
4392        $cred->password(_read_password("Password: ", $realm));
4393        $cred->may_save($may_save);
4394        $SVN::_Core::SVN_NO_ERROR;
4395}
4396
4397sub username {
4398        my ($cred, $realm, $may_save, $pool) = @_;
4399        $may_save = undef if $_no_auth_cache;
4400        if (defined $realm && length $realm) {
4401                print STDERR "Authentication realm: $realm\n";
4402        }
4403        my $username;
4404        if (defined $_username) {
4405                $username = $_username;
4406        } else {
4407                print STDERR "Username: ";
4408                STDERR->flush;
4409                chomp($username = <STDIN>);
4410        }
4411        $cred->username($username);
4412        $cred->may_save($may_save);
4413        $SVN::_Core::SVN_NO_ERROR;
4414}
4415
4416sub _read_password {
4417        my ($prompt, $realm) = @_;
4418        my $password = '';
4419        if (exists $ENV{GIT_ASKPASS}) {
4420                open(PH, "-|", $ENV{GIT_ASKPASS}, $prompt);
4421                $password = <PH>;
4422                $password =~ s/[\012\015]//; # \n\r
4423                close(PH);
4424        } else {
4425                print STDERR $prompt;
4426                STDERR->flush;
4427                require Term::ReadKey;
4428                Term::ReadKey::ReadMode('noecho');
4429                while (defined(my $key = Term::ReadKey::ReadKey(0))) {
4430                        last if $key =~ /[\012\015]/; # \n\r
4431                        $password .= $key;
4432                }
4433                Term::ReadKey::ReadMode('restore');
4434                print STDERR "\n";
4435                STDERR->flush;
4436        }
4437        $password;
4438}
4439
4440package SVN::Git::Fetcher;
4441use vars qw/@ISA $_ignore_regex $_preserve_empty_dirs $_placeholder_filename
4442            @deleted_gpath %added_placeholder $repo_id/;
4443use strict;
4444use warnings;
4445use Carp qw/croak/;
4446use File::Basename qw/dirname/;
4447use IO::File qw//;
4448
4449# file baton members: path, mode_a, mode_b, pool, fh, blob, base
4450sub new {
4451        my ($class, $git_svn, $switch_path) = @_;
4452        my $self = SVN::Delta::Editor->new;
4453        bless $self, $class;
4454        if (exists $git_svn->{last_commit}) {
4455                $self->{c} = $git_svn->{last_commit};
4456                $self->{empty_symlinks} =
4457                                  _mark_empty_symlinks($git_svn, $switch_path);
4458        }
4459
4460        # some options are read globally, but can be overridden locally
4461        # per [svn-remote "..."] section.  Command-line options will *NOT*
4462        # override options set in an [svn-remote "..."] section
4463        $repo_id = $git_svn->{repo_id};
4464        my $k = "svn-remote.$repo_id.ignore-paths";
4465        my $v = eval { command_oneline('config', '--get', $k) };
4466        $self->{ignore_regex} = $v;
4467
4468        $k = "svn-remote.$repo_id.preserve-empty-dirs";
4469        $v = eval { command_oneline('config', '--get', '--bool', $k) };
4470        if ($v && $v eq 'true') {
4471                $_preserve_empty_dirs = 1;
4472                $k = "svn-remote.$repo_id.placeholder-filename";
4473                $v = eval { command_oneline('config', '--get', $k) };
4474                $_placeholder_filename = $v;
4475        }
4476
4477        # Load the list of placeholder files added during previous invocations.
4478        $k = "svn-remote.$repo_id.added-placeholder";
4479        $v = eval { command_oneline('config', '--get-all', $k) };
4480        if ($_preserve_empty_dirs && $v) {
4481                # command() prints errors to stderr, so we only call it if
4482                # command_oneline() succeeded.
4483                my @v = command('config', '--get-all', $k);
4484                $added_placeholder{ dirname($_) } = $_ foreach @v;
4485        }
4486
4487        $self->{empty} = {};
4488        $self->{dir_prop} = {};
4489        $self->{file_prop} = {};
4490        $self->{absent_dir} = {};
4491        $self->{absent_file} = {};
4492        $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
4493        $self->{pathnameencoding} = Git::config('svn.pathnameencoding');
4494        $self;
4495}
4496
4497# this uses the Ra object, so it must be called before do_{switch,update},
4498# not inside them (when the Git::SVN::Fetcher object is passed) to
4499# do_{switch,update}
4500sub _mark_empty_symlinks {
4501        my ($git_svn, $switch_path) = @_;
4502        my $bool = Git::config_bool('svn.brokenSymlinkWorkaround');
4503        return {} if (!defined($bool)) || (defined($bool) && ! $bool);
4504
4505        my %ret;
4506        my ($rev, $cmt) = $git_svn->last_rev_commit;
4507        return {} unless ($rev && $cmt);
4508
4509        # allow the warning to be printed for each revision we fetch to
4510        # ensure the user sees it.  The user can also disable the workaround
4511        # on the repository even while git svn is running and the next
4512        # revision fetched will skip this expensive function.
4513        my $printed_warning;
4514        chomp(my $empty_blob = `git hash-object -t blob --stdin < /dev/null`);
4515        my ($ls, $ctx) = command_output_pipe(qw/ls-tree -r -z/, $cmt);
4516        local $/ = "\0";
4517        my $pfx = defined($switch_path) ? $switch_path : $git_svn->{path};
4518        $pfx .= '/' if length($pfx);
4519        while (<$ls>) {
4520                chomp;
4521                s/\A100644 blob $empty_blob\t//o or next;
4522                unless ($printed_warning) {
4523                        print STDERR "Scanning for empty symlinks, ",
4524                                     "this may take a while if you have ",
4525                                     "many empty files\n",
4526                                     "You may disable this with `",
4527                                     "git config svn.brokenSymlinkWorkaround ",
4528                                     "false'.\n",
4529                                     "This may be done in a different ",
4530                                     "terminal without restarting ",
4531                                     "git svn\n";
4532                        $printed_warning = 1;
4533                }
4534                my $path = $_;
4535                my (undef, $props) =
4536                               $git_svn->ra->get_file($pfx.$path, $rev, undef);
4537                if ($props->{'svn:special'}) {
4538                        $ret{$path} = 1;
4539                }
4540        }
4541        command_close_pipe($ls, $ctx);
4542        \%ret;
4543}
4544
4545# returns true if a given path is inside a ".git" directory
4546sub in_dot_git {
4547        $_[0] =~ m{(?:^|/)\.git(?:/|$)};
4548}
4549
4550# return value: 0 -- don't ignore, 1 -- ignore
4551sub is_path_ignored {
4552        my ($self, $path) = @_;
4553        return 1 if in_dot_git($path);
4554        return 1 if defined($self->{ignore_regex}) &&
4555                    $path =~ m!$self->{ignore_regex}!;
4556        return 0 unless defined($_ignore_regex);
4557        return 1 if $path =~ m!$_ignore_regex!o;
4558        return 0;
4559}
4560
4561sub set_path_strip {
4562        my ($self, $path) = @_;
4563        $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
4564}
4565
4566sub open_root {
4567        { path => '' };
4568}
4569
4570sub open_directory {
4571        my ($self, $path, $pb, $rev) = @_;
4572        { path => $path };
4573}
4574
4575sub git_path {
4576        my ($self, $path) = @_;
4577        if (my $enc = $self->{pathnameencoding}) {
4578                require Encode;
4579                Encode::from_to($path, 'UTF-8', $enc);
4580        }
4581        if ($self->{path_strip}) {
4582                $path =~ s!$self->{path_strip}!! or
4583                  die "Failed to strip path '$path' ($self->{path_strip})\n";
4584        }
4585        $path;
4586}
4587
4588sub delete_entry {
4589        my ($self, $path, $rev, $pb) = @_;
4590        return undef if $self->is_path_ignored($path);
4591
4592        my $gpath = $self->git_path($path);
4593        return undef if ($gpath eq '');
4594
4595        # remove entire directories.
4596        my ($tree) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
4597                         =~ /\A040000 tree ([a-f\d]{40})\t\Q$gpath\E\0/);
4598        if ($tree) {
4599                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
4600                                                     -r --name-only -z/,
4601                                                     $tree);
4602                local $/ = "\0";
4603                while (<$ls>) {
4604                        chomp;
4605                        my $rmpath = "$gpath/$_";
4606                        $self->{gii}->remove($rmpath);
4607                        print "\tD\t$rmpath\n" unless $::_q;
4608                }
4609                print "\tD\t$gpath/\n" unless $::_q;
4610                command_close_pipe($ls, $ctx);
4611        } else {
4612                $self->{gii}->remove($gpath);
4613                print "\tD\t$gpath\n" unless $::_q;
4614        }
4615        # Don't add to @deleted_gpath if we're deleting a placeholder file.
4616        push @deleted_gpath, $gpath unless $added_placeholder{dirname($path)};
4617        $self->{empty}->{$path} = 0;
4618        undef;
4619}
4620
4621sub open_file {
4622        my ($self, $path, $pb, $rev) = @_;
4623        my ($mode, $blob);
4624
4625        goto out if $self->is_path_ignored($path);
4626
4627        my $gpath = $self->git_path($path);
4628        ($mode, $blob) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
4629                             =~ /\A(\d{6}) blob ([a-f\d]{40})\t\Q$gpath\E\0/);
4630        unless (defined $mode && defined $blob) {
4631                die "$path was not found in commit $self->{c} (r$rev)\n";
4632        }
4633        if ($mode eq '100644' && $self->{empty_symlinks}->{$path}) {
4634                $mode = '120000';
4635        }
4636out:
4637        { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
4638          pool => SVN::Pool->new, action => 'M' };
4639}
4640
4641sub add_file {
4642        my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
4643        my $mode;
4644
4645        if (!$self->is_path_ignored($path)) {
4646                my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
4647                delete $self->{empty}->{$dir};
4648                $mode = '100644';
4649
4650                if ($added_placeholder{$dir}) {
4651                        # Remove our placeholder file, if we created one.
4652                        delete_entry($self, $added_placeholder{$dir})
4653                                unless $path eq $added_placeholder{$dir};
4654                        delete $added_placeholder{$dir}
4655                }
4656        }
4657
4658        { path => $path, mode_a => $mode, mode_b => $mode,
4659          pool => SVN::Pool->new, action => 'A' };
4660}
4661
4662sub add_directory {
4663        my ($self, $path, $cp_path, $cp_rev) = @_;
4664        goto out if $self->is_path_ignored($path);
4665        my $gpath = $self->git_path($path);
4666        if ($gpath eq '') {
4667                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
4668                                                     -r --name-only -z/,
4669                                                     $self->{c});
4670                local $/ = "\0";
4671                while (<$ls>) {
4672                        chomp;
4673                        $self->{gii}->remove($_);
4674                        print "\tD\t$_\n" unless $::_q;
4675                        push @deleted_gpath, $gpath;
4676                }
4677                command_close_pipe($ls, $ctx);
4678                $self->{empty}->{$path} = 0;
4679        }
4680        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
4681        delete $self->{empty}->{$dir};
4682        $self->{empty}->{$path} = 1;
4683
4684        if ($added_placeholder{$dir}) {
4685                # Remove our placeholder file, if we created one.
4686                delete_entry($self, $added_placeholder{$dir});
4687                delete $added_placeholder{$dir}
4688        }
4689
4690out:
4691        { path => $path };
4692}
4693
4694sub change_dir_prop {
4695        my ($self, $db, $prop, $value) = @_;
4696        return undef if $self->is_path_ignored($db->{path});
4697        $self->{dir_prop}->{$db->{path}} ||= {};
4698        $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
4699        undef;
4700}
4701
4702sub absent_directory {
4703        my ($self, $path, $pb) = @_;
4704        return undef if $self->is_path_ignored($path);
4705        $self->{absent_dir}->{$pb->{path}} ||= [];
4706        push @{$self->{absent_dir}->{$pb->{path}}}, $path;
4707        undef;
4708}
4709
4710sub absent_file {
4711        my ($self, $path, $pb) = @_;
4712        return undef if $self->is_path_ignored($path);
4713        $self->{absent_file}->{$pb->{path}} ||= [];
4714        push @{$self->{absent_file}->{$pb->{path}}}, $path;
4715        undef;
4716}
4717
4718sub change_file_prop {
4719        my ($self, $fb, $prop, $value) = @_;
4720        return undef if $self->is_path_ignored($fb->{path});
4721        if ($prop eq 'svn:executable') {
4722                if ($fb->{mode_b} != 120000) {
4723                        $fb->{mode_b} = defined $value ? 100755 : 100644;
4724                }
4725        } elsif ($prop eq 'svn:special') {
4726                $fb->{mode_b} = defined $value ? 120000 : 100644;
4727        } else {
4728                $self->{file_prop}->{$fb->{path}} ||= {};
4729                $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
4730        }
4731        undef;
4732}
4733
4734sub apply_textdelta {
4735        my ($self, $fb, $exp) = @_;
4736        return undef if $self->is_path_ignored($fb->{path});
4737        my $fh = $::_repository->temp_acquire('svn_delta');
4738        # $fh gets auto-closed() by SVN::TxDelta::apply(),
4739        # (but $base does not,) so dup() it for reading in close_file
4740        open my $dup, '<&', $fh or croak $!;
4741        my $base = $::_repository->temp_acquire('git_blob');
4742
4743        if ($fb->{blob}) {
4744                my ($base_is_link, $size);
4745
4746                if ($fb->{mode_a} eq '120000' &&
4747                    ! $self->{empty_symlinks}->{$fb->{path}}) {
4748                        print $base 'link ' or die "print $!\n";
4749                        $base_is_link = 1;
4750                }
4751        retry:
4752                $size = $::_repository->cat_blob($fb->{blob}, $base);
4753                die "Failed to read object $fb->{blob}" if ($size < 0);
4754
4755                if (defined $exp) {
4756                        seek $base, 0, 0 or croak $!;
4757                        my $got = ::md5sum($base);
4758                        if ($got ne $exp) {
4759                                my $err = "Checksum mismatch: ".
4760                                       "$fb->{path} $fb->{blob}\n" .
4761                                       "expected: $exp\n" .
4762                                       "     got: $got\n";
4763                                if ($base_is_link) {
4764                                        warn $err,
4765                                             "Retrying... (possibly ",
4766                                             "a bad symlink from SVN)\n";
4767                                        $::_repository->temp_reset($base);
4768                                        $base_is_link = 0;
4769                                        goto retry;
4770                                }
4771                                die $err;
4772                        }
4773                }
4774        }
4775        seek $base, 0, 0 or croak $!;
4776        $fb->{fh} = $fh;
4777        $fb->{base} = $base;
4778        [ SVN::TxDelta::apply($base, $dup, undef, $fb->{path}, $fb->{pool}) ];
4779}
4780
4781sub close_file {
4782        my ($self, $fb, $exp) = @_;
4783        return undef if $self->is_path_ignored($fb->{path});
4784
4785        my $hash;
4786        my $path = $self->git_path($fb->{path});
4787        if (my $fh = $fb->{fh}) {
4788                if (defined $exp) {
4789                        seek($fh, 0, 0) or croak $!;
4790                        my $got = ::md5sum($fh);
4791                        if ($got ne $exp) {
4792                                die "Checksum mismatch: $path\n",
4793                                    "expected: $exp\n    got: $got\n";
4794                        }
4795                }
4796                if ($fb->{mode_b} == 120000) {
4797                        sysseek($fh, 0, 0) or croak $!;
4798                        my $rd = sysread($fh, my $buf, 5);
4799
4800                        if (!defined $rd) {
4801                                croak "sysread: $!\n";
4802                        } elsif ($rd == 0) {
4803                                warn "$path has mode 120000",
4804                                     " but it points to nothing\n",
4805                                     "converting to an empty file with mode",
4806                                     " 100644\n";
4807                                $fb->{mode_b} = '100644';
4808                        } elsif ($buf ne 'link ') {
4809                                warn "$path has mode 120000",
4810                                     " but is not a link\n";
4811                        } else {
4812                                my $tmp_fh = $::_repository->temp_acquire(
4813                                        'svn_hash');
4814                                my $res;
4815                                while ($res = sysread($fh, my $str, 1024)) {
4816                                        my $out = syswrite($tmp_fh, $str, $res);
4817                                        defined($out) && $out == $res
4818                                                or croak("write ",
4819                                                        Git::temp_path($tmp_fh),
4820                                                        ": $!\n");
4821                                }
4822                                defined $res or croak $!;
4823
4824                                ($fh, $tmp_fh) = ($tmp_fh, $fh);
4825                                Git::temp_release($tmp_fh, 1);
4826                        }
4827                }
4828
4829                $hash = $::_repository->hash_and_insert_object(
4830                                Git::temp_path($fh));
4831                $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
4832
4833                Git::temp_release($fb->{base}, 1);
4834                Git::temp_release($fh, 1);
4835        } else {
4836                $hash = $fb->{blob} or die "no blob information\n";
4837        }
4838        $fb->{pool}->clear;
4839        $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
4840        print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
4841        undef;
4842}
4843
4844sub abort_edit {
4845        my $self = shift;
4846        $self->{nr} = $self->{gii}->{nr};
4847        delete $self->{gii};
4848        $self->SUPER::abort_edit(@_);
4849}
4850
4851sub close_edit {
4852        my $self = shift;
4853
4854        if ($_preserve_empty_dirs) {
4855                my @empty_dirs;
4856
4857                # Any entry flagged as empty that also has an associated
4858                # dir_prop represents a newly created empty directory.
4859                foreach my $i (keys %{$self->{empty}}) {
4860                        push @empty_dirs, $i if exists $self->{dir_prop}->{$i};
4861                }
4862
4863                # Search for directories that have become empty due subsequent
4864                # file deletes.
4865                push @empty_dirs, $self->find_empty_directories();
4866
4867                # Finally, add a placeholder file to each empty directory.
4868                $self->add_placeholder_file($_) foreach (@empty_dirs);
4869
4870                $self->stash_placeholder_list();
4871        }
4872
4873        $self->{git_commit_ok} = 1;
4874        $self->{nr} = $self->{gii}->{nr};
4875        delete $self->{gii};
4876        $self->SUPER::close_edit(@_);
4877}
4878
4879sub find_empty_directories {
4880        my ($self) = @_;
4881        my @empty_dirs;
4882        my %dirs = map { dirname($_) => 1 } @deleted_gpath;
4883
4884        foreach my $dir (sort keys %dirs) {
4885                next if $dir eq ".";
4886
4887                # If there have been any additions to this directory, there is
4888                # no reason to check if it is empty.
4889                my $skip_added = 0;
4890                foreach my $t (qw/dir_prop file_prop/) {
4891                        foreach my $path (keys %{ $self->{$t} }) {
4892                                if (exists $self->{$t}->{dirname($path)}) {
4893                                        $skip_added = 1;
4894                                        last;
4895                                }
4896                        }
4897                        last if $skip_added;
4898                }
4899                next if $skip_added;
4900
4901                # Use `git ls-tree` to get the filenames of this directory
4902                # that existed prior to this particular commit.
4903                my $ls = command('ls-tree', '-z', '--name-only',
4904                                 $self->{c}, "$dir/");
4905                my %files = map { $_ => 1 } split(/\0/, $ls);
4906
4907                # Remove the filenames that were deleted during this commit.
4908                delete $files{$_} foreach (@deleted_gpath);
4909
4910                # Report the directory if there are no filenames left.
4911                push @empty_dirs, $dir unless (scalar %files);
4912        }
4913        @empty_dirs;
4914}
4915
4916sub add_placeholder_file {
4917        my ($self, $dir) = @_;
4918        my $path = "$dir/$_placeholder_filename";
4919        my $gpath = $self->git_path($path);
4920
4921        my $fh = $::_repository->temp_acquire($gpath);
4922        my $hash = $::_repository->hash_and_insert_object(Git::temp_path($fh));
4923        Git::temp_release($fh, 1);
4924        $self->{gii}->update('100644', $hash, $gpath) or croak $!;
4925
4926        # The directory should no longer be considered empty.
4927        delete $self->{empty}->{$dir} if exists $self->{empty}->{$dir};
4928
4929        # Keep track of any placeholder files we create.
4930        $added_placeholder{$dir} = $path;
4931}
4932
4933sub stash_placeholder_list {
4934        my ($self) = @_;
4935        my $k = "svn-remote.$repo_id.added-placeholder";
4936        my $v = eval { command_oneline('config', '--get-all', $k) };
4937        command_noisy('config', '--unset-all', $k) if $v;
4938        foreach (values %added_placeholder) {
4939                command_noisy('config', '--add', $k, $_);
4940        }
4941}
4942
4943package SVN::Git::Editor;
4944use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
4945use strict;
4946use warnings;
4947use Carp qw/croak/;
4948use IO::File;
4949
4950sub new {
4951        my ($class, $opts) = @_;
4952        foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
4953                die "$_ required!\n" unless (defined $opts->{$_});
4954        }
4955
4956        my $pool = SVN::Pool->new;
4957        my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
4958        my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
4959                                     $opts->{r}, $mods);
4960
4961        # $opts->{ra} functions should not be used after this:
4962        my @ce  = $opts->{ra}->get_commit_editor($opts->{log},
4963                                                $opts->{editor_cb}, $pool);
4964        my $self = SVN::Delta::Editor->new(@ce, $pool);
4965        bless $self, $class;
4966        foreach (qw/svn_path r tree_a tree_b/) {
4967                $self->{$_} = $opts->{$_};
4968        }
4969        $self->{url} = $opts->{ra}->{url};
4970        $self->{mods} = $mods;
4971        $self->{types} = $types;
4972        $self->{pool} = $pool;
4973        $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
4974        $self->{rm} = { };
4975        $self->{path_prefix} = length $self->{svn_path} ?
4976                               "$self->{svn_path}/" : '';
4977        $self->{config} = $opts->{config};
4978        $self->{mergeinfo} = $opts->{mergeinfo};
4979        return $self;
4980}
4981
4982sub generate_diff {
4983        my ($tree_a, $tree_b) = @_;
4984        my @diff_tree = qw(diff-tree -z -r);
4985        if ($_cp_similarity) {
4986                push @diff_tree, "-C$_cp_similarity";
4987        } else {
4988                push @diff_tree, '-C';
4989        }
4990        push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
4991        push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
4992        push @diff_tree, $tree_a, $tree_b;
4993        my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
4994        local $/ = "\0";
4995        my $state = 'meta';
4996        my @mods;
4997        while (<$diff_fh>) {
4998                chomp $_; # this gets rid of the trailing "\0"
4999                if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
5000                                        ($::sha1)\s($::sha1)\s
5001                                        ([MTCRAD])\d*$/xo) {
5002                        push @mods, {   mode_a => $1, mode_b => $2,
5003                                        sha1_a => $3, sha1_b => $4,
5004                                        chg => $5 };
5005                        if ($5 =~ /^(?:C|R)$/) {
5006                                $state = 'file_a';
5007                        } else {
5008                                $state = 'file_b';
5009                        }
5010                } elsif ($state eq 'file_a') {
5011                        my $x = $mods[$#mods] or croak "Empty array\n";
5012                        if ($x->{chg} !~ /^(?:C|R)$/) {
5013                                croak "Error parsing $_, $x->{chg}\n";
5014                        }
5015                        $x->{file_a} = $_;
5016                        $state = 'file_b';
5017                } elsif ($state eq 'file_b') {
5018                        my $x = $mods[$#mods] or croak "Empty array\n";
5019                        if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
5020                                croak "Error parsing $_, $x->{chg}\n";
5021                        }
5022                        if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
5023                                croak "Error parsing $_, $x->{chg}\n";
5024                        }
5025                        $x->{file_b} = $_;
5026                        $state = 'meta';
5027                } else {
5028                        croak "Error parsing $_\n";
5029                }
5030        }
5031        command_close_pipe($diff_fh, $ctx);
5032        \@mods;
5033}
5034
5035sub check_diff_paths {
5036        my ($ra, $pfx, $rev, $mods) = @_;
5037        my %types;
5038        $pfx .= '/' if length $pfx;
5039
5040        sub type_diff_paths {
5041                my ($ra, $types, $path, $rev) = @_;
5042                my @p = split m#/+#, $path;
5043                my $c = shift @p;
5044                unless (defined $types->{$c}) {
5045                        $types->{$c} = $ra->check_path($c, $rev);
5046                }
5047                while (@p) {
5048                        $c .= '/' . shift @p;
5049                        next if defined $types->{$c};
5050                        $types->{$c} = $ra->check_path($c, $rev);
5051                }
5052        }
5053
5054        foreach my $m (@$mods) {
5055                foreach my $f (qw/file_a file_b/) {
5056                        next unless defined $m->{$f};
5057                        my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
5058                        if (length $pfx.$dir && ! defined $types{$dir}) {
5059                                type_diff_paths($ra, \%types, $pfx.$dir, $rev);
5060                        }
5061                }
5062        }
5063        \%types;
5064}
5065
5066sub split_path {
5067        return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
5068}
5069
5070sub repo_path {
5071        my ($self, $path) = @_;
5072        if (my $enc = $self->{pathnameencoding}) {
5073                require Encode;
5074                Encode::from_to($path, $enc, 'UTF-8');
5075        }
5076        $self->{path_prefix}.(defined $path ? $path : '');
5077}
5078
5079sub url_path {
5080        my ($self, $path) = @_;
5081        if ($self->{url} =~ m#^https?://#) {
5082                $path =~ s!([^~a-zA-Z0-9_./-])!uc sprintf("%%%02x",ord($1))!eg;
5083        }
5084        $self->{url} . '/' . $self->repo_path($path);
5085}
5086
5087sub rmdirs {
5088        my ($self) = @_;
5089        my $rm = $self->{rm};
5090        delete $rm->{''}; # we never delete the url we're tracking
5091        return unless %$rm;
5092
5093        foreach (keys %$rm) {
5094                my @d = split m#/#, $_;
5095                my $c = shift @d;
5096                $rm->{$c} = 1;
5097                while (@d) {
5098                        $c .= '/' . shift @d;
5099                        $rm->{$c} = 1;
5100                }
5101        }
5102        delete $rm->{$self->{svn_path}};
5103        delete $rm->{''}; # we never delete the url we're tracking
5104        return unless %$rm;
5105
5106        my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
5107                                             $self->{tree_b});
5108        local $/ = "\0";
5109        while (<$fh>) {
5110                chomp;
5111                my @dn = split m#/#, $_;
5112                while (pop @dn) {
5113                        delete $rm->{join '/', @dn};
5114                }
5115                unless (%$rm) {
5116                        close $fh;
5117                        return;
5118                }
5119        }
5120        command_close_pipe($fh, $ctx);
5121
5122        my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
5123        foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
5124                $self->close_directory($bat->{$d}, $p);
5125                my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
5126                print "\tD+\t$d/\n" unless $::_q;
5127                $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
5128                delete $bat->{$d};
5129        }
5130}
5131
5132sub open_or_add_dir {
5133        my ($self, $full_path, $baton) = @_;
5134        my $t = $self->{types}->{$full_path};
5135        if (!defined $t) {
5136                die "$full_path not known in r$self->{r} or we have a bug!\n";
5137        }
5138        {
5139                no warnings 'once';
5140                # SVN::Node::none and SVN::Node::file are used only once,
5141                # so we're shutting up Perl's warnings about them.
5142                if ($t == $SVN::Node::none) {
5143                        return $self->add_directory($full_path, $baton,
5144                            undef, -1, $self->{pool});
5145                } elsif ($t == $SVN::Node::dir) {
5146                        return $self->open_directory($full_path, $baton,
5147                            $self->{r}, $self->{pool});
5148                } # no warnings 'once'
5149                print STDERR "$full_path already exists in repository at ",
5150                    "r$self->{r} and it is not a directory (",
5151                    ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
5152        } # no warnings 'once'
5153        exit 1;
5154}
5155
5156sub ensure_path {
5157        my ($self, $path) = @_;
5158        my $bat = $self->{bat};
5159        my $repo_path = $self->repo_path($path);
5160        return $bat->{''} unless (length $repo_path);
5161        my @p = split m#/+#, $repo_path;
5162        my $c = shift @p;
5163        $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
5164        while (@p) {
5165                my $c0 = $c;
5166                $c .= '/' . shift @p;
5167                $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
5168        }
5169        return $bat->{$c};
5170}
5171
5172# Subroutine to convert a globbing pattern to a regular expression.
5173# From perl cookbook.
5174sub glob2pat {
5175        my $globstr = shift;
5176        my %patmap = ('*' => '.*', '?' => '.', '[' => '[', ']' => ']');
5177        $globstr =~ s{(.)} { $patmap{$1} || "\Q$1" }ge;
5178        return '^' . $globstr . '$';
5179}
5180
5181sub check_autoprop {
5182        my ($self, $pattern, $properties, $file, $fbat) = @_;
5183        # Convert the globbing pattern to a regular expression.
5184        my $regex = glob2pat($pattern);
5185        # Check if the pattern matches the file name.
5186        if($file =~ m/($regex)/) {
5187                # Parse the list of properties to set.
5188                my @props = split(/;/, $properties);
5189                foreach my $prop (@props) {
5190                        # Parse 'name=value' syntax and set the property.
5191                        if ($prop =~ /([^=]+)=(.*)/) {
5192                                my ($n,$v) = ($1,$2);
5193                                for ($n, $v) {
5194                                        s/^\s+//; s/\s+$//;
5195                                }
5196                                $self->change_file_prop($fbat, $n, $v);
5197                        }
5198                }
5199        }
5200}
5201
5202sub apply_autoprops {
5203        my ($self, $file, $fbat) = @_;
5204        my $conf_t = ${$self->{config}}{'config'};
5205        no warnings 'once';
5206        # Check [miscellany]/enable-auto-props in svn configuration.
5207        if (SVN::_Core::svn_config_get_bool(
5208                $conf_t,
5209                $SVN::_Core::SVN_CONFIG_SECTION_MISCELLANY,
5210                $SVN::_Core::SVN_CONFIG_OPTION_ENABLE_AUTO_PROPS,
5211                0)) {
5212                # Auto-props are enabled.  Enumerate them to look for matches.
5213                my $callback = sub {
5214                        $self->check_autoprop($_[0], $_[1], $file, $fbat);
5215                };
5216                SVN::_Core::svn_config_enumerate(
5217                        $conf_t,
5218                        $SVN::_Core::SVN_CONFIG_SECTION_AUTO_PROPS,
5219                        $callback);
5220        }
5221}
5222
5223sub A {
5224        my ($self, $m) = @_;
5225        my ($dir, $file) = split_path($m->{file_b});
5226        my $pbat = $self->ensure_path($dir);
5227        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
5228                                        undef, -1);
5229        print "\tA\t$m->{file_b}\n" unless $::_q;
5230        $self->apply_autoprops($file, $fbat);
5231        $self->chg_file($fbat, $m);
5232        $self->close_file($fbat,undef,$self->{pool});
5233}
5234
5235sub C {
5236        my ($self, $m) = @_;
5237        my ($dir, $file) = split_path($m->{file_b});
5238        my $pbat = $self->ensure_path($dir);
5239        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
5240                                $self->url_path($m->{file_a}), $self->{r});
5241        print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
5242        $self->chg_file($fbat, $m);
5243        $self->close_file($fbat,undef,$self->{pool});
5244}
5245
5246sub delete_entry {
5247        my ($self, $path, $pbat) = @_;
5248        my $rpath = $self->repo_path($path);
5249        my ($dir, $file) = split_path($rpath);
5250        $self->{rm}->{$dir} = 1;
5251        $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
5252}
5253
5254sub R {
5255        my ($self, $m) = @_;
5256        my ($dir, $file) = split_path($m->{file_b});
5257        my $pbat = $self->ensure_path($dir);
5258        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
5259                                $self->url_path($m->{file_a}), $self->{r});
5260        print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
5261        $self->apply_autoprops($file, $fbat);
5262        $self->chg_file($fbat, $m);
5263        $self->close_file($fbat,undef,$self->{pool});
5264
5265        ($dir, $file) = split_path($m->{file_a});
5266        $pbat = $self->ensure_path($dir);
5267        $self->delete_entry($m->{file_a}, $pbat);
5268}
5269
5270sub M {
5271        my ($self, $m) = @_;
5272        my ($dir, $file) = split_path($m->{file_b});
5273        my $pbat = $self->ensure_path($dir);
5274        my $fbat = $self->open_file($self->repo_path($m->{file_b}),
5275                                $pbat,$self->{r},$self->{pool});
5276        print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
5277        $self->chg_file($fbat, $m);
5278        $self->close_file($fbat,undef,$self->{pool});
5279}
5280
5281sub T { shift->M(@_) }
5282
5283sub change_file_prop {
5284        my ($self, $fbat, $pname, $pval) = @_;
5285        $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
5286}
5287
5288sub change_dir_prop {
5289        my ($self, $pbat, $pname, $pval) = @_;
5290        $self->SUPER::change_dir_prop($pbat, $pname, $pval, $self->{pool});
5291}
5292
5293sub _chg_file_get_blob ($$$$) {
5294        my ($self, $fbat, $m, $which) = @_;
5295        my $fh = $::_repository->temp_acquire("git_blob_$which");
5296        if ($m->{"mode_$which"} =~ /^120/) {
5297                print $fh 'link ' or croak $!;
5298                $self->change_file_prop($fbat,'svn:special','*');
5299        } elsif ($m->{mode_a} =~ /^120/ && $m->{"mode_$which"} !~ /^120/) {
5300                $self->change_file_prop($fbat,'svn:special',undef);
5301        }
5302        my $blob = $m->{"sha1_$which"};
5303        return ($fh,) if ($blob =~ /^0{40}$/);
5304        my $size = $::_repository->cat_blob($blob, $fh);
5305        croak "Failed to read object $blob" if ($size < 0);
5306        $fh->flush == 0 or croak $!;
5307        seek $fh, 0, 0 or croak $!;
5308
5309        my $exp = ::md5sum($fh);
5310        seek $fh, 0, 0 or croak $!;
5311        return ($fh, $exp);
5312}
5313
5314sub chg_file {
5315        my ($self, $fbat, $m) = @_;
5316        if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
5317                $self->change_file_prop($fbat,'svn:executable','*');
5318        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
5319                $self->change_file_prop($fbat,'svn:executable',undef);
5320        }
5321        my ($fh_a, $exp_a) = _chg_file_get_blob $self, $fbat, $m, 'a';
5322        my ($fh_b, $exp_b) = _chg_file_get_blob $self, $fbat, $m, 'b';
5323        my $pool = SVN::Pool->new;
5324        my $atd = $self->apply_textdelta($fbat, $exp_a, $pool);
5325        if (-s $fh_a) {
5326                my $txstream = SVN::TxDelta::new ($fh_a, $fh_b, $pool);
5327                my $res = SVN::TxDelta::send_txstream($txstream, @$atd, $pool);
5328                if (defined $res) {
5329                        die "Unexpected result from send_txstream: $res\n",
5330                            "(SVN::Core::VERSION: $SVN::Core::VERSION)\n";
5331                }
5332        } else {
5333                my $got = SVN::TxDelta::send_stream($fh_b, @$atd, $pool);
5334                die "Checksum mismatch\nexpected: $exp_b\ngot: $got\n"
5335                    if ($got ne $exp_b);
5336        }
5337        Git::temp_release($fh_b, 1);
5338        Git::temp_release($fh_a, 1);
5339        $pool->clear;
5340}
5341
5342sub D {
5343        my ($self, $m) = @_;
5344        my ($dir, $file) = split_path($m->{file_b});
5345        my $pbat = $self->ensure_path($dir);
5346        print "\tD\t$m->{file_b}\n" unless $::_q;
5347        $self->delete_entry($m->{file_b}, $pbat);
5348}
5349
5350sub close_edit {
5351        my ($self) = @_;
5352        my ($p,$bat) = ($self->{pool}, $self->{bat});
5353        foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
5354                next if $_ eq '';
5355                $self->close_directory($bat->{$_}, $p);
5356        }
5357        $self->close_directory($bat->{''}, $p);
5358        $self->SUPER::close_edit($p);
5359        $p->clear;
5360}
5361
5362sub abort_edit {
5363        my ($self) = @_;
5364        $self->SUPER::abort_edit($self->{pool});
5365}
5366
5367sub DESTROY {
5368        my $self = shift;
5369        $self->SUPER::DESTROY(@_);
5370        $self->{pool}->clear;
5371}
5372
5373# this drives the editor
5374sub apply_diff {
5375        my ($self) = @_;
5376        my $mods = $self->{mods};
5377        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
5378        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
5379                my $f = $m->{chg};
5380                if (defined $o{$f}) {
5381                        $self->$f($m);
5382                } else {
5383                        fatal("Invalid change type: $f");
5384                }
5385        }
5386
5387        if (defined($self->{mergeinfo})) {
5388                $self->change_dir_prop($self->{bat}{''}, "svn:mergeinfo",
5389                                       $self->{mergeinfo});
5390        }
5391        $self->rmdirs if $_rmdir;
5392        if (@$mods == 0) {
5393                $self->abort_edit;
5394        } else {
5395                $self->close_edit;
5396        }
5397        return scalar @$mods;
5398}
5399
5400package Git::SVN::Ra;
5401use vars qw/@ISA $config_dir $_ignore_refs_regex $_log_window_size/;
5402use strict;
5403use warnings;
5404my ($ra_invalid, $can_do_switch, %ignored_err, $RA);
5405
5406BEGIN {
5407        # enforce temporary pool usage for some simple functions
5408        no strict 'refs';
5409        for my $f (qw/rev_proplist get_latest_revnum get_uuid get_repos_root
5410                      get_file/) {
5411                my $SUPER = "SUPER::$f";
5412                *$f = sub {
5413                        my $self = shift;
5414                        my $pool = SVN::Pool->new;
5415                        my @ret = $self->$SUPER(@_,$pool);
5416                        $pool->clear;
5417                        wantarray ? @ret : $ret[0];
5418                };
5419        }
5420}
5421
5422sub _auth_providers () {
5423        [
5424          SVN::Client::get_simple_provider(),
5425          SVN::Client::get_ssl_server_trust_file_provider(),
5426          SVN::Client::get_simple_prompt_provider(
5427            \&Git::SVN::Prompt::simple, 2),
5428          SVN::Client::get_ssl_client_cert_file_provider(),
5429          SVN::Client::get_ssl_client_cert_prompt_provider(
5430            \&Git::SVN::Prompt::ssl_client_cert, 2),
5431          SVN::Client::get_ssl_client_cert_pw_file_provider(),
5432          SVN::Client::get_ssl_client_cert_pw_prompt_provider(
5433            \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
5434          SVN::Client::get_username_provider(),
5435          SVN::Client::get_ssl_server_trust_prompt_provider(
5436            \&Git::SVN::Prompt::ssl_server_trust),
5437          SVN::Client::get_username_prompt_provider(
5438            \&Git::SVN::Prompt::username, 2)
5439        ]
5440}
5441
5442sub escape_uri_only {
5443        my ($uri) = @_;
5444        my @tmp;
5445        foreach (split m{/}, $uri) {
5446                s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
5447                push @tmp, $_;
5448        }
5449        join('/', @tmp);
5450}
5451
5452sub escape_url {
5453        my ($url) = @_;
5454        if ($url =~ m#^(https?)://([^/]+)(.*)$#) {
5455                my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
5456                $url = "$scheme://$domain$uri";
5457        }
5458        $url;
5459}
5460
5461sub new {
5462        my ($class, $url) = @_;
5463        $url =~ s!/+$!!;
5464        return $RA if ($RA && $RA->{url} eq $url);
5465
5466        ::_req_svn();
5467
5468        SVN::_Core::svn_config_ensure($config_dir, undef);
5469        my ($baton, $callbacks) = SVN::Core::auth_open_helper(_auth_providers);
5470        my $config = SVN::Core::config_get_config($config_dir);
5471        $RA = undef;
5472        my $dont_store_passwords = 1;
5473        my $conf_t = ${$config}{'config'};
5474        {
5475                no warnings 'once';
5476                # The usage of $SVN::_Core::SVN_CONFIG_* variables
5477                # produces warnings that variables are used only once.
5478                # I had not found the better way to shut them up, so
5479                # the warnings of type 'once' are disabled in this block.
5480                if (SVN::_Core::svn_config_get_bool($conf_t,
5481                    $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
5482                    $SVN::_Core::SVN_CONFIG_OPTION_STORE_PASSWORDS,
5483                    1) == 0) {
5484                        SVN::_Core::svn_auth_set_parameter($baton,
5485                            $SVN::_Core::SVN_AUTH_PARAM_DONT_STORE_PASSWORDS,
5486                            bless (\$dont_store_passwords, "_p_void"));
5487                }
5488                if (SVN::_Core::svn_config_get_bool($conf_t,
5489                    $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
5490                    $SVN::_Core::SVN_CONFIG_OPTION_STORE_AUTH_CREDS,
5491                    1) == 0) {
5492                        $Git::SVN::Prompt::_no_auth_cache = 1;
5493                }
5494        } # no warnings 'once'
5495        my $self = SVN::Ra->new(url => escape_url($url), auth => $baton,
5496                              config => $config,
5497                              pool => SVN::Pool->new,
5498                              auth_provider_callbacks => $callbacks);
5499        $self->{url} = $url;
5500        $self->{svn_path} = $url;
5501        $self->{repos_root} = $self->get_repos_root;
5502        $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
5503        $self->{cache} = { check_path => { r => 0, data => {} },
5504                           get_dir => { r => 0, data => {} } };
5505        $RA = bless $self, $class;
5506}
5507
5508sub check_path {
5509        my ($self, $path, $r) = @_;
5510        my $cache = $self->{cache}->{check_path};
5511        if ($r == $cache->{r} && exists $cache->{data}->{$path}) {
5512                return $cache->{data}->{$path};
5513        }
5514        my $pool = SVN::Pool->new;
5515        my $t = $self->SUPER::check_path($path, $r, $pool);
5516        $pool->clear;
5517        if ($r != $cache->{r}) {
5518                %{$cache->{data}} = ();
5519                $cache->{r} = $r;
5520        }
5521        $cache->{data}->{$path} = $t;
5522}
5523
5524sub get_dir {
5525        my ($self, $dir, $r) = @_;
5526        my $cache = $self->{cache}->{get_dir};
5527        if ($r == $cache->{r}) {
5528                if (my $x = $cache->{data}->{$dir}) {
5529                        return wantarray ? @$x : $x->[0];
5530                }
5531        }
5532        my $pool = SVN::Pool->new;
5533        my ($d, undef, $props) = $self->SUPER::get_dir($dir, $r, $pool);
5534        my %dirents = map { $_ => { kind => $d->{$_}->kind } } keys %$d;
5535        $pool->clear;
5536        if ($r != $cache->{r}) {
5537                %{$cache->{data}} = ();
5538                $cache->{r} = $r;
5539        }
5540        $cache->{data}->{$dir} = [ \%dirents, $r, $props ];
5541        wantarray ? (\%dirents, $r, $props) : \%dirents;
5542}
5543
5544sub DESTROY {
5545        # do not call the real DESTROY since we store ourselves in $RA
5546}
5547
5548# get_log(paths, start, end, limit,
5549#         discover_changed_paths, strict_node_history, receiver)
5550sub get_log {
5551        my ($self, @args) = @_;
5552        my $pool = SVN::Pool->new;
5553
5554        # svn_log_changed_path_t objects passed to get_log are likely to be
5555        # overwritten even if only the refs are copied to an external variable,
5556        # so we should dup the structures in their entirety.  Using an
5557        # externally passed pool (instead of our temporary and quickly cleared
5558        # pool in Git::SVN::Ra) does not help matters at all...
5559        my $receiver = pop @args;
5560        my $prefix = "/".$self->{svn_path};
5561        $prefix =~ s#/+($)##;
5562        my $prefix_regex = qr#^\Q$prefix\E#;
5563        push(@args, sub {
5564                my ($paths) = $_[0];
5565                return &$receiver(@_) unless $paths;
5566                $_[0] = ();
5567                foreach my $p (keys %$paths) {
5568                        my $i = $paths->{$p};
5569                        # Make path relative to our url, not repos_root
5570                        $p =~ s/$prefix_regex//;
5571                        my %s = map { $_ => $i->$_; }
5572                                qw/copyfrom_path copyfrom_rev action/;
5573                        if ($s{'copyfrom_path'}) {
5574                                $s{'copyfrom_path'} =~ s/$prefix_regex//;
5575                        }
5576                        $_[0]{$p} = \%s;
5577                }
5578                &$receiver(@_);
5579        });
5580
5581
5582        # the limit parameter was not supported in SVN 1.1.x, so we
5583        # drop it.  Therefore, the receiver callback passed to it
5584        # is made aware of this limitation by being wrapped if
5585        # the limit passed to is being wrapped.
5586        if ($SVN::Core::VERSION le '1.2.0') {
5587                my $limit = splice(@args, 3, 1);
5588                if ($limit > 0) {
5589                        my $receiver = pop @args;
5590                        push(@args, sub { &$receiver(@_) if (--$limit >= 0) });
5591                }
5592        }
5593        my $ret = $self->SUPER::get_log(@args, $pool);
5594        $pool->clear;
5595        $ret;
5596}
5597
5598sub trees_match {
5599        my ($self, $url1, $rev1, $url2, $rev2) = @_;
5600        my $ctx = SVN::Client->new(auth => _auth_providers);
5601        my $out = IO::File->new_tmpfile;
5602
5603        # older SVN (1.1.x) doesn't take $pool as the last parameter for
5604        # $ctx->diff(), so we'll create a default one
5605        my $pool = SVN::Pool->new_default_sub;
5606
5607        $ra_invalid = 1; # this will open a new SVN::Ra connection to $url1
5608        $ctx->diff([], $url1, $rev1, $url2, $rev2, 1, 1, 0, $out, $out);
5609        $out->flush;
5610        my $ret = (($out->stat)[7] == 0);
5611        close $out or croak $!;
5612
5613        $ret;
5614}
5615
5616sub get_commit_editor {
5617        my ($self, $log, $cb, $pool) = @_;
5618        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
5619        $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
5620}
5621
5622sub gs_do_update {
5623        my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
5624        my $new = ($rev_a == $rev_b);
5625        my $path = $gs->{path};
5626
5627        if ($new && -e $gs->{index}) {
5628                unlink $gs->{index} or die
5629                  "Couldn't unlink index: $gs->{index}: $!\n";
5630        }
5631        my $pool = SVN::Pool->new;
5632        $editor->set_path_strip($path);
5633        my (@pc) = split m#/#, $path;
5634        my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
5635                                        1, $editor, $pool);
5636        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
5637
5638        # Since we can't rely on svn_ra_reparent being available, we'll
5639        # just have to do some magic with set_path to make it so
5640        # we only want a partial path.
5641        my $sp = '';
5642        my $final = join('/', @pc);
5643        while (@pc) {
5644                $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
5645                $sp .= '/' if length $sp;
5646                $sp .= shift @pc;
5647        }
5648        die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
5649
5650        $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
5651
5652        $reporter->finish_report($pool);
5653        $pool->clear;
5654        $editor->{git_commit_ok};
5655}
5656
5657# this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
5658# svn_ra_reparent didn't work before 1.4)
5659sub gs_do_switch {
5660        my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
5661        my $path = $gs->{path};
5662        my $pool = SVN::Pool->new;
5663
5664        my $full_url = $self->{url};
5665        my $old_url = $full_url;
5666        $full_url .= '/' . $path if length $path;
5667        my ($ra, $reparented);
5668
5669        if ($old_url =~ m#^svn(\+ssh)?://# ||
5670            ($full_url =~ m#^https?://# &&
5671             escape_url($full_url) ne $full_url)) {
5672                $_[0] = undef;
5673                $self = undef;
5674                $RA = undef;
5675                $ra = Git::SVN::Ra->new($full_url);
5676                $ra_invalid = 1;
5677        } elsif ($old_url ne $full_url) {
5678                SVN::_Ra::svn_ra_reparent($self->{session}, $full_url, $pool);
5679                $self->{url} = $full_url;
5680                $reparented = 1;
5681        }
5682
5683        $ra ||= $self;
5684        $url_b = escape_url($url_b);
5685        my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
5686        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
5687        $reporter->set_path('', $rev_a, 0, @lock, $pool);
5688        $reporter->finish_report($pool);
5689
5690        if ($reparented) {
5691                SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
5692                $self->{url} = $old_url;
5693        }
5694
5695        $pool->clear;
5696        $editor->{git_commit_ok};
5697}
5698
5699sub longest_common_path {
5700        my ($gsv, $globs) = @_;
5701        my %common;
5702        my $common_max = scalar @$gsv;
5703
5704        foreach my $gs (@$gsv) {
5705                my @tmp = split m#/#, $gs->{path};
5706                my $p = '';
5707                foreach (@tmp) {
5708                        $p .= length($p) ? "/$_" : $_;
5709                        $common{$p} ||= 0;
5710                        $common{$p}++;
5711                }
5712        }
5713        $globs ||= [];
5714        $common_max += scalar @$globs;
5715        foreach my $glob (@$globs) {
5716                my @tmp = split m#/#, $glob->{path}->{left};
5717                my $p = '';
5718                foreach (@tmp) {
5719                        $p .= length($p) ? "/$_" : $_;
5720                        $common{$p} ||= 0;
5721                        $common{$p}++;
5722                }
5723        }
5724
5725        my $longest_path = '';
5726        foreach (sort {length $b <=> length $a} keys %common) {
5727                if ($common{$_} == $common_max) {
5728                        $longest_path = $_;
5729                        last;
5730                }
5731        }
5732        $longest_path;
5733}
5734
5735sub gs_fetch_loop_common {
5736        my ($self, $base, $head, $gsv, $globs) = @_;
5737        return if ($base > $head);
5738        my $inc = $_log_window_size;
5739        my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
5740        my $longest_path = longest_common_path($gsv, $globs);
5741        my $ra_url = $self->{url};
5742        my $find_trailing_edge;
5743        while (1) {
5744                my %revs;
5745                my $err;
5746                my $err_handler = $SVN::Error::handler;
5747                $SVN::Error::handler = sub {
5748                        ($err) = @_;
5749                        skip_unknown_revs($err);
5750                };
5751                sub _cb {
5752                        my ($paths, $r, $author, $date, $log) = @_;
5753                        [ $paths,
5754                          { author => $author, date => $date, log => $log } ];
5755                }
5756                $self->get_log([$longest_path], $min, $max, 0, 1, 1,
5757                               sub { $revs{$_[1]} = _cb(@_) });
5758                if ($err) {
5759                        print "Checked through r$max\r";
5760                } else {
5761                        $find_trailing_edge = 1;
5762                }
5763                if ($err and $find_trailing_edge) {
5764                        print STDERR "Path '$longest_path' ",
5765                                     "was probably deleted:\n",
5766                                     $err->expanded_message,
5767                                     "\nWill attempt to follow ",
5768                                     "revisions r$min .. r$max ",
5769                                     "committed before the deletion\n";
5770                        my $hi = $max;
5771                        while (--$hi >= $min) {
5772                                my $ok;
5773                                $self->get_log([$longest_path], $min, $hi,
5774                                               0, 1, 1, sub {
5775                                               $ok = $_[1];
5776                                               $revs{$_[1]} = _cb(@_) });
5777                                if ($ok) {
5778                                        print STDERR "r$min .. r$ok OK\n";
5779                                        last;
5780                                }
5781                        }
5782                        $find_trailing_edge = 0;
5783                }
5784                $SVN::Error::handler = $err_handler;
5785
5786                my %exists = map { $_->{path} => $_ } @$gsv;
5787                foreach my $r (sort {$a <=> $b} keys %revs) {
5788                        my ($paths, $logged) = @{$revs{$r}};
5789
5790                        foreach my $gs ($self->match_globs(\%exists, $paths,
5791                                                           $globs, $r)) {
5792                                if ($gs->rev_map_max >= $r) {
5793                                        next;
5794                                }
5795                                next unless $gs->match_paths($paths, $r);
5796                                $gs->{logged_rev_props} = $logged;
5797                                if (my $last_commit = $gs->last_commit) {
5798                                        $gs->assert_index_clean($last_commit);
5799                                }
5800                                my $log_entry = $gs->do_fetch($paths, $r);
5801                                if ($log_entry) {
5802                                        $gs->do_git_commit($log_entry);
5803                                }
5804                                $INDEX_FILES{$gs->{index}} = 1;
5805                        }
5806                        foreach my $g (@$globs) {
5807                                my $k = "svn-remote.$g->{remote}." .
5808                                        "$g->{t}-maxRev";
5809                                Git::SVN::tmp_config($k, $r);
5810                        }
5811                        if ($ra_invalid) {
5812                                $_[0] = undef;
5813                                $self = undef;
5814                                $RA = undef;
5815                                $self = Git::SVN::Ra->new($ra_url);
5816                                $ra_invalid = undef;
5817                        }
5818                }
5819                # pre-fill the .rev_db since it'll eventually get filled in
5820                # with '0' x40 if something new gets committed
5821                foreach my $gs (@$gsv) {
5822                        next if $gs->rev_map_max >= $max;
5823                        next if defined $gs->rev_map_get($max);
5824                        $gs->rev_map_set($max, 0 x40);
5825                }
5826                foreach my $g (@$globs) {
5827                        my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
5828                        Git::SVN::tmp_config($k, $max);
5829                }
5830                last if $max >= $head;
5831                $min = $max + 1;
5832                $max += $inc;
5833                $max = $head if ($max > $head);
5834        }
5835        Git::SVN::gc();
5836}
5837
5838sub get_dir_globbed {
5839        my ($self, $left, $depth, $r) = @_;
5840
5841        my @x = eval { $self->get_dir($left, $r) };
5842        return unless scalar @x == 3;
5843        my $dirents = $x[0];
5844        my @finalents;
5845        foreach my $de (keys %$dirents) {
5846                next if $dirents->{$de}->{kind} != $SVN::Node::dir;
5847                if ($depth > 1) {
5848                        my @args = ("$left/$de", $depth - 1, $r);
5849                        foreach my $dir ($self->get_dir_globbed(@args)) {
5850                                push @finalents, "$de/$dir";
5851                        }
5852                } else {
5853                        push @finalents, $de;
5854                }
5855        }
5856        @finalents;
5857}
5858
5859# return value: 0 -- don't ignore, 1 -- ignore
5860sub is_ref_ignored {
5861        my ($g, $p) = @_;
5862        my $refname = $g->{ref}->full_path($p);
5863        return 1 if defined($g->{ignore_refs_regex}) &&
5864                    $refname =~ m!$g->{ignore_refs_regex}!;
5865        return 0 unless defined($_ignore_refs_regex);
5866        return 1 if $refname =~ m!$_ignore_refs_regex!o;
5867        return 0;
5868}
5869
5870sub match_globs {
5871        my ($self, $exists, $paths, $globs, $r) = @_;
5872
5873        sub get_dir_check {
5874                my ($self, $exists, $g, $r) = @_;
5875
5876                my @dirs = $self->get_dir_globbed($g->{path}->{left},
5877                                                  $g->{path}->{depth},
5878                                                  $r);
5879
5880                foreach my $de (@dirs) {
5881                        my $p = $g->{path}->full_path($de);
5882                        next if $exists->{$p};
5883                        next if (length $g->{path}->{right} &&
5884                                 ($self->check_path($p, $r) !=
5885                                  $SVN::Node::dir));
5886                        next unless $p =~ /$g->{path}->{regex}/;
5887                        $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
5888                                         $g->{ref}->full_path($de), 1);
5889                }
5890        }
5891        foreach my $g (@$globs) {
5892                if (my $path = $paths->{"/$g->{path}->{left}"}) {
5893                        if ($path->{action} =~ /^[AR]$/) {
5894                                get_dir_check($self, $exists, $g, $r);
5895                        }
5896                }
5897                foreach (keys %$paths) {
5898                        if (/$g->{path}->{left_regex}/ &&
5899                            !/$g->{path}->{regex}/) {
5900                                next if $paths->{$_}->{action} !~ /^[AR]$/;
5901                                get_dir_check($self, $exists, $g, $r);
5902                        }
5903                        next unless /$g->{path}->{regex}/;
5904                        my $p = $1;
5905                        my $pathname = $g->{path}->full_path($p);
5906                        next if is_ref_ignored($g, $p);
5907                        next if $exists->{$pathname};
5908                        next if ($self->check_path($pathname, $r) !=
5909                                 $SVN::Node::dir);
5910                        $exists->{$pathname} = Git::SVN->init(
5911                                              $self->{url}, $pathname, undef,
5912                                              $g->{ref}->full_path($p), 1);
5913                }
5914                my $c = '';
5915                foreach (split m#/#, $g->{path}->{left}) {
5916                        $c .= "/$_";
5917                        next unless ($paths->{$c} &&
5918                                     ($paths->{$c}->{action} =~ /^[AR]$/));
5919                        get_dir_check($self, $exists, $g, $r);
5920                }
5921        }
5922        values %$exists;
5923}
5924
5925sub minimize_url {
5926        my ($self) = @_;
5927        return $self->{url} if ($self->{url} eq $self->{repos_root});
5928        my $url = $self->{repos_root};
5929        my @components = split(m!/!, $self->{svn_path});
5930        my $c = '';
5931        do {
5932                $url .= "/$c" if length $c;
5933                eval {
5934                        my $ra = (ref $self)->new($url);
5935                        my $latest = $ra->get_latest_revnum;
5936                        $ra->get_log("", $latest, 0, 1, 0, 1, sub {});
5937                };
5938        } while ($@ && ($c = shift @components));
5939        $url;
5940}
5941
5942sub can_do_switch {
5943        my $self = shift;
5944        unless (defined $can_do_switch) {
5945                my $pool = SVN::Pool->new;
5946                my $rep = eval {
5947                        $self->do_switch(1, '', 0, $self->{url},
5948                                         SVN::Delta::Editor->new, $pool);
5949                };
5950                if ($@) {
5951                        $can_do_switch = 0;
5952                } else {
5953                        $rep->abort_report($pool);
5954                        $can_do_switch = 1;
5955                }
5956                $pool->clear;
5957        }
5958        $can_do_switch;
5959}
5960
5961sub skip_unknown_revs {
5962        my ($err) = @_;
5963        my $errno = $err->apr_err();
5964        # Maybe the branch we're tracking didn't
5965        # exist when the repo started, so it's
5966        # not an error if it doesn't, just continue
5967        #
5968        # Wonderfully consistent library, eh?
5969        # 160013 - svn:// and file://
5970        # 175002 - http(s)://
5971        # 175007 - http(s):// (this repo required authorization, too...)
5972        #   More codes may be discovered later...
5973        if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
5974                my $err_key = $err->expanded_message;
5975                # revision numbers change every time, filter them out
5976                $err_key =~ s/\d+/\0/g;
5977                $err_key = "$errno\0$err_key";
5978                unless ($ignored_err{$err_key}) {
5979                        warn "W: Ignoring error from SVN, path probably ",
5980                             "does not exist: ($errno): ",
5981                             $err->expanded_message,"\n";
5982                        warn "W: Do not be alarmed at the above message ",
5983                             "git-svn is just searching aggressively for ",
5984                             "old history.\n",
5985                             "This may take a while on large repositories\n";
5986                        $ignored_err{$err_key} = 1;
5987                }
5988                return;
5989        }
5990        die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
5991}
5992
5993package Git::SVN::Log;
5994use strict;
5995use warnings;
5996use POSIX qw/strftime/;
5997use Time::Local;
5998use constant commit_log_separator => ('-' x 72) . "\n";
5999use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
6000            %rusers $show_commit $incremental/;
6001my $l_fmt;
6002
6003sub cmt_showable {
6004        my ($c) = @_;
6005        return 1 if defined $c->{r};
6006
6007        # big commit message got truncated by the 16k pretty buffer in rev-list
6008        if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
6009                                $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
6010                @{$c->{l}} = ();
6011                my @log = command(qw/cat-file commit/, $c->{c});
6012
6013                # shift off the headers
6014                shift @log while ($log[0] ne '');
6015                shift @log;
6016
6017                # TODO: make $c->{l} not have a trailing newline in the future
6018                @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
6019
6020                (undef, $c->{r}, undef) = ::extract_metadata(
6021                                (grep(/^git-svn-id: /, @log))[-1]);
6022        }
6023        return defined $c->{r};
6024}
6025
6026sub log_use_color {
6027        return $color || Git->repository->get_colorbool('color.diff');
6028}
6029
6030sub git_svn_log_cmd {
6031        my ($r_min, $r_max, @args) = @_;
6032        my $head = 'HEAD';
6033        my (@files, @log_opts);
6034        foreach my $x (@args) {
6035                if ($x eq '--' || @files) {
6036                        push @files, $x;
6037                } else {
6038                        if (::verify_ref("$x^0")) {
6039                                $head = $x;
6040                        } else {
6041                                push @log_opts, $x;
6042                        }
6043                }
6044        }
6045
6046        my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
6047        $gs ||= Git::SVN->_new;
6048        my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
6049                   $gs->refname);
6050        push @cmd, '-r' unless $non_recursive;
6051        push @cmd, qw/--raw --name-status/ if $verbose;
6052        push @cmd, '--color' if log_use_color();
6053        push @cmd, @log_opts;
6054        if (defined $r_max && $r_max == $r_min) {
6055                push @cmd, '--max-count=1';
6056                if (my $c = $gs->rev_map_get($r_max)) {
6057                        push @cmd, $c;
6058                }
6059        } elsif (defined $r_max) {
6060                if ($r_max < $r_min) {
6061                        ($r_min, $r_max) = ($r_max, $r_min);
6062                }
6063                my (undef, $c_max) = $gs->find_rev_before($r_max, 1, $r_min);
6064                my (undef, $c_min) = $gs->find_rev_after($r_min, 1, $r_max);
6065                # If there are no commits in the range, both $c_max and $c_min
6066                # will be undefined.  If there is at least 1 commit in the
6067                # range, both will be defined.
6068                return () if !defined $c_min || !defined $c_max;
6069                if ($c_min eq $c_max) {
6070                        push @cmd, '--max-count=1', $c_min;
6071                } else {
6072                        push @cmd, '--boundary', "$c_min..$c_max";
6073                }
6074        }
6075        return (@cmd, @files);
6076}
6077
6078# adapted from pager.c
6079sub config_pager {
6080        if (! -t *STDOUT) {
6081                $ENV{GIT_PAGER_IN_USE} = 'false';
6082                $pager = undef;
6083                return;
6084        }
6085        chomp($pager = command_oneline(qw(var GIT_PAGER)));
6086        if ($pager eq 'cat') {
6087                $pager = undef;
6088        }
6089        $ENV{GIT_PAGER_IN_USE} = defined($pager);
6090}
6091
6092sub run_pager {
6093        return unless defined $pager;
6094        pipe my ($rfd, $wfd) or return;
6095        defined(my $pid = fork) or ::fatal "Can't fork: $!";
6096        if (!$pid) {
6097                open STDOUT, '>&', $wfd or
6098                                     ::fatal "Can't redirect to stdout: $!";
6099                return;
6100        }
6101        open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!";
6102        $ENV{LESS} ||= 'FRSX';
6103        exec $pager or ::fatal "Can't run pager: $! ($pager)";
6104}
6105
6106sub format_svn_date {
6107        # some systmes don't handle or mishandle %z, so be creative.
6108        my $t = shift || time;
6109        my $gm = timelocal(gmtime($t));
6110        my $sign = qw( + + - )[ $t <=> $gm ];
6111        my $gmoff = sprintf("%s%02d%02d", $sign, (gmtime(abs($t - $gm)))[2,1]);
6112        return strftime("%Y-%m-%d %H:%M:%S $gmoff (%a, %d %b %Y)", localtime($t));
6113}
6114
6115sub parse_git_date {
6116        my ($t, $tz) = @_;
6117        # Date::Parse isn't in the standard Perl distro :(
6118        if ($tz =~ s/^\+//) {
6119                $t += tz_to_s_offset($tz);
6120        } elsif ($tz =~ s/^\-//) {
6121                $t -= tz_to_s_offset($tz);
6122        }
6123        return $t;
6124}
6125
6126sub set_local_timezone {
6127        if (defined $TZ) {
6128                $ENV{TZ} = $TZ;
6129        } else {
6130                delete $ENV{TZ};
6131        }
6132}
6133
6134sub tz_to_s_offset {
6135        my ($tz) = @_;
6136        $tz =~ s/(\d\d)$//;
6137        return ($1 * 60) + ($tz * 3600);
6138}
6139
6140sub get_author_info {
6141        my ($dest, $author, $t, $tz) = @_;
6142        $author =~ s/(?:^\s*|\s*$)//g;
6143        $dest->{a_raw} = $author;
6144        my $au;
6145        if ($::_authors) {
6146                $au = $rusers{$author} || undef;
6147        }
6148        if (!$au) {
6149                ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
6150        }
6151        $dest->{t} = $t;
6152        $dest->{tz} = $tz;
6153        $dest->{a} = $au;
6154        $dest->{t_utc} = parse_git_date($t, $tz);
6155}
6156
6157sub process_commit {
6158        my ($c, $r_min, $r_max, $defer) = @_;
6159        if (defined $r_min && defined $r_max) {
6160                if ($r_min == $c->{r} && $r_min == $r_max) {
6161                        show_commit($c);
6162                        return 0;
6163                }
6164                return 1 if $r_min == $r_max;
6165                if ($r_min < $r_max) {
6166                        # we need to reverse the print order
6167                        return 0 if (defined $limit && --$limit < 0);
6168                        push @$defer, $c;
6169                        return 1;
6170                }
6171                if ($r_min != $r_max) {
6172                        return 1 if ($r_min < $c->{r});
6173                        return 1 if ($r_max > $c->{r});
6174                }
6175        }
6176        return 0 if (defined $limit && --$limit < 0);
6177        show_commit($c);
6178        return 1;
6179}
6180
6181sub show_commit {
6182        my $c = shift;
6183        if ($oneline) {
6184                my $x = "\n";
6185                if (my $l = $c->{l}) {
6186                        while ($l->[0] =~ /^\s*$/) { shift @$l }
6187                        $x = $l->[0];
6188                }
6189                $l_fmt ||= 'A' . length($c->{r});
6190                print 'r',pack($l_fmt, $c->{r}),' | ';
6191                print "$c->{c} | " if $show_commit;
6192                print $x;
6193        } else {
6194                show_commit_normal($c);
6195        }
6196}
6197
6198sub show_commit_changed_paths {
6199        my ($c) = @_;
6200        return unless $c->{changed};
6201        print "Changed paths:\n", @{$c->{changed}};
6202}
6203
6204sub show_commit_normal {
6205        my ($c) = @_;
6206        print commit_log_separator, "r$c->{r} | ";
6207        print "$c->{c} | " if $show_commit;
6208        print "$c->{a} | ", format_svn_date($c->{t_utc}), ' | ';
6209        my $nr_line = 0;
6210
6211        if (my $l = $c->{l}) {
6212                while ($l->[$#$l] eq "\n" && $#$l > 0
6213                                          && $l->[($#$l - 1)] eq "\n") {
6214                        pop @$l;
6215                }
6216                $nr_line = scalar @$l;
6217                if (!$nr_line) {
6218                        print "1 line\n\n\n";
6219                } else {
6220                        if ($nr_line == 1) {
6221                                $nr_line = '1 line';
6222                        } else {
6223                                $nr_line .= ' lines';
6224                        }
6225                        print $nr_line, "\n";
6226                        show_commit_changed_paths($c);
6227                        print "\n";
6228                        print $_ foreach @$l;
6229                }
6230        } else {
6231                print "1 line\n";
6232                show_commit_changed_paths($c);
6233                print "\n";
6234
6235        }
6236        foreach my $x (qw/raw stat diff/) {
6237                if ($c->{$x}) {
6238                        print "\n";
6239                        print $_ foreach @{$c->{$x}}
6240                }
6241        }
6242}
6243
6244sub cmd_show_log {
6245        my (@args) = @_;
6246        my ($r_min, $r_max);
6247        my $r_last = -1; # prevent dupes
6248        set_local_timezone();
6249        if (defined $::_revision) {
6250                if ($::_revision =~ /^(\d+):(\d+)$/) {
6251                        ($r_min, $r_max) = ($1, $2);
6252                } elsif ($::_revision =~ /^\d+$/) {
6253                        $r_min = $r_max = $::_revision;
6254                } else {
6255                        ::fatal "-r$::_revision is not supported, use ",
6256                                "standard 'git log' arguments instead";
6257                }
6258        }
6259
6260        config_pager();
6261        @args = git_svn_log_cmd($r_min, $r_max, @args);
6262        if (!@args) {
6263                print commit_log_separator unless $incremental || $oneline;
6264                return;
6265        }
6266        my $log = command_output_pipe(@args);
6267        run_pager();
6268        my (@k, $c, $d, $stat);
6269        my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
6270        while (<$log>) {
6271                if (/^${esc_color}commit (?:- )?($::sha1_short)/o) {
6272                        my $cmt = $1;
6273                        if ($c && cmt_showable($c) && $c->{r} != $r_last) {
6274                                $r_last = $c->{r};
6275                                process_commit($c, $r_min, $r_max, \@k) or
6276                                                                goto out;
6277                        }
6278                        $d = undef;
6279                        $c = { c => $cmt };
6280                } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
6281                        get_author_info($c, $1, $2, $3);
6282                } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
6283                        # ignore
6284                } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
6285                        push @{$c->{raw}}, $_;
6286                } elsif (/^${esc_color}[ACRMDT]\t/) {
6287                        # we could add $SVN->{svn_path} here, but that requires
6288                        # remote access at the moment (repo_path_split)...
6289                        s#^(${esc_color})([ACRMDT])\t#$1   $2 #o;
6290                        push @{$c->{changed}}, $_;
6291                } elsif (/^${esc_color}diff /o) {
6292                        $d = 1;
6293                        push @{$c->{diff}}, $_;
6294                } elsif ($d) {
6295                        push @{$c->{diff}}, $_;
6296                } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
6297                          $esc_color*[\+\-]*$esc_color$/x) {
6298                        $stat = 1;
6299                        push @{$c->{stat}}, $_;
6300                } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
6301                        push @{$c->{stat}}, $_;
6302                        $stat = undef;
6303                } elsif (/^${esc_color}    (git-svn-id:.+)$/o) {
6304                        ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
6305                } elsif (s/^${esc_color}    //o) {
6306                        push @{$c->{l}}, $_;
6307                }
6308        }
6309        if ($c && defined $c->{r} && $c->{r} != $r_last) {
6310                $r_last = $c->{r};
6311                process_commit($c, $r_min, $r_max, \@k);
6312        }
6313        if (@k) {
6314                ($r_min, $r_max) = ($r_max, $r_min);
6315                process_commit($_, $r_min, $r_max) foreach reverse @k;
6316        }
6317out:
6318        close $log;
6319        print commit_log_separator unless $incremental || $oneline;
6320}
6321
6322sub cmd_blame {
6323        my $path = pop;
6324
6325        config_pager();
6326        run_pager();
6327
6328        my ($fh, $ctx, $rev);
6329
6330        if ($_git_format) {
6331                ($fh, $ctx) = command_output_pipe('blame', @_, $path);
6332                while (my $line = <$fh>) {
6333                        if ($line =~ /^\^?([[:xdigit:]]+)\s/) {
6334                                # Uncommitted edits show up as a rev ID of
6335                                # all zeros, which we can't look up with
6336                                # cmt_metadata
6337                                if ($1 !~ /^0+$/) {
6338                                        (undef, $rev, undef) =
6339                                                ::cmt_metadata($1);
6340                                        $rev = '0' if (!$rev);
6341                                } else {
6342                                        $rev = '0';
6343                                }
6344                                $rev = sprintf('%-10s', $rev);
6345                                $line =~ s/^\^?[[:xdigit:]]+(\s)/$rev$1/;
6346                        }
6347                        print $line;
6348                }
6349        } else {
6350                ($fh, $ctx) = command_output_pipe('blame', '-p', @_, 'HEAD',
6351                                                  '--', $path);
6352                my ($sha1);
6353                my %authors;
6354                my @buffer;
6355                my %dsha; #distinct sha keys
6356
6357                while (my $line = <$fh>) {
6358                        push @buffer, $line;
6359                        if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
6360                                $dsha{$1} = 1;
6361                        }
6362                }
6363
6364                my $s2r = ::cmt_sha2rev_batch([keys %dsha]);
6365
6366                foreach my $line (@buffer) {
6367                        if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
6368                                $rev = $s2r->{$1};
6369                                $rev = '0' if (!$rev)
6370                        }
6371                        elsif ($line =~ /^author (.*)/) {
6372                                $authors{$rev} = $1;
6373                                $authors{$rev} =~ s/\s/_/g;
6374                        }
6375                        elsif ($line =~ /^\t(.*)$/) {
6376                                printf("%6s %10s %s\n", $rev, $authors{$rev}, $1);
6377                        }
6378                }
6379        }
6380        command_close_pipe($fh, $ctx);
6381}
6382
6383package Git::SVN::Migration;
6384# these version numbers do NOT correspond to actual version numbers
6385# of git nor git-svn.  They are just relative.
6386#
6387# v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
6388#
6389# v1 layout: .git/$id/info/url, refs/remotes/$id
6390#
6391# v2 layout: .git/svn/$id/info/url, refs/remotes/$id
6392#
6393# v3 layout: .git/svn/$id, refs/remotes/$id
6394#            - info/url may remain for backwards compatibility
6395#            - this is what we migrate up to this layout automatically,
6396#            - this will be used by git svn init on single branches
6397# v3.1 layout (auto migrated):
6398#            - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
6399#              for backwards compatibility
6400#
6401# v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
6402#            - this is only created for newly multi-init-ed
6403#              repositories.  Similar in spirit to the
6404#              --use-separate-remotes option in git-clone (now default)
6405#            - we do not automatically migrate to this (following
6406#              the example set by core git)
6407#
6408# v5 layout: .rev_db.$UUID => .rev_map.$UUID
6409#            - newer, more-efficient format that uses 24-bytes per record
6410#              with no filler space.
6411#            - use xxd -c24 < .rev_map.$UUID to view and debug
6412#            - This is a one-way migration, repositories updated to the
6413#              new format will not be able to use old git-svn without
6414#              rebuilding the .rev_db.  Rebuilding the rev_db is not
6415#              possible if noMetadata or useSvmProps are set; but should
6416#              be no problem for users that use the (sensible) defaults.
6417use strict;
6418use warnings;
6419use Carp qw/croak/;
6420use File::Path qw/mkpath/;
6421use File::Basename qw/dirname basename/;
6422use vars qw/$_minimize/;
6423
6424sub migrate_from_v0 {
6425        my $git_dir = $ENV{GIT_DIR};
6426        return undef unless -d $git_dir;
6427        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
6428        my $migrated = 0;
6429        while (<$fh>) {
6430                chomp;
6431                my ($id, $orig_ref) = ($_, $_);
6432                next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
6433                next unless -f "$git_dir/$id/info/url";
6434                my $new_ref = "refs/remotes/$id";
6435                if (::verify_ref("$new_ref^0")) {
6436                        print STDERR "W: $orig_ref is probably an old ",
6437                                     "branch used by an ancient version of ",
6438                                     "git-svn.\n",
6439                                     "However, $new_ref also exists.\n",
6440                                     "We will not be able ",
6441                                     "to use this branch until this ",
6442                                     "ambiguity is resolved.\n";
6443                        next;
6444                }
6445                print STDERR "Migrating from v0 layout...\n" if !$migrated;
6446                print STDERR "Renaming ref: $orig_ref => $new_ref\n";
6447                command_noisy('update-ref', $new_ref, $orig_ref);
6448                command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
6449                $migrated++;
6450        }
6451        command_close_pipe($fh, $ctx);
6452        print STDERR "Done migrating from v0 layout...\n" if $migrated;
6453        $migrated;
6454}
6455
6456sub migrate_from_v1 {
6457        my $git_dir = $ENV{GIT_DIR};
6458        my $migrated = 0;
6459        return $migrated unless -d $git_dir;
6460        my $svn_dir = "$git_dir/svn";
6461
6462        # just in case somebody used 'svn' as their $id at some point...
6463        return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
6464
6465        print STDERR "Migrating from a git-svn v1 layout...\n";
6466        mkpath([$svn_dir]);
6467        print STDERR "Data from a previous version of git-svn exists, but\n\t",
6468                     "$svn_dir\n\t(required for this version ",
6469                     "($::VERSION) of git-svn) does not exist.\n";
6470        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
6471        while (<$fh>) {
6472                my $x = $_;
6473                next unless $x =~ s#^refs/remotes/##;
6474                chomp $x;
6475                next unless -f "$git_dir/$x/info/url";
6476                my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
6477                next unless $u;
6478                my $dn = dirname("$git_dir/svn/$x");
6479                mkpath([$dn]) unless -d $dn;
6480                if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
6481                        mkpath(["$git_dir/svn/svn"]);
6482                        print STDERR " - $git_dir/$x/info => ",
6483                                        "$git_dir/svn/$x/info\n";
6484                        rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
6485                               croak "$!: $x";
6486                        # don't worry too much about these, they probably
6487                        # don't exist with repos this old (save for index,
6488                        # and we can easily regenerate that)
6489                        foreach my $f (qw/unhandled.log index .rev_db/) {
6490                                rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
6491                        }
6492                } else {
6493                        print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
6494                        rename "$git_dir/$x", "$git_dir/svn/$x" or
6495                               croak "$!: $x";
6496                }
6497                $migrated++;
6498        }
6499        command_close_pipe($fh, $ctx);
6500        print STDERR "Done migrating from a git-svn v1 layout\n";
6501        $migrated;
6502}
6503
6504sub read_old_urls {
6505        my ($l_map, $pfx, $path) = @_;
6506        my @dir;
6507        foreach (<$path/*>) {
6508                if (-r "$_/info/url") {
6509                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
6510                        my $ref_id = $pfx . basename $_;
6511                        my $url = ::file_to_s("$_/info/url");
6512                        $l_map->{$ref_id} = $url;
6513                } elsif (-d $_) {
6514                        push @dir, $_;
6515                }
6516        }
6517        foreach (@dir) {
6518                my $x = $_;
6519                $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
6520                read_old_urls($l_map, $x, $_);
6521        }
6522}
6523
6524sub migrate_from_v2 {
6525        my @cfg = command(qw/config -l/);
6526        return if grep /^svn-remote\..+\.url=/, @cfg;
6527        my %l_map;
6528        read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
6529        my $migrated = 0;
6530
6531        foreach my $ref_id (sort keys %l_map) {
6532                eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
6533                if ($@) {
6534                        Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
6535                }
6536                $migrated++;
6537        }
6538        $migrated;
6539}
6540
6541sub minimize_connections {
6542        my $r = Git::SVN::read_all_remotes();
6543        my $new_urls = {};
6544        my $root_repos = {};
6545        foreach my $repo_id (keys %$r) {
6546                my $url = $r->{$repo_id}->{url} or next;
6547                my $fetch = $r->{$repo_id}->{fetch} or next;
6548                my $ra = Git::SVN::Ra->new($url);
6549
6550                # skip existing cases where we already connect to the root
6551                if (($ra->{url} eq $ra->{repos_root}) ||
6552                    ($ra->{repos_root} eq $repo_id)) {
6553                        $root_repos->{$ra->{url}} = $repo_id;
6554                        next;
6555                }
6556
6557                my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
6558                my $root_path = $ra->{url};
6559                $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
6560                foreach my $path (keys %$fetch) {
6561                        my $ref_id = $fetch->{$path};
6562                        my $gs = Git::SVN->new($ref_id, $repo_id, $path);
6563
6564                        # make sure we can read when connecting to
6565                        # a higher level of a repository
6566                        my ($last_rev, undef) = $gs->last_rev_commit;
6567                        if (!defined $last_rev) {
6568                                $last_rev = eval {
6569                                        $root_ra->get_latest_revnum;
6570                                };
6571                                next if $@;
6572                        }
6573                        my $new = $root_path;
6574                        $new .= length $path ? "/$path" : '';
6575                        eval {
6576                                $root_ra->get_log([$new], $last_rev, $last_rev,
6577                                                  0, 0, 1, sub { });
6578                        };
6579                        next if $@;
6580                        $new_urls->{$ra->{repos_root}}->{$new} =
6581                                { ref_id => $ref_id,
6582                                  old_repo_id => $repo_id,
6583                                  old_path => $path };
6584                }
6585        }
6586
6587        my @emptied;
6588        foreach my $url (keys %$new_urls) {
6589                # see if we can re-use an existing [svn-remote "repo_id"]
6590                # instead of creating a(n ugly) new section:
6591                my $repo_id = $root_repos->{$url} || $url;
6592
6593                my $fetch = $new_urls->{$url};
6594                foreach my $path (keys %$fetch) {
6595                        my $x = $fetch->{$path};
6596                        Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
6597                        my $pfx = "svn-remote.$x->{old_repo_id}";
6598
6599                        my $old_fetch = quotemeta("$x->{old_path}:".
6600                                                  "$x->{ref_id}");
6601                        command_noisy(qw/config --unset/,
6602                                      "$pfx.fetch", '^'. $old_fetch . '$');
6603                        delete $r->{$x->{old_repo_id}}->
6604                               {fetch}->{$x->{old_path}};
6605                        if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
6606                                command_noisy(qw/config --unset/,
6607                                              "$pfx.url");
6608                                push @emptied, $x->{old_repo_id}
6609                        }
6610                }
6611        }
6612        if (@emptied) {
6613                my $file = $ENV{GIT_CONFIG} || "$ENV{GIT_DIR}/config";
6614                print STDERR <<EOF;
6615The following [svn-remote] sections in your config file ($file) are empty
6616and can be safely removed:
6617EOF
6618                print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
6619        }
6620}
6621
6622sub migration_check {
6623        migrate_from_v0();
6624        migrate_from_v1();
6625        migrate_from_v2();
6626        minimize_connections() if $_minimize;
6627}
6628
6629package Git::IndexInfo;
6630use strict;
6631use warnings;
6632use Git qw/command_input_pipe command_close_pipe/;
6633
6634sub new {
6635        my ($class) = @_;
6636        my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
6637        bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
6638}
6639
6640sub remove {
6641        my ($self, $path) = @_;
6642        if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
6643                return ++$self->{nr};
6644        }
6645        undef;
6646}
6647
6648sub update {
6649        my ($self, $mode, $hash, $path) = @_;
6650        if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
6651                return ++$self->{nr};
6652        }
6653        undef;
6654}
6655
6656sub DESTROY {
6657        my ($self) = @_;
6658        command_close_pipe($self->{gui}, $self->{ctx});
6659}
6660
6661package Git::SVN::GlobSpec;
6662use strict;
6663use warnings;
6664
6665sub new {
6666        my ($class, $glob, $pattern_ok) = @_;
6667        my $re = $glob;
6668        $re =~ s!/+$!!g; # no need for trailing slashes
6669        my (@left, @right, @patterns);
6670        my $state = "left";
6671        my $die_msg = "Only one set of wildcard directories " .
6672                                "(e.g. '*' or '*/*/*') is supported: '$glob'\n";
6673        for my $part (split(m|/|, $glob)) {
6674                if ($part =~ /\*/ && $part ne "*") {
6675                        die "Invalid pattern in '$glob': $part\n";
6676                } elsif ($pattern_ok && $part =~ /[{}]/ &&
6677                         $part !~ /^\{[^{}]+\}/) {
6678                        die "Invalid pattern in '$glob': $part\n";
6679                }
6680                if ($part eq "*") {
6681                        die $die_msg if $state eq "right";
6682                        $state = "pattern";
6683                        push(@patterns, "[^/]*");
6684                } elsif ($pattern_ok && $part =~ /^\{(.*)\}$/) {
6685                        die $die_msg if $state eq "right";
6686                        $state = "pattern";
6687                        my $p = quotemeta($1);
6688                        $p =~ s/\\,/|/g;
6689                        push(@patterns, "(?:$p)");
6690                } else {
6691                        if ($state eq "left") {
6692                                push(@left, $part);
6693                        } else {
6694                                push(@right, $part);
6695                                $state = "right";
6696                        }
6697                }
6698        }
6699        my $depth = @patterns;
6700        if ($depth == 0) {
6701                die "One '*' is needed in glob: '$glob'\n";
6702        }
6703        my $left = join('/', @left);
6704        my $right = join('/', @right);
6705        $re = join('/', @patterns);
6706        $re = join('\/',
6707                   grep(length, quotemeta($left), "($re)", quotemeta($right)));
6708        my $left_re = qr/^\/\Q$left\E(\/|$)/;
6709        bless { left => $left, right => $right, left_regex => $left_re,
6710                regex => qr/$re/, glob => $glob, depth => $depth }, $class;
6711}
6712
6713sub full_path {
6714        my ($self, $path) = @_;
6715        return (length $self->{left} ? "$self->{left}/" : '') .
6716               $path . (length $self->{right} ? "/$self->{right}" : '');
6717}
6718
6719__END__
6720
6721Data structures:
6722
6723
6724$remotes = { # returned by read_all_remotes()
6725        'svn' => {
6726                # svn-remote.svn.url=https://svn.musicpd.org
6727                url => 'https://svn.musicpd.org',
6728                # svn-remote.svn.fetch=mpd/trunk:trunk
6729                fetch => {
6730                        'mpd/trunk' => 'trunk',
6731                },
6732                # svn-remote.svn.tags=mpd/tags/*:tags/*
6733                tags => {
6734                        path => {
6735                                left => 'mpd/tags',
6736                                right => '',
6737                                regex => qr!mpd/tags/([^/]+)$!,
6738                                glob => 'tags/*',
6739                        },
6740                        ref => {
6741                                left => 'tags',
6742                                right => '',
6743                                regex => qr!tags/([^/]+)$!,
6744                                glob => 'tags/*',
6745                        },
6746                }
6747        }
6748};
6749
6750$log_entry hashref as returned by libsvn_log_entry()
6751{
6752        log => 'whitespace-formatted log entry
6753',                                              # trailing newline is preserved
6754        revision => '8',                        # integer
6755        date => '2004-02-24T17:01:44.108345Z',  # commit date
6756        author => 'committer name'
6757};
6758
6759
6760# this is generated by generate_diff();
6761@mods = array of diff-index line hashes, each element represents one line
6762        of diff-index output
6763
6764diff-index line ($m hash)
6765{
6766        mode_a => first column of diff-index output, no leading ':',
6767        mode_b => second column of diff-index output,
6768        sha1_b => sha1sum of the final blob,
6769        chg => change type [MCRADT],
6770        file_a => original file name of a file (iff chg is 'C' or 'R')
6771        file_b => new/current file name of a file (any chg)
6772}
6773;
6774
6775# retval of read_url_paths{,_all}();
6776$l_map = {
6777        # repository root url
6778        'https://svn.musicpd.org' => {
6779                # repository path               # GIT_SVN_ID
6780                'mpd/trunk'             =>      'trunk',
6781                'mpd/tags/0.11.5'       =>      'tags/0.11.5',
6782        },
6783}
6784
6785Notes:
6786        I don't trust the each() function on unless I created %hash myself
6787        because the internal iterator may not have started at base.