Andrew's git - gitweb.git/blob

b2931cd5aa99b6a0cf08fe5ba8485ad76b39847c
   1#!/usr/bin/env perl
   2# Copyright (C) 2006, Eric Wong <normalperson@yhbt.net>
   3# License: GPL v2 or later
   4use warnings;
   5use strict;
   6use vars qw/    $AUTHOR $VERSION
   7                $sha1 $sha1_short $_revision
   8                $_q $_authors %users/;
   9$AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
  10$VERSION = '@@GIT_VERSION@@';
  11
  12$ENV{GIT_DIR} ||= '.git';
  13$Git::SVN::default_repo_id = 'svn';
  14$Git::SVN::default_ref_id = $ENV{GIT_SVN_ID} || 'git-svn';
  15
  16$Git::SVN::Log::TZ = $ENV{TZ};
  17$ENV{TZ} = 'UTC';
  18$| = 1; # unbuffer STDOUT
  19
  20sub fatal (@) { print STDERR @_; exit 1 }
  21require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
  22require SVN::Ra;
  23require SVN::Delta;
  24if ($SVN::Core::VERSION lt '1.1.0') {
  25        fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)\n";
  26}
  27push @Git::SVN::Ra::ISA, 'SVN::Ra';
  28push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
  29push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
  30use Carp qw/croak/;
  31use IO::File qw//;
  32use File::Basename qw/dirname basename/;
  33use File::Path qw/mkpath/;
  34use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev pass_through/;
  35use IPC::Open3;
  36use Git;
  37
  38BEGIN {
  39        my $s;
  40        foreach (qw/command command_oneline command_noisy command_output_pipe
  41                    command_input_pipe command_close_pipe/) {
  42                $s .= "*SVN::Git::Editor::$_ = *SVN::Git::Fetcher::$_ = ".
  43                      "*Git::SVN::Migration::$_ = ".
  44                      "*Git::SVN::Log::$_ = *Git::SVN::$_ = *$_ = *Git::$_; ";
  45        }
  46        eval $s;
  47}
  48
  49my ($SVN);
  50
  51$sha1 = qr/[a-f\d]{40}/;
  52$sha1_short = qr/[a-f\d]{4,40}/;
  53my ($_stdin, $_help, $_edit,
  54        $_message, $_file,
  55        $_template, $_shared,
  56        $_version, $_fetch_all,
  57        $_merge, $_strategy, $_dry_run,
  58        $_prefix);
  59$Git::SVN::_follow_parent = 1;
  60my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
  61                    'config-dir=s' => \$Git::SVN::Ra::config_dir,
  62                    'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache );
  63my %fc_opts = ( 'follow-parent|follow!' => \$Git::SVN::_follow_parent,
  64                'authors-file|A=s' => \$_authors,
  65                'repack:i' => \$Git::SVN::_repack,
  66                'noMetadata' => \$Git::SVN::_no_metadata,
  67                'useSvmProps' => \$Git::SVN::_use_svm_props,
  68                'quiet|q' => \$_q,
  69                'repack-flags|repack-args|repack-opts=s' =>
  70                   \$Git::SVN::_repack_flags,
  71                %remote_opts );
  72
  73my ($_trunk, $_tags, $_branches);
  74my %init_opts = ( 'template=s' => \$_template, 'shared:s' => \$_shared,
  75                  'trunk|T=s' => \$_trunk, 'tags|t=s' => \$_tags,
  76                  'branches|b=s' => \$_branches, 'prefix=s' => \$_prefix,
  77                  %remote_opts );
  78my %cmt_opts = ( 'edit|e' => \$_edit,
  79                'rmdir' => \$SVN::Git::Editor::_rmdir,
  80                'find-copies-harder' => \$SVN::Git::Editor::_find_copies_harder,
  81                'l=i' => \$SVN::Git::Editor::_rename_limit,
  82                'copy-similarity|C=i'=> \$SVN::Git::Editor::_cp_similarity
  83);
  84
  85my %cmd = (
  86        fetch => [ \&cmd_fetch, "Download new revisions from SVN",
  87                        { 'revision|r=s' => \$_revision,
  88                          'all|a' => \$_fetch_all,
  89                           %fc_opts } ],
  90        init => [ \&cmd_init, "Initialize a repo for tracking" .
  91                          " (requires URL argument)",
  92                          \%init_opts ],
  93        'multi-init' => [ \&cmd_multi_init,
  94                          "Deprecated alias for ".
  95                          "'$0 init -T<trunk> -b<branches> -t<tags>'",
  96                          \%init_opts ],
  97        dcommit => [ \&cmd_dcommit,
  98                     'Commit several diffs to merge with upstream',
  99                        { 'merge|m|M' => \$_merge,
 100                          'strategy|s=s' => \$_strategy,
 101                          'dry-run|n' => \$_dry_run,
 102                        %cmt_opts, %fc_opts } ],
 103        'set-tree' => [ \&cmd_set_tree,
 104                        "Set an SVN repository to a git tree-ish",
 105                        { 'stdin|' => \$_stdin, %cmt_opts, %fc_opts, } ],
 106        'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
 107                        { 'revision|r=i' => \$_revision } ],
 108        'multi-fetch' => [ \&cmd_multi_fetch,
 109                           "Deprecated alias for $0 fetch --all",
 110                           { 'revision|r=s' => \$_revision, %fc_opts } ],
 111        'migrate' => [ sub { },
 112                       # no-op, we automatically run this anyways,
 113                       'Migrate configuration/metadata/layout from
 114                        previous versions of git-svn',
 115                        \%remote_opts ],
 116        'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
 117                        { 'limit=i' => \$Git::SVN::Log::limit,
 118                          'revision|r=s' => \$_revision,
 119                          'verbose|v' => \$Git::SVN::Log::verbose,
 120                          'incremental' => \$Git::SVN::Log::incremental,
 121                          'oneline' => \$Git::SVN::Log::oneline,
 122                          'show-commit' => \$Git::SVN::Log::show_commit,
 123                          'non-recursive' => \$Git::SVN::Log::non_recursive,
 124                          'authors-file|A=s' => \$_authors,
 125                          'color' => \$Git::SVN::Log::color,
 126                          'pager=s' => \$Git::SVN::Log::pager,
 127                        } ],
 128        'commit-diff' => [ \&cmd_commit_diff,
 129                           'Commit a diff between two trees',
 130                        { 'message|m=s' => \$_message,
 131                          'file|F=s' => \$_file,
 132                          'revision|r=s' => \$_revision,
 133                        %cmt_opts } ],
 134);
 135
 136my $cmd;
 137for (my $i = 0; $i < @ARGV; $i++) {
 138        if (defined $cmd{$ARGV[$i]}) {
 139                $cmd = $ARGV[$i];
 140                splice @ARGV, $i, 1;
 141                last;
 142        }
 143};
 144
 145my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
 146
 147read_repo_config(\%opts);
 148my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
 149                    'minimize-connections' => \$Git::SVN::Migration::_minimize,
 150                    'id|i=s' => \$Git::SVN::default_ref_id,
 151                    'svn-remote|remote|R=s' => \$Git::SVN::default_repo_id);
 152exit 1 if (!$rv && $cmd ne 'log');
 153
 154usage(0) if $_help;
 155version() if $_version;
 156usage(1) unless defined $cmd;
 157load_authors() if $_authors;
 158unless ($cmd =~ /^(?:init|multi-init|commit-diff)$/) {
 159        Git::SVN::Migration::migration_check();
 160}
 161Git::SVN::init_vars();
 162eval {
 163        Git::SVN::verify_remotes_sanity();
 164        $cmd{$cmd}->[0]->(@ARGV);
 165};
 166fatal $@ if $@;
 167exit 0;
 168
 169####################### primary functions ######################
 170sub usage {
 171        my $exit = shift || 0;
 172        my $fd = $exit ? \*STDERR : \*STDOUT;
 173        print $fd <<"";
 174git-svn - bidirectional operations between a single Subversion tree and git
 175Usage: $0 <command> [options] [arguments]\n
 176
 177        print $fd "Available commands:\n" unless $cmd;
 178
 179        foreach (sort keys %cmd) {
 180                next if $cmd && $cmd ne $_;
 181                print $fd '  ',pack('A17',$_),$cmd{$_}->[1],"\n";
 182                foreach (keys %{$cmd{$_}->[2]}) {
 183                        next if /^multi-/; # don't show deprecated commands
 184                        # prints out arguments as they should be passed:
 185                        my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
 186                        print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
 187                                                        "--$_" : "-$_" }
 188                                                split /\|/,$_)," $x\n";
 189                }
 190        }
 191        print $fd <<"";
 192\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
 193arbitrary identifier if you're tracking multiple SVN branches/repositories in
 194one git repository and want to keep them separate.  See git-svn(1) for more
 195information.
 196
 197        exit $exit;
 198}
 199
 200sub version {
 201        print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
 202        exit 0;
 203}
 204
 205sub do_git_init_db {
 206        unless (-d $ENV{GIT_DIR}) {
 207                my @init_db = ('init');
 208                push @init_db, "--template=$_template" if defined $_template;
 209                if (defined $_shared) {
 210                        if ($_shared =~ /[a-z]/) {
 211                                push @init_db, "--shared=$_shared";
 212                        } else {
 213                                push @init_db, "--shared";
 214                        }
 215                }
 216                command_noisy(@init_db);
 217        }
 218}
 219
 220sub init_subdir {
 221        my $repo_path = shift or return;
 222        mkpath([$repo_path]) unless -d $repo_path;
 223        chdir $repo_path or die "Couldn't chdir to $repo_path: $!\n";
 224        $ENV{GIT_DIR} = $repo_path . "/.git";
 225}
 226
 227sub cmd_init {
 228        if (defined $_trunk || defined $_branches || defined $_tags) {
 229                return cmd_multi_init(@_);
 230        }
 231        my $url = shift or die "SVN repository location required ",
 232                               "as a command-line argument\n";
 233        init_subdir(@_);
 234        do_git_init_db();
 235
 236        Git::SVN->init($url);
 237}
 238
 239sub cmd_fetch {
 240        if (grep /^\d+=./, @_) {
 241                die "'<rev>=<commit>' fetch arguments are ",
 242                    "no longer supported.\n";
 243        }
 244        my ($remote) = @_;
 245        if (@_ > 1) {
 246                die "Usage: $0 fetch [--all|-a] [svn-remote]\n";
 247        }
 248        $remote ||= $Git::SVN::default_repo_id;
 249        if ($_fetch_all) {
 250                cmd_multi_fetch();
 251        } else {
 252                Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
 253        }
 254}
 255
 256sub cmd_set_tree {
 257        my (@commits) = @_;
 258        if ($_stdin || !@commits) {
 259                print "Reading from stdin...\n";
 260                @commits = ();
 261                while (<STDIN>) {
 262                        if (/\b($sha1_short)\b/o) {
 263                                unshift @commits, $1;
 264                        }
 265                }
 266        }
 267        my @revs;
 268        foreach my $c (@commits) {
 269                my @tmp = command('rev-parse',$c);
 270                if (scalar @tmp == 1) {
 271                        push @revs, $tmp[0];
 272                } elsif (scalar @tmp > 1) {
 273                        push @revs, reverse(command('rev-list',@tmp));
 274                } else {
 275                        fatal "Failed to rev-parse $c\n";
 276                }
 277        }
 278        my $gs = Git::SVN->new;
 279        my ($r_last, $cmt_last) = $gs->last_rev_commit;
 280        $gs->fetch;
 281        if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
 282                fatal "There are new revisions that were fetched ",
 283                      "and need to be merged (or acknowledged) ",
 284                      "before committing.\nlast rev: $r_last\n",
 285                      " current: $gs->{last_rev}\n";
 286        }
 287        $gs->set_tree($_) foreach @revs;
 288        print "Done committing ",scalar @revs," revisions to SVN\n";
 289}
 290
 291sub cmd_dcommit {
 292        my $head = shift;
 293        $head ||= 'HEAD';
 294        my ($url, $rev, $uuid);
 295        my ($fh, $ctx) = command_output_pipe('rev-list', $head);
 296        my @refs;
 297        my $c;
 298        while (<$fh>) {
 299                $c = $_;
 300                chomp $c;
 301                ($url, $rev, $uuid) = cmt_metadata($c);
 302                last if (defined $url && defined $rev && defined $uuid);
 303                unshift @refs, $c;
 304        }
 305        close $fh; # most likely breaking the pipe
 306        unless (defined $url && defined $rev && defined $uuid) {
 307                die "Unable to determine upstream SVN information from ",
 308                    "$head history:\n  $ctx\n";
 309        }
 310        my $gs = Git::SVN->find_by_url($url) or
 311                           die "Can't determine fetch information for $url\n";
 312        my $last_rev;
 313        foreach my $d (@refs) {
 314                if (!verify_ref("$d~1")) {
 315                        fatal "Commit $d\n",
 316                              "has no parent commit, and therefore ",
 317                              "nothing to diff against.\n",
 318                              "You should be working from a repository ",
 319                              "originally created by git-svn\n";
 320                }
 321                unless (defined $last_rev) {
 322                        (undef, $last_rev, undef) = cmt_metadata("$d~1");
 323                        unless (defined $last_rev) {
 324                                fatal "Unable to extract revision information ",
 325                                      "from commit $d~1\n";
 326                        }
 327                }
 328                if ($_dry_run) {
 329                        print "diff-tree $d~1 $d\n";
 330                } else {
 331                        my %ed_opts = ( r => $last_rev,
 332                                        log => get_commit_entry($d)->{log},
 333                                        ra => Git::SVN::Ra->new($url),
 334                                        tree_a => "$d~1",
 335                                        tree_b => $d,
 336                                        editor_cb => sub {
 337                                               print "Committed r$_[0]\n";
 338                                               $last_rev = $_[0]; },
 339                                        svn_path => '');
 340                        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
 341                                print "No changes\n$d~1 == $d\n";
 342                        }
 343                }
 344        }
 345        return if $_dry_run;
 346        $gs->fetch;
 347        # we always want to rebase against the current HEAD, not any
 348        # head that was passed to us
 349        my @diff = command('diff-tree', 'HEAD', $gs->refname, '--');
 350        my @finish;
 351        if (@diff) {
 352                @finish = qw/rebase/;
 353                push @finish, qw/--merge/ if $_merge;
 354                push @finish, "--strategy=$_strategy" if $_strategy;
 355                print STDERR "W: HEAD and ", $gs->refname, " differ, ",
 356                             "using @finish:\n", "@diff";
 357        } else {
 358                print "No changes between current HEAD and ",
 359                      $gs->refname, "\nResetting to the latest ",
 360                      $gs->refname, "\n";
 361                @finish = qw/reset --mixed/;
 362        }
 363        command_noisy(@finish, $gs->refname);
 364}
 365
 366sub cmd_show_ignore {
 367        my $gs = Git::SVN->new;
 368        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 369        $gs->traverse_ignore(\*STDOUT, '', $r);
 370}
 371
 372sub cmd_multi_init {
 373        my $url = shift;
 374        unless (defined $_trunk || defined $_branches || defined $_tags) {
 375                usage(1);
 376        }
 377        do_git_init_db();
 378        $_prefix = '' unless defined $_prefix;
 379        if (defined $url) {
 380                $url =~ s#/+$##;
 381                init_subdir(@_);
 382        }
 383        if (defined $_trunk) {
 384                my $trunk_ref = $_prefix . 'trunk';
 385                # try both old-style and new-style lookups:
 386                my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
 387                unless ($gs_trunk) {
 388                        my ($trunk_url, $trunk_path) =
 389                                              complete_svn_url($url, $_trunk);
 390                        $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
 391                                                   undef, $trunk_ref);
 392                }
 393        }
 394        return unless defined $_branches || defined $_tags;
 395        my $ra = $url ? Git::SVN::Ra->new($url) : undef;
 396        complete_url_ls_init($ra, $_branches, '--branches/-b', $_prefix);
 397        complete_url_ls_init($ra, $_tags, '--tags/-t', $_prefix . 'tags/');
 398}
 399
 400sub cmd_multi_fetch {
 401        my $remotes = Git::SVN::read_all_remotes();
 402        foreach my $repo_id (sort keys %$remotes) {
 403                if ($remotes->{$repo_id}->{url}) {
 404                        Git::SVN::fetch_all($repo_id, $remotes);
 405                }
 406        }
 407}
 408
 409# this command is special because it requires no metadata
 410sub cmd_commit_diff {
 411        my ($ta, $tb, $url) = @_;
 412        my $usage = "Usage: $0 commit-diff -r<revision> ".
 413                    "<tree-ish> <tree-ish> [<URL>]\n";
 414        fatal($usage) if (!defined $ta || !defined $tb);
 415        my $svn_path;
 416        if (!defined $url) {
 417                my $gs = eval { Git::SVN->new };
 418                if (!$gs) {
 419                        fatal("Needed URL or usable git-svn --id in ",
 420                              "the command-line\n", $usage);
 421                }
 422                $url = $gs->{url};
 423                $svn_path = $gs->{path};
 424        }
 425        unless (defined $_revision) {
 426                fatal("-r|--revision is a required argument\n", $usage);
 427        }
 428        if (defined $_message && defined $_file) {
 429                fatal("Both --message/-m and --file/-F specified ",
 430                      "for the commit message.\n",
 431                      "I have no idea what you mean\n");
 432        }
 433        if (defined $_file) {
 434                $_message = file_to_s($_file);
 435        } else {
 436                $_message ||= get_commit_entry($tb)->{log};
 437        }
 438        my $ra ||= Git::SVN::Ra->new($url);
 439        $svn_path ||= $ra->{svn_path};
 440        my $r = $_revision;
 441        if ($r eq 'HEAD') {
 442                $r = $ra->get_latest_revnum;
 443        } elsif ($r !~ /^\d+$/) {
 444                die "revision argument: $r not understood by git-svn\n";
 445        }
 446        my %ed_opts = ( r => $r,
 447                        log => $_message,
 448                        ra => $ra,
 449                        tree_a => $ta,
 450                        tree_b => $tb,
 451                        editor_cb => sub { print "Committed r$_[0]\n" },
 452                        svn_path => $svn_path );
 453        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
 454                print "No changes\n$ta == $tb\n";
 455        }
 456}
 457
 458########################### utility functions #########################
 459
 460sub complete_svn_url {
 461        my ($url, $path) = @_;
 462        $path =~ s#/+$##;
 463        if ($path !~ m#^[a-z\+]+://#) {
 464                if (!defined $url || $url !~ m#^[a-z\+]+://#) {
 465                        fatal("E: '$path' is not a complete URL ",
 466                              "and a separate URL is not specified\n");
 467                }
 468                return ($url, $path);
 469        }
 470        return ($path, '');
 471}
 472
 473sub complete_url_ls_init {
 474        my ($ra, $repo_path, $switch, $pfx) = @_;
 475        unless ($repo_path) {
 476                print STDERR "W: $switch not specified\n";
 477                return;
 478        }
 479        $repo_path =~ s#/+$##;
 480        if ($repo_path =~ m#^[a-z\+]+://#) {
 481                $ra = Git::SVN::Ra->new($repo_path);
 482                $repo_path = '';
 483        } else {
 484                $repo_path =~ s#^/+##;
 485                unless ($ra) {
 486                        fatal("E: '$repo_path' is not a complete URL ",
 487                              "and a separate URL is not specified\n");
 488                }
 489        }
 490        my $r = defined $_revision ? $_revision : $ra->get_latest_revnum;
 491        my ($dirent, undef, undef) = $ra->get_dir($repo_path, $r);
 492        my $url = $ra->{url};
 493        my $remote_id;
 494        my $remote_path;
 495        foreach my $d (sort keys %$dirent) {
 496                next if ($dirent->{$d}->kind != $SVN::Node::dir);
 497                my $path =  "$repo_path/$d";
 498                my $ref = "$pfx$d";
 499                my $gs = eval { Git::SVN->new($ref) };
 500                # don't try to init already existing refs
 501                unless ($gs) {
 502                        print "init $url/$path => $ref\n";
 503                        $gs = Git::SVN->init($url, $path, undef, $ref, 1);
 504                }
 505                if ($gs) {
 506                        my $k = "svn-remote.$gs->{repo_id}.url";
 507                        my $orig_url = eval {
 508                                command_oneline(qw/config --get/, $k)
 509                        };
 510                        if ($orig_url && ($orig_url ne $gs->{url})) {
 511                                die "$k already set: $orig_url\n",
 512                                    "wanted to set to: $gs->{url}\n";
 513                        }
 514                        unless ($orig_url) {
 515                                command_oneline('config', $k, $gs->{url});
 516                        }
 517                        $remote_id = $gs->{repo_id};
 518                        last;
 519                }
 520        }
 521        if (defined $remote_id) {
 522                $remote_path = "$ra->{svn_path}/$repo_path/*";
 523                $remote_path =~ s#/+#/#g;
 524                $remote_path =~ s#^/##g;
 525                my ($n) = ($switch =~ /^--(\w+)/);
 526                if (length $pfx && $pfx !~ m#/$#) {
 527                        die "--prefix='$pfx' must have a trailing slash '/'\n";
 528                }
 529                command_noisy('config', "svn-remote.$remote_id.$n",
 530                                        "$remote_path:refs/remotes/$pfx*");
 531        }
 532}
 533
 534sub verify_ref {
 535        my ($ref) = @_;
 536        eval { command_oneline([ 'rev-parse', '--verify', $ref ],
 537                               { STDERR => 0 }); };
 538}
 539
 540sub get_tree_from_treeish {
 541        my ($treeish) = @_;
 542        # $treeish can be a symbolic ref, too:
 543        my $type = command_oneline(qw/cat-file -t/, $treeish);
 544        my $expected;
 545        while ($type eq 'tag') {
 546                ($treeish, $type) = command(qw/cat-file tag/, $treeish);
 547        }
 548        if ($type eq 'commit') {
 549                $expected = (grep /^tree /, command(qw/cat-file commit/,
 550                                                    $treeish))[0];
 551                ($expected) = ($expected =~ /^tree ($sha1)$/o);
 552                die "Unable to get tree from $treeish\n" unless $expected;
 553        } elsif ($type eq 'tree') {
 554                $expected = $treeish;
 555        } else {
 556                die "$treeish is a $type, expected tree, tag or commit\n";
 557        }
 558        return $expected;
 559}
 560
 561sub get_commit_entry {
 562        my ($treeish) = shift;
 563        my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
 564        my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
 565        my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
 566        open my $log_fh, '>', $commit_editmsg or croak $!;
 567
 568        my $type = command_oneline(qw/cat-file -t/, $treeish);
 569        if ($type eq 'commit' || $type eq 'tag') {
 570                my ($msg_fh, $ctx) = command_output_pipe('cat-file',
 571                                                         $type, $treeish);
 572                my $in_msg = 0;
 573                while (<$msg_fh>) {
 574                        if (!$in_msg) {
 575                                $in_msg = 1 if (/^\s*$/);
 576                        } elsif (/^git-svn-id: /) {
 577                                # skip this for now, we regenerate the
 578                                # correct one on re-fetch anyways
 579                                # TODO: set *:merge properties or like...
 580                        } else {
 581                                print $log_fh $_ or croak $!;
 582                        }
 583                }
 584                command_close_pipe($msg_fh, $ctx);
 585        }
 586        close $log_fh or croak $!;
 587
 588        if ($_edit || ($type eq 'tree')) {
 589                my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
 590                # TODO: strip out spaces, comments, like git-commit.sh
 591                system($editor, $commit_editmsg);
 592        }
 593        rename $commit_editmsg, $commit_msg or croak $!;
 594        open $log_fh, '<', $commit_msg or croak $!;
 595        { local $/; chomp($log_entry{log} = <$log_fh>); }
 596        close $log_fh or croak $!;
 597        unlink $commit_msg;
 598        \%log_entry;
 599}
 600
 601sub s_to_file {
 602        my ($str, $file, $mode) = @_;
 603        open my $fd,'>',$file or croak $!;
 604        print $fd $str,"\n" or croak $!;
 605        close $fd or croak $!;
 606        chmod ($mode &~ umask, $file) if (defined $mode);
 607}
 608
 609sub file_to_s {
 610        my $file = shift;
 611        open my $fd,'<',$file or croak "$!: file: $file\n";
 612        local $/;
 613        my $ret = <$fd>;
 614        close $fd or croak $!;
 615        $ret =~ s/\s*$//s;
 616        return $ret;
 617}
 618
 619# '<svn username> = real-name <email address>' mapping based on git-svnimport:
 620sub load_authors {
 621        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
 622        my $log = $cmd eq 'log';
 623        while (<$authors>) {
 624                chomp;
 625                next unless /^(\S+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
 626                my ($user, $name, $email) = ($1, $2, $3);
 627                if ($log) {
 628                        $Git::SVN::Log::rusers{"$name <$email>"} = $user;
 629                } else {
 630                        $users{$user} = [$name, $email];
 631                }
 632        }
 633        close $authors or croak $!;
 634}
 635
 636# convert GetOpt::Long specs for use by git-config
 637sub read_repo_config {
 638        return unless -d $ENV{GIT_DIR};
 639        my $opts = shift;
 640        my @config_only;
 641        foreach my $o (keys %$opts) {
 642                # if we have mixedCase and a long option-only, then
 643                # it's a config-only variable that we don't need for
 644                # the command-line.
 645                push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
 646                my $v = $opts->{$o};
 647                my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
 648                $key =~ s/-//g;
 649                my $arg = 'git-config';
 650                $arg .= ' --int' if ($o =~ /[:=]i$/);
 651                $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
 652                if (ref $v eq 'ARRAY') {
 653                        chomp(my @tmp = `$arg --get-all svn.$key`);
 654                        @$v = @tmp if @tmp;
 655                } else {
 656                        chomp(my $tmp = `$arg --get svn.$key`);
 657                        if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
 658                                $$v = $tmp;
 659                        }
 660                }
 661        }
 662        delete @$opts{@config_only} if @config_only;
 663}
 664
 665sub extract_metadata {
 666        my $id = shift or return (undef, undef, undef);
 667        my ($url, $rev, $uuid) = ($id =~ /^git-svn-id:\s(\S+?)\@(\d+)
 668                                                        \s([a-f\d\-]+)$/x);
 669        if (!defined $rev || !$uuid || !$url) {
 670                # some of the original repositories I made had
 671                # identifiers like this:
 672                ($rev, $uuid) = ($id =~/^git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
 673        }
 674        return ($url, $rev, $uuid);
 675}
 676
 677sub cmt_metadata {
 678        return extract_metadata((grep(/^git-svn-id: /,
 679                command(qw/cat-file commit/, shift)))[-1]);
 680}
 681
 682package Git::SVN;
 683use strict;
 684use warnings;
 685use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
 686            $_repack $_repack_flags $_use_svm_props/;
 687use Carp qw/croak/;
 688use File::Path qw/mkpath/;
 689use File::Copy qw/copy/;
 690use IPC::Open3;
 691
 692my $_repack_nr;
 693# properties that we do not log:
 694my %SKIP_PROP;
 695BEGIN {
 696        %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
 697                                        svn:special svn:executable
 698                                        svn:entry:committed-rev
 699                                        svn:entry:last-author
 700                                        svn:entry:uuid
 701                                        svn:entry:committed-date/;
 702
 703        # some options are read globally, but can be overridden locally
 704        # per [svn-remote "..."] section.  Command-line options will *NOT*
 705        # override options set in an [svn-remote "..."] section
 706        my $e;
 707        foreach (qw/follow_parent no_metadata use_svm_props/) {
 708                my $key = $_;
 709                $key =~ tr/_//d;
 710                $e .= "sub $_ {
 711                        my (\$self) = \@_;
 712                        return \$self->{-$_} if exists \$self->{-$_};
 713                        my \$k = \"svn-remote.\$self->{repo_id}\.$key\";
 714                        eval { command_oneline(qw/config --get/, \$k) };
 715                        if (\$@) {
 716                                \$self->{-$_} = \$Git::SVN::_$_;
 717                        } else {
 718                                my \$v = command_oneline(qw/config --bool/,\$k);
 719                                \$self->{-$_} = \$v eq 'false' ? 0 : 1;
 720                        }
 721                        return \$self->{-$_} }\n";
 722        }
 723        $e .= "1;\n";
 724        eval $e or die $@;
 725}
 726
 727my %LOCKFILES;
 728END { unlink keys %LOCKFILES if %LOCKFILES }
 729
 730sub resolve_local_globs {
 731        my ($url, $fetch, $glob_spec) = @_;
 732        return unless defined $glob_spec;
 733        my $ref = $glob_spec->{ref};
 734        my $path = $glob_spec->{path};
 735        foreach (command(qw#for-each-ref --format=%(refname) refs/remotes#)) {
 736                next unless m#^refs/remotes/$ref->{regex}$#;
 737                my $p = $1;
 738                my $pathname = $path->full_path($p);
 739                my $refname = $ref->full_path($p);
 740                if (my $existing = $fetch->{$pathname}) {
 741                        if ($existing ne $refname) {
 742                                die "Refspec conflict:\n",
 743                                    "existing: refs/remotes/$existing\n",
 744                                    " globbed: refs/remotes/$refname\n";
 745                        }
 746                        my $u = (::cmt_metadata("refs/remotes/$refname"))[0];
 747                        $u =~ s!^\Q$url\E(/|$)!! or die
 748                          "refs/remotes/$refname: '$url' not found in '$u'\n";
 749                        if ($pathname ne $u) {
 750                                warn "W: Refspec glob conflict ",
 751                                     "(ref: refs/remotes/$refname):\n",
 752                                     "expected path: $pathname\n",
 753                                     "    real path: $u\n",
 754                                     "Continuing ahead with $u\n";
 755                                next;
 756                        }
 757                } else {
 758                        $fetch->{$pathname} = $refname;
 759                }
 760        }
 761}
 762
 763sub parse_revision_argument {
 764        my ($base, $head) = @_;
 765        if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
 766                return ($base, $head);
 767        }
 768        return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
 769        return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
 770        return ($head, $head) if ($::_revision eq 'HEAD');
 771        return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
 772        return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
 773        die "revision argument: $::_revision not understood by git-svn\n";
 774}
 775
 776sub fetch_all {
 777        my ($repo_id, $remotes) = @_;
 778        my $remote = $remotes->{$repo_id};
 779        my $fetch = $remote->{fetch};
 780        my $url = $remote->{url};
 781        my (@gs, @globs);
 782        my $ra = Git::SVN::Ra->new($url);
 783        my $uuid = $ra->get_uuid;
 784        my $head = $ra->get_latest_revnum;
 785        my $base = $head;
 786
 787        # read the max revs for wildcard expansion (branches/*, tags/*)
 788        foreach my $t (qw/branches tags/) {
 789                defined $remote->{$t} or next;
 790                push @globs, $remote->{$t};
 791                my $max_rev = eval { tmp_config(qw/--int --get/,
 792                                         "svn-remote.$repo_id.${t}-maxRev") };
 793                if (defined $max_rev && ($max_rev < $base)) {
 794                        $base = $max_rev;
 795                }
 796        }
 797
 798        if ($fetch) {
 799                foreach my $p (sort keys %$fetch) {
 800                        my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
 801                        my $lr = $gs->rev_db_max;
 802                        if (defined $lr) {
 803                                $base = $lr if ($lr < $base);
 804                        }
 805                        push @gs, $gs;
 806                }
 807        }
 808
 809        ($base, $head) = parse_revision_argument($base, $head);
 810        $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
 811}
 812
 813sub read_all_remotes {
 814        my $r = {};
 815        foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
 816                if (m!^(.+)\.fetch=\s*(.*)\s*:\s*refs/remotes/(.+)\s*$!) {
 817                        $r->{$1}->{fetch}->{$2} = $3;
 818                } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
 819                        $r->{$1}->{url} = $2;
 820                } elsif (m!^(.+)\.(branches|tags)=
 821                           (.*):refs/remotes/(.+)\s*$/!x) {
 822                        my ($p, $g) = ($3, $4);
 823                        my $rs = $r->{$1}->{$2} = {
 824                                          t => $2,
 825                                          remote => $1,
 826                                          path => Git::SVN::GlobSpec->new($p),
 827                                          ref => Git::SVN::GlobSpec->new($g) };
 828                        if (length($rs->{ref}->{right}) != 0) {
 829                                die "The '*' glob character must be the last ",
 830                                    "character of '$g'\n";
 831                        }
 832                }
 833        }
 834        $r;
 835}
 836
 837sub init_vars {
 838        if (defined $_repack) {
 839                $_repack = 1000 if ($_repack <= 0);
 840                $_repack_nr = $_repack;
 841                $_repack_flags ||= '-d';
 842        }
 843}
 844
 845sub verify_remotes_sanity {
 846        return unless -d $ENV{GIT_DIR};
 847        my %seen;
 848        foreach (command(qw/config -l/)) {
 849                if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
 850                        if ($seen{$1}) {
 851                                die "Remote ref refs/remote/$1 is tracked by",
 852                                    "\n  \"$_\"\nand\n  \"$seen{$1}\"\n",
 853                                    "Please resolve this ambiguity in ",
 854                                    "your git configuration file before ",
 855                                    "continuing\n";
 856                        }
 857                        $seen{$1} = $_;
 858                }
 859        }
 860}
 861
 862# we allow more chars than remotes2config.sh...
 863sub sanitize_remote_name {
 864        my ($name) = @_;
 865        $name =~ tr{A-Za-z0-9:,/+-}{.}c;
 866        $name;
 867}
 868
 869sub find_existing_remote {
 870        my ($url, $remotes) = @_;
 871        my $existing;
 872        foreach my $repo_id (keys %$remotes) {
 873                my $u = $remotes->{$repo_id}->{url} or next;
 874                next if $u ne $url;
 875                $existing = $repo_id;
 876                last;
 877        }
 878        $existing;
 879}
 880
 881sub init_remote_config {
 882        my ($self, $url, $no_write) = @_;
 883        $url =~ s!/+$!!; # strip trailing slash
 884        my $r = read_all_remotes();
 885        my $existing = find_existing_remote($url, $r);
 886        if ($existing) {
 887                unless ($no_write) {
 888                        print STDERR "Using existing ",
 889                                     "[svn-remote \"$existing\"]\n";
 890                }
 891                $self->{repo_id} = $existing;
 892        } else {
 893                my $min_url = Git::SVN::Ra->new($url)->minimize_url;
 894                $existing = find_existing_remote($min_url, $r);
 895                if ($existing) {
 896                        unless ($no_write) {
 897                                print STDERR "Using existing ",
 898                                             "[svn-remote \"$existing\"]\n";
 899                        }
 900                        $self->{repo_id} = $existing;
 901                }
 902                if ($min_url ne $url) {
 903                        unless ($no_write) {
 904                                print STDERR "Using higher level of URL: ",
 905                                             "$url => $min_url\n";
 906                        }
 907                        my $old_path = $self->{path};
 908                        $self->{path} = $url;
 909                        $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
 910                        if (length $old_path) {
 911                                $self->{path} .= "/$old_path";
 912                        }
 913                        $url = $min_url;
 914                }
 915        }
 916        my $orig_url;
 917        if (!$existing) {
 918                # verify that we aren't overwriting anything:
 919                $orig_url = eval {
 920                        command_oneline('config', '--get',
 921                                        "svn-remote.$self->{repo_id}.url")
 922                };
 923                if ($orig_url && ($orig_url ne $url)) {
 924                        die "svn-remote.$self->{repo_id}.url already set: ",
 925                            "$orig_url\nwanted to set to: $url\n";
 926                }
 927        }
 928        my ($xrepo_id, $xpath) = find_ref($self->refname);
 929        if (defined $xpath) {
 930                die "svn-remote.$xrepo_id.fetch already set to track ",
 931                    "$xpath:refs/remotes/", $self->refname, "\n";
 932        }
 933        unless ($no_write) {
 934                command_noisy('config',
 935                              "svn-remote.$self->{repo_id}.url", $url);
 936                command_noisy('config', '--add',
 937                              "svn-remote.$self->{repo_id}.fetch",
 938                              "$self->{path}:".$self->refname);
 939        }
 940        $self->{url} = $url;
 941}
 942
 943sub find_by_url { # repos_root and, path are optional
 944        my ($class, $full_url, $repos_root, $path) = @_;
 945        my $remotes = read_all_remotes();
 946        if (defined $full_url && defined $repos_root && !defined $path) {
 947                $path = $full_url;
 948                $path =~ s#^\Q$repos_root\E(?:/|$)##;
 949        }
 950        foreach my $repo_id (keys %$remotes) {
 951                my $u = $remotes->{$repo_id}->{url} or next;
 952                next if defined $repos_root && $repos_root ne $u;
 953
 954                my $fetch = $remotes->{$repo_id}->{fetch} || {};
 955                foreach (qw/branches tags/) {
 956                        resolve_local_globs($u, $fetch,
 957                                            $remotes->{$repo_id}->{$_});
 958                }
 959                my $p = $path;
 960                unless (defined $p) {
 961                        $p = $full_url;
 962                        $p =~ s#^\Q$u\E(?:/|$)## or next;
 963                }
 964                foreach my $f (keys %$fetch) {
 965                        next if $f ne $p;
 966                        return Git::SVN->new($fetch->{$f}, $repo_id, $f);
 967                }
 968        }
 969        undef;
 970}
 971
 972sub init {
 973        my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
 974        my $self = _new($class, $repo_id, $ref_id, $path);
 975        if (defined $url) {
 976                $self->init_remote_config($url, $no_write);
 977        }
 978        $self;
 979}
 980
 981sub find_ref {
 982        my ($ref_id) = @_;
 983        foreach (command(qw/config -l/)) {
 984                next unless m!^svn-remote\.(.+)\.fetch=
 985                              \s*(.*)\s*:\s*refs/remotes/(.+)\s*$!x;
 986                my ($repo_id, $path, $ref) = ($1, $2, $3);
 987                if ($ref eq $ref_id) {
 988                        $path = '' if ($path =~ m#^\./?#);
 989                        return ($repo_id, $path);
 990                }
 991        }
 992        (undef, undef, undef);
 993}
 994
 995sub new {
 996        my ($class, $ref_id, $repo_id, $path) = @_;
 997        if (defined $ref_id && !defined $repo_id && !defined $path) {
 998                ($repo_id, $path) = find_ref($ref_id);
 999                if (!defined $repo_id) {
1000                        die "Could not find a \"svn-remote.*.fetch\" key ",
1001                            "in the repository configuration matching: ",
1002                            "refs/remotes/$ref_id\n";
1003                }
1004        }
1005        my $self = _new($class, $repo_id, $ref_id, $path);
1006        if (!defined $self->{path} || !length $self->{path}) {
1007                my $fetch = command_oneline('config', '--get',
1008                                            "svn-remote.$repo_id.fetch",
1009                                            ":refs/remotes/$ref_id\$") or
1010                     die "Failed to read \"svn-remote.$repo_id.fetch\" ",
1011                         "\":refs/remotes/$ref_id\$\" in config\n";
1012                ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
1013        }
1014        $self->{url} = command_oneline('config', '--get',
1015                                       "svn-remote.$repo_id.url") or
1016                  die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
1017        if ((-z $self->db_path || ! -e $self->db_path) &&
1018            ::verify_ref($self->refname.'^0')) {
1019                $self->rebuild;
1020        }
1021        $self;
1022}
1023
1024sub refname { "refs/remotes/$_[0]->{ref_id}" }
1025
1026sub svm_uuid {
1027        my ($self) = @_;
1028        return $self->{svm}->{uuid} if $self->svm;
1029        $self->ra;
1030        unless ($self->{svm}) {
1031                die "SVM UUID not cached, and reading remotely failed\n";
1032        }
1033        $self->{svm}->{uuid};
1034}
1035
1036sub svm {
1037        my ($self) = @_;
1038        return $self->{svm} if $self->{svm};
1039        my $svm;
1040        # see if we have it in our config, first:
1041        eval {
1042                my $section = "svn-remote.$self->{repo_id}";
1043                $svm = {
1044                  source => tmp_config('--get', "$section.svm-source"),
1045                  uuid => tmp_config('--get', "$section.svm-uuid"),
1046                }
1047        };
1048        $self->{svm} = $svm if ($svm && $svm->{source} && $svm->{uuid});
1049        $self->{svm};
1050}
1051
1052sub _set_svm_vars {
1053        my ($self, $ra) = @_;
1054        return $ra if $self->svm;
1055
1056        my @err = ( "useSvmProps set, but failed to read SVM properties\n",
1057                    "(svm:source, svm:mirror, svm:mirror) ",
1058                    "from the following URLs:\n" );
1059        sub read_svm_props {
1060                my ($self, $props) = @_;
1061                my $src = $props->{'svm:source'};
1062                my $mirror = $props->{'svm:mirror'};
1063                my $uuid = $props->{'svm:uuid'};
1064                return undef if (!$src || !$mirror || !$uuid);
1065
1066                chomp($src, $mirror, $uuid);
1067
1068                $uuid =~ m{^[0-9a-f\-]{30,}$}
1069                    or die "doesn't look right - svm:uuid is '$uuid'\n";
1070                # don't know what a '!' is there for, also the
1071                # username is of no interest
1072                $src =~ s{/?!$}{$mirror};
1073                $src =~ s{/+$}{}; # no trailing slashes please
1074                $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
1075
1076                my $section = "svn-remote.$self->{repo_id}";
1077                tmp_config('--add', "$section.svm-source", $src);
1078                tmp_config('--add', "$section.svm-uuid", $uuid);
1079                $self->{svm} = { source => $src , uuid => $uuid };
1080                return 1;
1081        }
1082
1083        my $r = $ra->get_latest_revnum;
1084        my $path = $self->{path};
1085        my @tried_a = ($path);
1086        while (length $path) {
1087                if ($self->read_svm_props(($ra->get_dir($path, $r))[2])) {
1088                        return $ra;
1089                }
1090                $path =~ s#/?[^/]+$## && push @tried_a, $path;
1091        }
1092        if ($self->read_svm_props(($ra->get_dir('', $r))[2])) {
1093                return $ra;
1094        }
1095
1096        if ($ra->{repos_root} eq $self->{url}) {
1097                die @err, map { "  $self->{url}/$_\n" } @tried_a, "\n";
1098        }
1099
1100        # nope, make sure we're connected to the repository root:
1101        my $ok;
1102        my @tried_b;
1103        $path = $ra->{svn_path};
1104        $path =~ s#/?[^/]+$##; # we already tried this one above
1105        $ra = Git::SVN::Ra->new($ra->{repos_root});
1106        while (length $path) {
1107                $ok = $self->read_svm_props(($ra->get_dir($path, $r))[2]);
1108                last if $ok;
1109                $path =~ s#/?[^/]+$## && push @tried_b, $path;
1110        }
1111        $ok = $self->read_svm_props(($ra->get_dir('', $r))[2]) unless $ok;
1112        if (!$ok) {
1113                die @err, map { "  $self->{url}/$_\n" } @tried_a, "\n",
1114                          map { "  $ra->{url}/$_\n" } @tried_b, "\n"
1115        }
1116        Git::SVN::Ra->new($self->{url});
1117}
1118
1119# this allows us to memoize our SVN::Ra UUID locally and avoid a
1120# remote lookup (useful for 'git svn log').
1121sub ra_uuid {
1122        my ($self) = @_;
1123        unless ($self->{ra_uuid}) {
1124                my $key = "svn-remote.$self->{repo_id}.uuid";
1125                my $uuid = eval { tmp_config('--get', $key) };
1126                if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/) {
1127                        $self->{ra_uuid} = $uuid;
1128                } else {
1129                        die "ra_uuid called without URL\n" unless $self->{url};
1130                        $self->{ra_uuid} = $self->ra->get_uuid;
1131                        tmp_config('--add', $key, $self->{ra_uuid});
1132                }
1133        }
1134        $self->{ra_uuid};
1135}
1136
1137sub ra {
1138        my ($self) = shift;
1139        my $ra = Git::SVN::Ra->new($self->{url});
1140        if ($self->use_svm_props && !$self->{svm}) {
1141                if ($self->no_metadata) {
1142                        die "Can't have both 'noMetadata' and ",
1143                            "'useSvmProps' options set!\n";
1144                }
1145                $ra = $self->_set_svm_vars($ra);
1146                $self->{-want_revprops} = 1;
1147        }
1148        $ra;
1149}
1150
1151sub rel_path {
1152        my ($self) = @_;
1153        my $repos_root = $self->ra->{repos_root};
1154        return $self->{path} if ($self->{url} eq $repos_root);
1155        die "BUG: rel_path failed! repos_root: $repos_root, Ra URL: ",
1156            $self->ra->{url}, " path: $self->{path},  URL: $self->{url}\n";
1157}
1158
1159sub traverse_ignore {
1160        my ($self, $fh, $path, $r) = @_;
1161        $path =~ s#^/+##g;
1162        my $ra = $self->ra;
1163        my ($dirent, undef, $props) = $ra->get_dir($path, $r);
1164        my $p = $path;
1165        $p =~ s#^\Q$ra->{svn_path}\E/##;
1166        print $fh length $p ? "\n# $p\n" : "\n# /\n";
1167        if (my $s = $props->{'svn:ignore'}) {
1168                $s =~ s/[\r\n]+/\n/g;
1169                chomp $s;
1170                if (length $p == 0) {
1171                        $s =~ s#\n#\n/$p#g;
1172                        print $fh "/$s\n";
1173                } else {
1174                        $s =~ s#\n#\n/$p/#g;
1175                        print $fh "/$p/$s\n";
1176                }
1177        }
1178        foreach (sort keys %$dirent) {
1179                next if $dirent->{$_}->kind != $SVN::Node::dir;
1180                $self->traverse_ignore($fh, "$path/$_", $r);
1181        }
1182}
1183
1184sub last_rev { ($_[0]->last_rev_commit)[0] }
1185sub last_commit { ($_[0]->last_rev_commit)[1] }
1186
1187# returns the newest SVN revision number and newest commit SHA1
1188sub last_rev_commit {
1189        my ($self) = @_;
1190        if (defined $self->{last_rev} && defined $self->{last_commit}) {
1191                return ($self->{last_rev}, $self->{last_commit});
1192        }
1193        my $c = ::verify_ref($self->refname.'^0');
1194        if ($c && !$self->use_svm_props && !$self->no_metadata) {
1195                my $rev = (::cmt_metadata($c))[1];
1196                if (defined $rev) {
1197                        ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
1198                        return ($rev, $c);
1199                }
1200        }
1201        my $db_path = $self->db_path;
1202        unless (-e $db_path) {
1203                ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
1204                return (undef, undef);
1205        }
1206        my $offset = -41; # from tail
1207        my $rl;
1208        open my $fh, '<', $db_path or croak "$db_path not readable: $!\n";
1209        sysseek($fh, $offset, 2); # don't care for errors
1210        sysread($fh, $rl, 41) == 41 or return (undef, undef);
1211        chomp $rl;
1212        while (('0' x40) eq $rl && sysseek($fh, 0, 1) != 0) {
1213                $offset -= 41;
1214                sysseek($fh, $offset, 2); # don't care for errors
1215                sysread($fh, $rl, 41) == 41 or return (undef, undef);
1216                chomp $rl;
1217        }
1218        if ($c && $c ne $rl) {
1219                die "$db_path and ", $self->refname,
1220                    " inconsistent!:\n$c != $rl\n";
1221        }
1222        my $rev = sysseek($fh, 0, 1) or croak $!;
1223        $rev =  ($rev - 41) / 41;
1224        close $fh or croak $!;
1225        ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
1226        return ($rev, $c);
1227}
1228
1229sub get_fetch_range {
1230        my ($self, $min, $max) = @_;
1231        $max ||= $self->ra->get_latest_revnum;
1232        $min ||= $self->rev_db_max;
1233        (++$min, $max);
1234}
1235
1236sub tmp_config {
1237        my (@args) = @_;
1238        my $config = "$ENV{GIT_DIR}/svn/config";
1239        unless (-f $config) {
1240                open my $fh, '>', $config or
1241                    die "Can't open $config: $!\n";
1242                print $fh "; This file is used internally by git-svn\n" or
1243                      die "Couldn't write to $config: $!\n";
1244                print $fh "; You should not have to edit it\n" or
1245                      die "Couldn't write to $config: $!\n";
1246                close $fh or die "Couldn't close $config: $!\n";
1247        }
1248        my $old_config = $ENV{GIT_CONFIG};
1249        $ENV{GIT_CONFIG} = $config;
1250        $@ = undef;
1251        my @ret = eval { command('config', @args) };
1252        my $err = $@;
1253        if (defined $old_config) {
1254                $ENV{GIT_CONFIG} = $old_config;
1255        } else {
1256                delete $ENV{GIT_CONFIG};
1257        }
1258        die $err if $err;
1259        wantarray ? @ret : $ret[0];
1260}
1261
1262sub tmp_index_do {
1263        my ($self, $sub) = @_;
1264        my $old_index = $ENV{GIT_INDEX_FILE};
1265        $ENV{GIT_INDEX_FILE} = $self->{index};
1266        $@ = undef;
1267        my @ret = eval { &$sub };
1268        my $err = $@;
1269        if (defined $old_index) {
1270                $ENV{GIT_INDEX_FILE} = $old_index;
1271        } else {
1272                delete $ENV{GIT_INDEX_FILE};
1273        }
1274        die $err if $err;
1275        wantarray ? @ret : $ret[0];
1276}
1277
1278sub assert_index_clean {
1279        my ($self, $treeish) = @_;
1280
1281        $self->tmp_index_do(sub {
1282                command_noisy('read-tree', $treeish) unless -e $self->{index};
1283                my $x = command_oneline('write-tree');
1284                my ($y) = (command(qw/cat-file commit/, $treeish) =~
1285                           /^tree ($::sha1)/mo);
1286                if ($y ne $x) {
1287                        unlink $self->{index} or croak $!;
1288                        command_noisy('read-tree', $treeish);
1289                }
1290                $x = command_oneline('write-tree');
1291                if ($y ne $x) {
1292                        ::fatal "trees ($treeish) $y != $x\n",
1293                                "Something is seriously wrong...\n";
1294                }
1295        });
1296}
1297
1298sub get_commit_parents {
1299        my ($self, $log_entry) = @_;
1300        my (%seen, @ret, @tmp);
1301        # legacy support for 'set-tree'; this is only used by set_tree_cb:
1302        if (my $ip = $self->{inject_parents}) {
1303                if (my $commit = delete $ip->{$log_entry->{revision}}) {
1304                        push @tmp, $commit;
1305                }
1306        }
1307        if (my $cur = ::verify_ref($self->refname.'^0')) {
1308                push @tmp, $cur;
1309        }
1310        push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
1311        while (my $p = shift @tmp) {
1312                next if $seen{$p};
1313                $seen{$p} = 1;
1314                push @ret, $p;
1315                # MAXPARENT is defined to 16 in commit-tree.c:
1316                last if @ret >= 16;
1317        }
1318        if (@tmp) {
1319                die "r$log_entry->{revision}: No room for parents:\n\t",
1320                    join("\n\t", @tmp), "\n";
1321        }
1322        @ret;
1323}
1324
1325sub full_url {
1326        my ($self) = @_;
1327        $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
1328}
1329
1330sub do_git_commit {
1331        my ($self, $log_entry) = @_;
1332        my $lr = $self->last_rev;
1333        if (defined $lr && $lr >= $log_entry->{revision}) {
1334                die "Last fetched revision of ", $self->refname,
1335                    " was r$lr, but we are about to fetch: ",
1336                    "r$log_entry->{revision}!\n";
1337        }
1338        if (my $c = $self->rev_db_get($log_entry->{revision})) {
1339                croak "$log_entry->{revision} = $c already exists! ",
1340                      "Why are we refetching it?\n";
1341        }
1342        $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} = $log_entry->{name};
1343        $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} =
1344                                                          $log_entry->{email};
1345        $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
1346
1347        my $tree = $log_entry->{tree};
1348        if (!defined $tree) {
1349                $tree = $self->tmp_index_do(sub {
1350                                            command_oneline('write-tree') });
1351        }
1352        die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
1353
1354        my @exec = ('git-commit-tree', $tree);
1355        foreach ($self->get_commit_parents($log_entry)) {
1356                push @exec, '-p', $_;
1357        }
1358        defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
1359                                                                   or croak $!;
1360        print $msg_fh $log_entry->{log} or croak $!;
1361        unless ($self->no_metadata) {
1362                print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
1363                              or croak $!;
1364        }
1365        $msg_fh->flush == 0 or croak $!;
1366        close $msg_fh or croak $!;
1367        chomp(my $commit = do { local $/; <$out_fh> });
1368        close $out_fh or croak $!;
1369        waitpid $pid, 0;
1370        croak $? if $?;
1371        if ($commit !~ /^$::sha1$/o) {
1372                die "Failed to commit, invalid sha1: $commit\n";
1373        }
1374
1375        $self->rev_db_set($log_entry->{revision}, $commit, 1);
1376
1377        $self->{last_rev} = $log_entry->{revision};
1378        $self->{last_commit} = $commit;
1379        print "r$log_entry->{revision}";
1380        if (defined $log_entry->{svm_revision}) {
1381                 print " (\@$log_entry->{svm_revision})";
1382                 $self->rev_db_set($log_entry->{svm_revision}, $commit,
1383                                   0, $self->svm_uuid);
1384        }
1385        print " = $commit ($self->{ref_id})\n";
1386        if (defined $_repack && (--$_repack_nr == 0)) {
1387                $_repack_nr = $_repack;
1388                # repack doesn't use any arguments with spaces in them, does it?
1389                print "Running git repack $_repack_flags ...\n";
1390                command_noisy('repack', split(/\s+/, $_repack_flags));
1391                print "Done repacking\n";
1392        }
1393        return $commit;
1394}
1395
1396sub match_paths {
1397        my ($self, $paths, $r) = @_;
1398        return 1 if $self->{path} eq '';
1399        if (my $path = $paths->{"/$self->{path}"}) {
1400                return ($path->{action} eq 'D') ? 0 : 1;
1401        }
1402        $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
1403        if (grep /$self->{path_regex}/, keys %$paths) {
1404                return 1;
1405        }
1406        my $c = '';
1407        foreach (split m#/#, $self->{path}) {
1408                $c .= "/$_";
1409                next unless ($paths->{$c} &&
1410                             ($paths->{$c}->{action} =~ /^[AR]$/));
1411                if ($self->ra->check_path($self->{path}, $r) ==
1412                    $SVN::Node::dir) {
1413                        return 1;
1414                }
1415        }
1416        return 0;
1417}
1418
1419sub find_parent_branch {
1420        my ($self, $paths, $rev) = @_;
1421        return undef unless $self->follow_parent;
1422        unless (defined $paths) {
1423                my $err_handler = $SVN::Error::handler;
1424                $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
1425                $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1, sub {
1426                                   $paths =
1427                                      Git::SVN::Ra::dup_changed_paths($_[0]) });
1428                $SVN::Error::handler = $err_handler;
1429        }
1430        return undef unless defined $paths;
1431
1432        # look for a parent from another branch:
1433        my @b_path_components = split m#/#, $self->rel_path;
1434        my @a_path_components;
1435        my $i;
1436        while (@b_path_components) {
1437                $i = $paths->{'/'.join('/', @b_path_components)};
1438                last if $i && defined $i->{copyfrom_path};
1439                unshift(@a_path_components, pop(@b_path_components));
1440        }
1441        return undef unless defined $i && defined $i->{copyfrom_path};
1442        my $branch_from = $i->{copyfrom_path};
1443        if (@a_path_components) {
1444                print STDERR "branch_from: $branch_from => ";
1445                $branch_from .= '/'.join('/', @a_path_components);
1446                print STDERR $branch_from, "\n";
1447        }
1448        my $r = $i->{copyfrom_rev};
1449        my $repos_root = $self->ra->{repos_root};
1450        my $url = $self->ra->{url};
1451        my $new_url = $repos_root . $branch_from;
1452        print STDERR  "Found possible branch point: ",
1453                      "$new_url => ", $self->full_url, ", $r\n";
1454        $branch_from =~ s#^/##;
1455        my $gs = Git::SVN->find_by_url($new_url, $repos_root, $branch_from);
1456        unless ($gs) {
1457                my $ref_id = $self->{ref_id};
1458                $ref_id =~ s/\@\d+$//;
1459                $ref_id .= "\@$r";
1460                # just grow a tail if we're not unique enough :x
1461                $ref_id .= '-' while find_ref($ref_id);
1462                print STDERR "Initializing parent: $ref_id\n";
1463                $gs = Git::SVN->init($new_url, '', $ref_id, $ref_id, 1);
1464        }
1465        my ($r0, $parent) = $gs->find_rev_before($r, 1);
1466        if (!defined $r0 || !defined $parent) {
1467                $gs->fetch(0, $r);
1468                ($r0, $parent) = $gs->last_rev_commit;
1469        }
1470        if (defined $r0 && defined $parent) {
1471                print STDERR "Found branch parent: ($self->{ref_id}) $parent\n";
1472                $self->assert_index_clean($parent);
1473                my $ed;
1474                if ($self->ra->can_do_switch) {
1475                        print STDERR "Following parent with do_switch\n";
1476                        # do_switch works with svn/trunk >= r22312, but that
1477                        # is not included with SVN 1.4.3 (the latest version
1478                        # at the moment), so we can't rely on it
1479                        $self->{last_commit} = $parent;
1480                        $ed = SVN::Git::Fetcher->new($self);
1481                        $gs->ra->gs_do_switch($r0, $rev, $gs,
1482                                              $self->full_url, $ed)
1483                          or die "SVN connection failed somewhere...\n";
1484                } else {
1485                        print STDERR "Following parent with do_update\n";
1486                        $ed = SVN::Git::Fetcher->new($self);
1487                        $self->ra->gs_do_update($rev, $rev, $self, $ed)
1488                          or die "SVN connection failed somewhere...\n";
1489                }
1490                print STDERR "Successfully followed parent\n";
1491                return $self->make_log_entry($rev, [$parent], $ed);
1492        }
1493        return undef;
1494}
1495
1496sub do_fetch {
1497        my ($self, $paths, $rev) = @_;
1498        my $ed;
1499        my ($last_rev, @parents);
1500        if (my $lc = $self->last_commit) {
1501                # we can have a branch that was deleted, then re-added
1502                # under the same name but copied from another path, in
1503                # which case we'll have multiple parents (we don't
1504                # want to break the original ref, nor lose copypath info):
1505                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
1506                        push @{$log_entry->{parents}}, $lc;
1507                        return $log_entry;
1508                }
1509                $ed = SVN::Git::Fetcher->new($self);
1510                $last_rev = $self->{last_rev};
1511                $ed->{c} = $lc;
1512                @parents = ($lc);
1513        } else {
1514                $last_rev = $rev;
1515                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
1516                        return $log_entry;
1517                }
1518                $ed = SVN::Git::Fetcher->new($self);
1519        }
1520        unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
1521                die "SVN connection failed somewhere...\n";
1522        }
1523        $self->make_log_entry($rev, \@parents, $ed);
1524}
1525
1526sub get_untracked {
1527        my ($self, $ed) = @_;
1528        my @out;
1529        my $h = $ed->{empty};
1530        foreach (sort keys %$h) {
1531                my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
1532                push @out, "  $act: " . uri_encode($_);
1533                warn "W: $act: $_\n";
1534        }
1535        foreach my $t (qw/dir_prop file_prop/) {
1536                $h = $ed->{$t} or next;
1537                foreach my $path (sort keys %$h) {
1538                        my $ppath = $path eq '' ? '.' : $path;
1539                        foreach my $prop (sort keys %{$h->{$path}}) {
1540                                next if $SKIP_PROP{$prop};
1541                                my $v = $h->{$path}->{$prop};
1542                                my $t_ppath_prop = "$t: " .
1543                                                    uri_encode($ppath) . ' ' .
1544                                                    uri_encode($prop);
1545                                if (defined $v) {
1546                                        push @out, "  +$t_ppath_prop " .
1547                                                   uri_encode($v);
1548                                } else {
1549                                        push @out, "  -$t_ppath_prop";
1550                                }
1551                        }
1552                }
1553        }
1554        foreach my $t (qw/absent_file absent_directory/) {
1555                $h = $ed->{$t} or next;
1556                foreach my $parent (sort keys %$h) {
1557                        foreach my $path (sort @{$h->{$parent}}) {
1558                                push @out, "  $t: " .
1559                                           uri_encode("$parent/$path");
1560                                warn "W: $t: $parent/$path ",
1561                                     "Insufficient permissions?\n";
1562                        }
1563                }
1564        }
1565        \@out;
1566}
1567
1568sub parse_svn_date {
1569        my $date = shift || return '+0000 1970-01-01 00:00:00';
1570        my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
1571                                            (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x) or
1572                                         croak "Unable to parse date: $date\n";
1573        "+0000 $Y-$m-$d $H:$M:$S";
1574}
1575
1576sub check_author {
1577        my ($author) = @_;
1578        if (!defined $author || length $author == 0) {
1579                $author = '(no author)';
1580        }
1581        if (defined $::_authors && ! defined $::users{$author}) {
1582                die "Author: $author not defined in $::_authors file\n";
1583        }
1584        $author;
1585}
1586
1587sub make_log_entry {
1588        my ($self, $rev, $parents, $ed) = @_;
1589        my $untracked = $self->get_untracked($ed);
1590
1591        open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
1592        print $un "r$rev\n" or croak $!;
1593        print $un $_, "\n" foreach @$untracked;
1594        my %log_entry = ( parents => $parents || [], revision => $rev,
1595                          log => '');
1596
1597        my $headrev;
1598        my $logged = delete $self->{logged_rev_props};
1599        if (!$logged || $self->{-want_revprops}) {
1600                my $rp = $self->ra->rev_proplist($rev);
1601                foreach (sort keys %$rp) {
1602                        my $v = $rp->{$_};
1603                        if (/^svn:(author|date|log)$/) {
1604                                $log_entry{$1} = $v;
1605                        } elsif ($_ eq 'svm:headrev') {
1606                                $headrev = $v;
1607                        } else {
1608                                print $un "  rev_prop: ", uri_encode($_), ' ',
1609                                          uri_encode($v), "\n";
1610                        }
1611                }
1612        } else {
1613                map { $log_entry{$_} = $logged->{$_} } keys %$logged;
1614        }
1615        close $un or croak $!;
1616
1617        $log_entry{date} = parse_svn_date($log_entry{date});
1618        $log_entry{log} .= "\n";
1619        my $author = $log_entry{author} = check_author($log_entry{author});
1620        my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
1621                                                       : ($author, undef);
1622        if (defined $headrev && $self->use_svm_props) {
1623                my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$};
1624                if ($uuid ne $self->{svm}->{uuid}) {
1625                        die "UUID mismatch on SVM path:\n",
1626                            "expected: $self->{svm}->{uuid}\n",
1627                            "     got: $uuid\n";
1628                }
1629                my $full_url = $self->{svm}->{source};
1630                $full_url .= "/$self->{path}" if length $self->{path};
1631                $log_entry{metadata} = "$full_url\@$r $uuid";
1632                $log_entry{svm_revision} = $r;
1633                $email ||= "$author\@$uuid"
1634        } else {
1635                $log_entry{metadata} = $self->full_url . "\@$rev " .
1636                                       $self->ra->get_uuid;
1637                $email ||= "$author\@" . $self->ra->get_uuid;
1638        }
1639        $log_entry{name} = $name;
1640        $log_entry{email} = $email;
1641        \%log_entry;
1642}
1643
1644sub fetch {
1645        my ($self, $min_rev, $max_rev, @parents) = @_;
1646        my ($last_rev, $last_commit) = $self->last_rev_commit;
1647        my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
1648        $self->ra->gs_fetch_loop_common($base, $head, [$self]);
1649}
1650
1651sub set_tree_cb {
1652        my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
1653        $self->{inject_parents} = { $rev => $tree };
1654        $self->fetch(undef, undef);
1655}
1656
1657sub set_tree {
1658        my ($self, $tree) = (shift, shift);
1659        my $log_entry = ::get_commit_entry($tree);
1660        unless ($self->{last_rev}) {
1661                fatal("Must have an existing revision to commit\n");
1662        }
1663        my %ed_opts = ( r => $self->{last_rev},
1664                        log => $log_entry->{log},
1665                        ra => $self->ra,
1666                        tree_a => $self->{last_commit},
1667                        tree_b => $tree,
1668                        editor_cb => sub {
1669                               $self->set_tree_cb($log_entry, $tree, @_) },
1670                        svn_path => $self->{path} );
1671        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
1672                print "No changes\nr$self->{last_rev} = $tree\n";
1673        }
1674}
1675
1676sub rebuild {
1677        my ($self) = @_;
1678        my $db_path = $self->db_path;
1679        if (-f $self->{db_root}) {
1680                rename $self->{db_root}, $db_path or die
1681                     "rename $self->{db_root} => $db_path failed: $!\n";
1682                my ($dir, $base) = ($db_path =~ m#^(.*?)/?([^/]+)$#);
1683                symlink $base, $self->{db_root} or die
1684                     "symlink $base => $self->{db_root} failed: $!\n";
1685                return;
1686        }
1687        print "Rebuilding $db_path ...\n";
1688        my ($rev_list, $ctx) = command_output_pipe("rev-list", $self->refname);
1689        my $latest;
1690        my $full_url = $self->full_url;
1691        my $svn_uuid;
1692        while (<$rev_list>) {
1693                chomp;
1694                my $c = $_;
1695                die "Non-SHA1: $c\n" unless $c =~ /^$::sha1$/o;
1696                my ($url, $rev, $uuid) = ::cmt_metadata($c);
1697
1698                # ignore merges (from set-tree)
1699                next if (!defined $rev || !$uuid);
1700
1701                # if we merged or otherwise started elsewhere, this is
1702                # how we break out of it
1703                if ((defined $svn_uuid && ($uuid ne $svn_uuid)) ||
1704                    ($full_url && $url && ($url ne $full_url))) {
1705                        next;
1706                }
1707                $latest ||= $rev;
1708                $svn_uuid ||= $uuid;
1709
1710                $self->rev_db_set($rev, $c);
1711                print "r$rev = $c\n";
1712        }
1713        command_close_pipe($rev_list, $ctx);
1714        print "Done rebuilding $db_path\n";
1715}
1716
1717# rev_db:
1718# Tie::File seems to be prone to offset errors if revisions get sparse,
1719# it's not that fast, either.  Tie::File is also not in Perl 5.6.  So
1720# one of my favorite modules is out :<  Next up would be one of the DBM
1721# modules, but I'm not sure which is most portable...  So I'll just
1722# go with something that's plain-text, but still capable of
1723# being randomly accessed.  So here's my ultra-simple fixed-width
1724# database.  All records are 40 characters + "\n", so it's easy to seek
1725# to a revision: (41 * rev) is the byte offset.
1726# A record of 40 0s denotes an empty revision.
1727# And yes, it's still pretty fast (faster than Tie::File).
1728# These files are disposable unless noMetadata or useSvmProps is set
1729
1730sub _rev_db_set {
1731        my ($fh, $rev, $commit) = @_;
1732        my $offset = $rev * 41;
1733        # assume that append is the common case:
1734        seek $fh, 0, 2 or croak $!;
1735        my $pos = tell $fh;
1736        if ($pos < $offset) {
1737                for (1 .. (($offset - $pos) / 41)) {
1738                        print $fh (('0' x 40),"\n") or croak $!;
1739                }
1740        }
1741        seek $fh, $offset, 0 or croak $!;
1742        print $fh $commit,"\n" or croak $!;
1743}
1744
1745sub mkfile {
1746        my ($path) = @_;
1747        unless (-e $path) {
1748                my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
1749                mkpath([$dir]) unless -d $dir;
1750                open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
1751                close $fh or die "Couldn't close (create) $path: $!\n";
1752        }
1753}
1754
1755sub rev_db_set {
1756        my ($self, $rev, $commit, $update_ref, $uuid) = @_;
1757        length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
1758        my $db = $self->db_path($uuid);
1759        my $db_lock = "$db.lock";
1760        my $sig;
1761        if ($update_ref) {
1762                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
1763                            $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
1764        }
1765        mkfile($db);
1766
1767        $LOCKFILES{$db_lock} = 1;
1768        my $sync;
1769        # both of these options make our .rev_db file very, very important
1770        # and we can't afford to lose it because rebuild() won't work
1771        if ($self->use_svm_props || $self->no_metadata) {
1772                $sync = 1;
1773                copy($db, $db_lock) or die "rev_db_set(@_): ",
1774                                           "Failed to copy: ",
1775                                           "$db => $db_lock ($!)\n";
1776        } else {
1777                rename $db, $db_lock or die "rev_db_set(@_): ",
1778                                            "Failed to rename: ",
1779                                            "$db => $db_lock ($!)\n";
1780        }
1781        open my $fh, '+<', $db_lock or die "Couldn't open $db_lock: $!\n";
1782        _rev_db_set($fh, $rev, $commit);
1783        if ($sync) {
1784                $fh->flush or die "Couldn't flush $db_lock: $!\n";
1785                $fh->sync or die "Couldn't sync $db_lock: $!\n";
1786        }
1787        close $fh or croak $!;
1788        if ($update_ref) {
1789                command_noisy('update-ref', '-m', "r$rev",
1790                              $self->refname, $commit);
1791        }
1792        rename $db_lock, $db or die "rev_db_set(@_): ", "Failed to rename: ",
1793                                    "$db_lock => $db ($!)\n";
1794        delete $LOCKFILES{$db_lock};
1795        if ($update_ref) {
1796                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
1797                            $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
1798                kill $sig, $$ if defined $sig;
1799        }
1800}
1801
1802sub rev_db_max {
1803        my ($self) = @_;
1804        my $db_path = $self->db_path;
1805        my @stat = stat $db_path or return 0;
1806        ($stat[7] % 41) == 0 or die "$db_path inconsistent size: $stat[7]\n";
1807        my $max = $stat[7] / 41;
1808        (($max > 0) ? $max - 1 : 0);
1809}
1810
1811sub rev_db_get {
1812        my ($self, $rev, $uuid) = @_;
1813        my $ret;
1814        my $offset = $rev * 41;
1815        my $db_path = $self->db_path($uuid);
1816        return undef unless -e $db_path;
1817        open my $fh, '<', $db_path or croak $!;
1818        if (sysseek($fh, $offset, 0) == $offset) {
1819                my $read = sysread($fh, $ret, 40);
1820                $ret = undef if ($read != 40 || $ret eq ('0'x40));
1821        }
1822        close $fh or croak $!;
1823        $ret;
1824}
1825
1826sub find_rev_before {
1827        my ($self, $rev, $eq_ok) = @_;
1828        --$rev unless $eq_ok;
1829        while ($rev > 0) {
1830                if (my $c = $self->rev_db_get($rev)) {
1831                        return ($rev, $c);
1832                }
1833                --$rev;
1834        }
1835        return (undef, undef);
1836}
1837
1838sub _new {
1839        my ($class, $repo_id, $ref_id, $path) = @_;
1840        unless (defined $repo_id && length $repo_id) {
1841                $repo_id = $Git::SVN::default_repo_id;
1842        }
1843        unless (defined $ref_id && length $ref_id) {
1844                $_[2] = $ref_id = $Git::SVN::default_ref_id;
1845        }
1846        $_[1] = $repo_id = sanitize_remote_name($repo_id);
1847        my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
1848        $_[3] = $path = '' unless (defined $path);
1849        mkpath([$dir]);
1850        bless {
1851                ref_id => $ref_id, dir => $dir, index => "$dir/index",
1852                path => $path, config => "$ENV{GIT_DIR}/svn/config",
1853                db_root => "$dir/.rev_db", repo_id => $repo_id }, $class;
1854}
1855
1856sub db_path {
1857        my ($self, $uuid) = @_;
1858        $uuid ||= $self->ra_uuid;
1859        "$self->{db_root}.$uuid";
1860}
1861
1862sub uri_encode {
1863        my ($f) = @_;
1864        $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
1865        $f
1866}
1867
1868package Git::SVN::Prompt;
1869use strict;
1870use warnings;
1871require SVN::Core;
1872use vars qw/$_no_auth_cache $_username/;
1873
1874sub simple {
1875        my ($cred, $realm, $default_username, $may_save, $pool) = @_;
1876        $may_save = undef if $_no_auth_cache;
1877        $default_username = $_username if defined $_username;
1878        if (defined $default_username && length $default_username) {
1879                if (defined $realm && length $realm) {
1880                        print STDERR "Authentication realm: $realm\n";
1881                        STDERR->flush;
1882                }
1883                $cred->username($default_username);
1884        } else {
1885                username($cred, $realm, $may_save, $pool);
1886        }
1887        $cred->password(_read_password("Password for '" .
1888                                       $cred->username . "': ", $realm));
1889        $cred->may_save($may_save);
1890        $SVN::_Core::SVN_NO_ERROR;
1891}
1892
1893sub ssl_server_trust {
1894        my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
1895        $may_save = undef if $_no_auth_cache;
1896        print STDERR "Error validating server certificate for '$realm':\n";
1897        if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
1898                print STDERR " - The certificate is not issued by a trusted ",
1899                      "authority. Use the\n",
1900                      "   fingerprint to validate the certificate manually!\n";
1901        }
1902        if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
1903                print STDERR " - The certificate hostname does not match.\n";
1904        }
1905        if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
1906                print STDERR " - The certificate is not yet valid.\n";
1907        }
1908        if ($failures & $SVN::Auth::SSL::EXPIRED) {
1909                print STDERR " - The certificate has expired.\n";
1910        }
1911        if ($failures & $SVN::Auth::SSL::OTHER) {
1912                print STDERR " - The certificate has an unknown error.\n";
1913        }
1914        printf STDERR
1915                "Certificate information:\n".
1916                " - Hostname: %s\n".
1917                " - Valid: from %s until %s\n".
1918                " - Issuer: %s\n".
1919                " - Fingerprint: %s\n",
1920                map $cert_info->$_, qw(hostname valid_from valid_until
1921                                       issuer_dname fingerprint);
1922        my $choice;
1923prompt:
1924        print STDERR $may_save ?
1925              "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
1926              "(R)eject or accept (t)emporarily? ";
1927        STDERR->flush;
1928        $choice = lc(substr(<STDIN> || 'R', 0, 1));
1929        if ($choice =~ /^t$/i) {
1930                $cred->may_save(undef);
1931        } elsif ($choice =~ /^r$/i) {
1932                return -1;
1933        } elsif ($may_save && $choice =~ /^p$/i) {
1934                $cred->may_save($may_save);
1935        } else {
1936                goto prompt;
1937        }
1938        $cred->accepted_failures($failures);
1939        $SVN::_Core::SVN_NO_ERROR;
1940}
1941
1942sub ssl_client_cert {
1943        my ($cred, $realm, $may_save, $pool) = @_;
1944        $may_save = undef if $_no_auth_cache;
1945        print STDERR "Client certificate filename: ";
1946        STDERR->flush;
1947        chomp(my $filename = <STDIN>);
1948        $cred->cert_file($filename);
1949        $cred->may_save($may_save);
1950        $SVN::_Core::SVN_NO_ERROR;
1951}
1952
1953sub ssl_client_cert_pw {
1954        my ($cred, $realm, $may_save, $pool) = @_;
1955        $may_save = undef if $_no_auth_cache;
1956        $cred->password(_read_password("Password: ", $realm));
1957        $cred->may_save($may_save);
1958        $SVN::_Core::SVN_NO_ERROR;
1959}
1960
1961sub username {
1962        my ($cred, $realm, $may_save, $pool) = @_;
1963        $may_save = undef if $_no_auth_cache;
1964        if (defined $realm && length $realm) {
1965                print STDERR "Authentication realm: $realm\n";
1966        }
1967        my $username;
1968        if (defined $_username) {
1969                $username = $_username;
1970        } else {
1971                print STDERR "Username: ";
1972                STDERR->flush;
1973                chomp($username = <STDIN>);
1974        }
1975        $cred->username($username);
1976        $cred->may_save($may_save);
1977        $SVN::_Core::SVN_NO_ERROR;
1978}
1979
1980sub _read_password {
1981        my ($prompt, $realm) = @_;
1982        print STDERR $prompt;
1983        STDERR->flush;
1984        require Term::ReadKey;
1985        Term::ReadKey::ReadMode('noecho');
1986        my $password = '';
1987        while (defined(my $key = Term::ReadKey::ReadKey(0))) {
1988                last if $key =~ /[\012\015]/; # \n\r
1989                $password .= $key;
1990        }
1991        Term::ReadKey::ReadMode('restore');
1992        print STDERR "\n";
1993        STDERR->flush;
1994        $password;
1995}
1996
1997package main;
1998
1999{
2000        my $kill_stupid_warnings = $SVN::Node::none.$SVN::Node::file.
2001                                $SVN::Node::dir.$SVN::Node::unknown.
2002                                $SVN::Node::none.$SVN::Node::file.
2003                                $SVN::Node::dir.$SVN::Node::unknown.
2004                                $SVN::Auth::SSL::CNMISMATCH.
2005                                $SVN::Auth::SSL::NOTYETVALID.
2006                                $SVN::Auth::SSL::EXPIRED.
2007                                $SVN::Auth::SSL::UNKNOWNCA.
2008                                $SVN::Auth::SSL::OTHER;
2009}
2010
2011package SVN::Git::Fetcher;
2012use vars qw/@ISA/;
2013use strict;
2014use warnings;
2015use Carp qw/croak/;
2016use IO::File qw//;
2017use Digest::MD5;
2018
2019# file baton members: path, mode_a, mode_b, pool, fh, blob, base
2020sub new {
2021        my ($class, $git_svn) = @_;
2022        my $self = SVN::Delta::Editor->new;
2023        bless $self, $class;
2024        $self->{c} = $git_svn->{last_commit} if exists $git_svn->{last_commit};
2025        $self->{empty} = {};
2026        $self->{dir_prop} = {};
2027        $self->{file_prop} = {};
2028        $self->{absent_dir} = {};
2029        $self->{absent_file} = {};
2030        $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
2031        $self;
2032}
2033
2034sub set_path_strip {
2035        my ($self, $path) = @_;
2036        $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
2037}
2038
2039sub open_root {
2040        { path => '' };
2041}
2042
2043sub open_directory {
2044        my ($self, $path, $pb, $rev) = @_;
2045        { path => $path };
2046}
2047
2048sub git_path {
2049        my ($self, $path) = @_;
2050        if ($self->{path_strip}) {
2051                $path =~ s!$self->{path_strip}!! or
2052                  die "Failed to strip path '$path' ($self->{path_strip})\n";
2053        }
2054        $path;
2055}
2056
2057sub delete_entry {
2058        my ($self, $path, $rev, $pb) = @_;
2059
2060        my $gpath = $self->git_path($path);
2061        return undef if ($gpath eq '');
2062
2063        # remove entire directories.
2064        if (command('ls-tree', $self->{c}, '--', $gpath) =~ /^040000 tree/) {
2065                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
2066                                                     -r --name-only -z/,
2067                                                     $self->{c}, '--', $gpath);
2068                local $/ = "\0";
2069                while (<$ls>) {
2070                        chomp;
2071                        $self->{gii}->remove($_);
2072                        print "\tD\t$_\n" unless $::_q;
2073                }
2074                print "\tD\t$gpath/\n" unless $::_q;
2075                command_close_pipe($ls, $ctx);
2076                $self->{empty}->{$path} = 0
2077        } else {
2078                $self->{gii}->remove($gpath);
2079                print "\tD\t$gpath\n" unless $::_q;
2080        }
2081        undef;
2082}
2083
2084sub open_file {
2085        my ($self, $path, $pb, $rev) = @_;
2086        my $gpath = $self->git_path($path);
2087        my ($mode, $blob) = (command('ls-tree', $self->{c}, '--', $gpath)
2088                             =~ /^(\d{6}) blob ([a-f\d]{40})\t/);
2089        unless (defined $mode && defined $blob) {
2090                die "$path was not found in commit $self->{c} (r$rev)\n";
2091        }
2092        { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
2093          pool => SVN::Pool->new, action => 'M' };
2094}
2095
2096sub add_file {
2097        my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
2098        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
2099        delete $self->{empty}->{$dir};
2100        { path => $path, mode_a => 100644, mode_b => 100644,
2101          pool => SVN::Pool->new, action => 'A' };
2102}
2103
2104sub add_directory {
2105        my ($self, $path, $cp_path, $cp_rev) = @_;
2106        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
2107        delete $self->{empty}->{$dir};
2108        $self->{empty}->{$path} = 1;
2109        { path => $path };
2110}
2111
2112sub change_dir_prop {
2113        my ($self, $db, $prop, $value) = @_;
2114        $self->{dir_prop}->{$db->{path}} ||= {};
2115        $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
2116        undef;
2117}
2118
2119sub absent_directory {
2120        my ($self, $path, $pb) = @_;
2121        $self->{absent_dir}->{$pb->{path}} ||= [];
2122        push @{$self->{absent_dir}->{$pb->{path}}}, $path;
2123        undef;
2124}
2125
2126sub absent_file {
2127        my ($self, $path, $pb) = @_;
2128        $self->{absent_file}->{$pb->{path}} ||= [];
2129        push @{$self->{absent_file}->{$pb->{path}}}, $path;
2130        undef;
2131}
2132
2133sub change_file_prop {
2134        my ($self, $fb, $prop, $value) = @_;
2135        if ($prop eq 'svn:executable') {
2136                if ($fb->{mode_b} != 120000) {
2137                        $fb->{mode_b} = defined $value ? 100755 : 100644;
2138                }
2139        } elsif ($prop eq 'svn:special') {
2140                $fb->{mode_b} = defined $value ? 120000 : 100644;
2141        } else {
2142                $self->{file_prop}->{$fb->{path}} ||= {};
2143                $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
2144        }
2145        undef;
2146}
2147
2148sub apply_textdelta {
2149        my ($self, $fb, $exp) = @_;
2150        my $fh = IO::File->new_tmpfile;
2151        $fh->autoflush(1);
2152        # $fh gets auto-closed() by SVN::TxDelta::apply(),
2153        # (but $base does not,) so dup() it for reading in close_file
2154        open my $dup, '<&', $fh or croak $!;
2155        my $base = IO::File->new_tmpfile;
2156        $base->autoflush(1);
2157        if ($fb->{blob}) {
2158                defined (my $pid = fork) or croak $!;
2159                if (!$pid) {
2160                        open STDOUT, '>&', $base or croak $!;
2161                        print STDOUT 'link ' if ($fb->{mode_a} == 120000);
2162                        exec qw/git-cat-file blob/, $fb->{blob} or croak $!;
2163                }
2164                waitpid $pid, 0;
2165                croak $? if $?;
2166
2167                if (defined $exp) {
2168                        seek $base, 0, 0 or croak $!;
2169                        my $md5 = Digest::MD5->new;
2170                        $md5->addfile($base);
2171                        my $got = $md5->hexdigest;
2172                        die "Checksum mismatch: $fb->{path} $fb->{blob}\n",
2173                            "expected: $exp\n",
2174                            "     got: $got\n" if ($got ne $exp);
2175                }
2176        }
2177        seek $base, 0, 0 or croak $!;
2178        $fb->{fh} = $dup;
2179        $fb->{base} = $base;
2180        [ SVN::TxDelta::apply($base, $fh, undef, $fb->{path}, $fb->{pool}) ];
2181}
2182
2183sub close_file {
2184        my ($self, $fb, $exp) = @_;
2185        my $hash;
2186        my $path = $self->git_path($fb->{path});
2187        if (my $fh = $fb->{fh}) {
2188                seek($fh, 0, 0) or croak $!;
2189                my $md5 = Digest::MD5->new;
2190                $md5->addfile($fh);
2191                my $got = $md5->hexdigest;
2192                die "Checksum mismatch: $path\n",
2193                    "expected: $exp\n    got: $got\n" if ($got ne $exp);
2194                seek($fh, 0, 0) or croak $!;
2195                if ($fb->{mode_b} == 120000) {
2196                        read($fh, my $buf, 5) == 5 or croak $!;
2197                        $buf eq 'link ' or die "$path has mode 120000",
2198                                               "but is not a link\n";
2199                }
2200                defined(my $pid = open my $out,'-|') or die "Can't fork: $!\n";
2201                if (!$pid) {
2202                        open STDIN, '<&', $fh or croak $!;
2203                        exec qw/git-hash-object -w --stdin/ or croak $!;
2204                }
2205                chomp($hash = do { local $/; <$out> });
2206                close $out or croak $!;
2207                close $fh or croak $!;
2208                $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
2209                close $fb->{base} or croak $!;
2210        } else {
2211                $hash = $fb->{blob} or die "no blob information\n";
2212        }
2213        $fb->{pool}->clear;
2214        $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
2215        print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
2216        undef;
2217}
2218
2219sub abort_edit {
2220        my $self = shift;
2221        $self->{nr} = $self->{gii}->{nr};
2222        delete $self->{gii};
2223        $self->SUPER::abort_edit(@_);
2224}
2225
2226sub close_edit {
2227        my $self = shift;
2228        $self->{git_commit_ok} = 1;
2229        $self->{nr} = $self->{gii}->{nr};
2230        delete $self->{gii};
2231        $self->SUPER::close_edit(@_);
2232}
2233
2234package SVN::Git::Editor;
2235use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
2236use strict;
2237use warnings;
2238use Carp qw/croak/;
2239use IO::File;
2240use Digest::MD5;
2241
2242sub new {
2243        my ($class, $opts) = @_;
2244        foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
2245                die "$_ required!\n" unless (defined $opts->{$_});
2246        }
2247
2248        my $pool = SVN::Pool->new;
2249        my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
2250        my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
2251                                     $opts->{r}, $mods);
2252
2253        # $opts->{ra} functions should not be used after this:
2254        my @ce  = $opts->{ra}->get_commit_editor($opts->{log},
2255                                                $opts->{editor_cb}, $pool);
2256        my $self = SVN::Delta::Editor->new(@ce, $pool);
2257        bless $self, $class;
2258        foreach (qw/svn_path r tree_a tree_b/) {
2259                $self->{$_} = $opts->{$_};
2260        }
2261        $self->{url} = $opts->{ra}->{url};
2262        $self->{mods} = $mods;
2263        $self->{types} = $types;
2264        $self->{pool} = $pool;
2265        $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
2266        $self->{rm} = { };
2267        $self->{path_prefix} = length $self->{svn_path} ?
2268                               "$self->{svn_path}/" : '';
2269        return $self;
2270}
2271
2272sub generate_diff {
2273        my ($tree_a, $tree_b) = @_;
2274        my @diff_tree = qw(diff-tree -z -r);
2275        if ($_cp_similarity) {
2276                push @diff_tree, "-C$_cp_similarity";
2277        } else {
2278                push @diff_tree, '-C';
2279        }
2280        push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
2281        push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
2282        push @diff_tree, $tree_a, $tree_b;
2283        my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
2284        local $/ = "\0";
2285        my $state = 'meta';
2286        my @mods;
2287        while (<$diff_fh>) {
2288                chomp $_; # this gets rid of the trailing "\0"
2289                if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
2290                                        $::sha1\s($::sha1)\s
2291                                        ([MTCRAD])\d*$/xo) {
2292                        push @mods, {   mode_a => $1, mode_b => $2,
2293                                        sha1_b => $3, chg => $4 };
2294                        if ($4 =~ /^(?:C|R)$/) {
2295                                $state = 'file_a';
2296                        } else {
2297                                $state = 'file_b';
2298                        }
2299                } elsif ($state eq 'file_a') {
2300                        my $x = $mods[$#mods] or croak "Empty array\n";
2301                        if ($x->{chg} !~ /^(?:C|R)$/) {
2302                                croak "Error parsing $_, $x->{chg}\n";
2303                        }
2304                        $x->{file_a} = $_;
2305                        $state = 'file_b';
2306                } elsif ($state eq 'file_b') {
2307                        my $x = $mods[$#mods] or croak "Empty array\n";
2308                        if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
2309                                croak "Error parsing $_, $x->{chg}\n";
2310                        }
2311                        if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
2312                                croak "Error parsing $_, $x->{chg}\n";
2313                        }
2314                        $x->{file_b} = $_;
2315                        $state = 'meta';
2316                } else {
2317                        croak "Error parsing $_\n";
2318                }
2319        }
2320        command_close_pipe($diff_fh, $ctx);
2321        \@mods;
2322}
2323
2324sub check_diff_paths {
2325        my ($ra, $pfx, $rev, $mods) = @_;
2326        my %types;
2327        $pfx .= '/' if length $pfx;
2328
2329        sub type_diff_paths {
2330                my ($ra, $types, $path, $rev) = @_;
2331                my @p = split m#/+#, $path;
2332                my $c = shift @p;
2333                unless (defined $types->{$c}) {
2334                        $types->{$c} = $ra->check_path($c, $rev);
2335                }
2336                while (@p) {
2337                        $c .= '/' . shift @p;
2338                        next if defined $types->{$c};
2339                        $types->{$c} = $ra->check_path($c, $rev);
2340                }
2341        }
2342
2343        foreach my $m (@$mods) {
2344                foreach my $f (qw/file_a file_b/) {
2345                        next unless defined $m->{$f};
2346                        my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
2347                        if (length $pfx.$dir && ! defined $types{$dir}) {
2348                                type_diff_paths($ra, \%types, $pfx.$dir, $rev);
2349                        }
2350                }
2351        }
2352        \%types;
2353}
2354
2355sub split_path {
2356        return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
2357}
2358
2359sub repo_path {
2360        my ($self, $path) = @_;
2361        $self->{path_prefix}.(defined $path ? $path : '');
2362}
2363
2364sub url_path {
2365        my ($self, $path) = @_;
2366        $self->{url} . '/' . $self->repo_path($path);
2367}
2368
2369sub rmdirs {
2370        my ($self) = @_;
2371        my $rm = $self->{rm};
2372        delete $rm->{''}; # we never delete the url we're tracking
2373        return unless %$rm;
2374
2375        foreach (keys %$rm) {
2376                my @d = split m#/#, $_;
2377                my $c = shift @d;
2378                $rm->{$c} = 1;
2379                while (@d) {
2380                        $c .= '/' . shift @d;
2381                        $rm->{$c} = 1;
2382                }
2383        }
2384        delete $rm->{$self->{svn_path}};
2385        delete $rm->{''}; # we never delete the url we're tracking
2386        return unless %$rm;
2387
2388        my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
2389                                             $self->{tree_b});
2390        local $/ = "\0";
2391        while (<$fh>) {
2392                chomp;
2393                my @dn = split m#/#, $_;
2394                while (pop @dn) {
2395                        delete $rm->{join '/', @dn};
2396                }
2397                unless (%$rm) {
2398                        close $fh;
2399                        return;
2400                }
2401        }
2402        command_close_pipe($fh, $ctx);
2403
2404        my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
2405        foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
2406                $self->close_directory($bat->{$d}, $p);
2407                my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
2408                print "\tD+\t$d/\n" unless $::_q;
2409                $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
2410                delete $bat->{$d};
2411        }
2412}
2413
2414sub open_or_add_dir {
2415        my ($self, $full_path, $baton) = @_;
2416        my $t = $self->{types}->{$full_path};
2417        if (!defined $t) {
2418                die "$full_path not known in r$self->{r} or we have a bug!\n";
2419        }
2420        if ($t == $SVN::Node::none) {
2421                return $self->add_directory($full_path, $baton,
2422                                                undef, -1, $self->{pool});
2423        } elsif ($t == $SVN::Node::dir) {
2424                return $self->open_directory($full_path, $baton,
2425                                                $self->{r}, $self->{pool});
2426        }
2427        print STDERR "$full_path already exists in repository at ",
2428                "r$self->{r} and it is not a directory (",
2429                ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
2430        exit 1;
2431}
2432
2433sub ensure_path {
2434        my ($self, $path) = @_;
2435        my $bat = $self->{bat};
2436        my $repo_path = $self->repo_path($path);
2437        return $bat->{''} unless (length $repo_path);
2438        my @p = split m#/+#, $repo_path;
2439        my $c = shift @p;
2440        $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
2441        while (@p) {
2442                my $c0 = $c;
2443                $c .= '/' . shift @p;
2444                $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
2445        }
2446        return $bat->{$c};
2447}
2448
2449sub A {
2450        my ($self, $m) = @_;
2451        my ($dir, $file) = split_path($m->{file_b});
2452        my $pbat = $self->ensure_path($dir);
2453        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2454                                        undef, -1);
2455        print "\tA\t$m->{file_b}\n" unless $::_q;
2456        $self->chg_file($fbat, $m);
2457        $self->close_file($fbat,undef,$self->{pool});
2458}
2459
2460sub C {
2461        my ($self, $m) = @_;
2462        my ($dir, $file) = split_path($m->{file_b});
2463        my $pbat = $self->ensure_path($dir);
2464        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2465                                $self->url_path($m->{file_a}), $self->{r});
2466        print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
2467        $self->chg_file($fbat, $m);
2468        $self->close_file($fbat,undef,$self->{pool});
2469}
2470
2471sub delete_entry {
2472        my ($self, $path, $pbat) = @_;
2473        my $rpath = $self->repo_path($path);
2474        my ($dir, $file) = split_path($rpath);
2475        $self->{rm}->{$dir} = 1;
2476        $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
2477}
2478
2479sub R {
2480        my ($self, $m) = @_;
2481        my ($dir, $file) = split_path($m->{file_b});
2482        my $pbat = $self->ensure_path($dir);
2483        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2484                                $self->url_path($m->{file_a}), $self->{r});
2485        print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
2486        $self->chg_file($fbat, $m);
2487        $self->close_file($fbat,undef,$self->{pool});
2488
2489        ($dir, $file) = split_path($m->{file_a});
2490        $pbat = $self->ensure_path($dir);
2491        $self->delete_entry($m->{file_a}, $pbat);
2492}
2493
2494sub M {
2495        my ($self, $m) = @_;
2496        my ($dir, $file) = split_path($m->{file_b});
2497        my $pbat = $self->ensure_path($dir);
2498        my $fbat = $self->open_file($self->repo_path($m->{file_b}),
2499                                $pbat,$self->{r},$self->{pool});
2500        print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
2501        $self->chg_file($fbat, $m);
2502        $self->close_file($fbat,undef,$self->{pool});
2503}
2504
2505sub T { shift->M(@_) }
2506
2507sub change_file_prop {
2508        my ($self, $fbat, $pname, $pval) = @_;
2509        $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
2510}
2511
2512sub chg_file {
2513        my ($self, $fbat, $m) = @_;
2514        if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
2515                $self->change_file_prop($fbat,'svn:executable','*');
2516        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
2517                $self->change_file_prop($fbat,'svn:executable',undef);
2518        }
2519        my $fh = IO::File->new_tmpfile or croak $!;
2520        if ($m->{mode_b} =~ /^120/) {
2521                print $fh 'link ' or croak $!;
2522                $self->change_file_prop($fbat,'svn:special','*');
2523        } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
2524                $self->change_file_prop($fbat,'svn:special',undef);
2525        }
2526        defined(my $pid = fork) or croak $!;
2527        if (!$pid) {
2528                open STDOUT, '>&', $fh or croak $!;
2529                exec qw/git-cat-file blob/, $m->{sha1_b} or croak $!;
2530        }
2531        waitpid $pid, 0;
2532        croak $? if $?;
2533        $fh->flush == 0 or croak $!;
2534        seek $fh, 0, 0 or croak $!;
2535
2536        my $md5 = Digest::MD5->new;
2537        $md5->addfile($fh) or croak $!;
2538        seek $fh, 0, 0 or croak $!;
2539
2540        my $exp = $md5->hexdigest;
2541        my $pool = SVN::Pool->new;
2542        my $atd = $self->apply_textdelta($fbat, undef, $pool);
2543        my $got = SVN::TxDelta::send_stream($fh, @$atd, $pool);
2544        die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
2545        $pool->clear;
2546
2547        close $fh or croak $!;
2548}
2549
2550sub D {
2551        my ($self, $m) = @_;
2552        my ($dir, $file) = split_path($m->{file_b});
2553        my $pbat = $self->ensure_path($dir);
2554        print "\tD\t$m->{file_b}\n" unless $::_q;
2555        $self->delete_entry($m->{file_b}, $pbat);
2556}
2557
2558sub close_edit {
2559        my ($self) = @_;
2560        my ($p,$bat) = ($self->{pool}, $self->{bat});
2561        foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
2562                $self->close_directory($bat->{$_}, $p);
2563        }
2564        $self->SUPER::close_edit($p);
2565        $p->clear;
2566}
2567
2568sub abort_edit {
2569        my ($self) = @_;
2570        $self->SUPER::abort_edit($self->{pool});
2571}
2572
2573sub DESTROY {
2574        my $self = shift;
2575        $self->SUPER::DESTROY(@_);
2576        $self->{pool}->clear;
2577}
2578
2579# this drives the editor
2580sub apply_diff {
2581        my ($self) = @_;
2582        my $mods = $self->{mods};
2583        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
2584        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
2585                my $f = $m->{chg};
2586                if (defined $o{$f}) {
2587                        $self->$f($m);
2588                } else {
2589                        fatal("Invalid change type: $f\n");
2590                }
2591        }
2592        $self->rmdirs if $_rmdir;
2593        if (@$mods == 0) {
2594                $self->abort_edit;
2595        } else {
2596                $self->close_edit;
2597        }
2598        return scalar @$mods;
2599}
2600
2601package Git::SVN::Ra;
2602use vars qw/@ISA $config_dir/;
2603use strict;
2604use warnings;
2605my ($can_do_switch);
2606my $RA;
2607
2608BEGIN {
2609        # enforce temporary pool usage for some simple functions
2610        my $e;
2611        foreach (qw/get_latest_revnum get_uuid get_repos_root/) {
2612                $e .= "sub $_ {
2613                        my \$self = shift;
2614                        my \$pool = SVN::Pool->new;
2615                        my \@ret = \$self->SUPER::$_(\@_,\$pool);
2616                        \$pool->clear;
2617                        wantarray ? \@ret : \$ret[0]; }\n";
2618        }
2619
2620        # get_dir needs $pool held in cache for dirents to work,
2621        # check_path is cacheable and rev_proplist is close enough
2622        # for our purposes.
2623        foreach (qw/check_path get_dir rev_proplist/) {
2624                $e .= "my \%${_}_cache; my \$${_}_rev = 0; sub $_ {
2625                        my \$self = shift;
2626                        my \$r = pop;
2627                        my \$k = join(\"\\0\", \@_);
2628                        if (my \$x = \$${_}_cache{\$r}->{\$k}) {
2629                                return wantarray ? \@\$x : \$x->[0];
2630                        }
2631                        my \$pool = SVN::Pool->new;
2632                        my \@ret = \$self->SUPER::$_(\@_, \$r, \$pool);
2633                        if (\$r != \$${_}_rev) {
2634                                \%${_}_cache = ( pool => [] );
2635                                \$${_}_rev = \$r;
2636                        }
2637                        \$${_}_cache{\$r}->{\$k} = \\\@ret;
2638                        push \@{\$${_}_cache{pool}}, \$pool;
2639                        wantarray ? \@ret : \$ret[0]; }\n";
2640        }
2641        $e .= "\n1;";
2642        eval $e or die $@;
2643}
2644
2645sub new {
2646        my ($class, $url) = @_;
2647        $url =~ s!/+$!!;
2648        return $RA if ($RA && $RA->{url} eq $url);
2649
2650        SVN::_Core::svn_config_ensure($config_dir, undef);
2651        my ($baton, $callbacks) = SVN::Core::auth_open_helper([
2652            SVN::Client::get_simple_provider(),
2653            SVN::Client::get_ssl_server_trust_file_provider(),
2654            SVN::Client::get_simple_prompt_provider(
2655              \&Git::SVN::Prompt::simple, 2),
2656            SVN::Client::get_ssl_client_cert_prompt_provider(
2657              \&Git::SVN::Prompt::ssl_client_cert, 2),
2658            SVN::Client::get_ssl_client_cert_pw_prompt_provider(
2659              \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
2660            SVN::Client::get_username_provider(),
2661            SVN::Client::get_ssl_server_trust_prompt_provider(
2662              \&Git::SVN::Prompt::ssl_server_trust),
2663            SVN::Client::get_username_prompt_provider(
2664              \&Git::SVN::Prompt::username, 2),
2665          ]);
2666        my $config = SVN::Core::config_get_config($config_dir);
2667        my $self = SVN::Ra->new(url => $url, auth => $baton,
2668                              config => $config,
2669                              pool => SVN::Pool->new,
2670                              auth_provider_callbacks => $callbacks);
2671        $self->{svn_path} = $url;
2672        $self->{repos_root} = $self->get_repos_root;
2673        $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
2674        $RA = bless $self, $class;
2675}
2676
2677sub DESTROY {
2678        # do not call the real DESTROY since we store ourselves in $RA
2679}
2680
2681sub get_log {
2682        my ($self, @args) = @_;
2683        my $pool = SVN::Pool->new;
2684        splice(@args, 3, 1) if ($SVN::Core::VERSION le '1.2.0');
2685        my $ret = $self->SUPER::get_log(@args, $pool);
2686        $pool->clear;
2687        $ret;
2688}
2689
2690sub get_commit_editor {
2691        my ($self, $log, $cb, $pool) = @_;
2692        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
2693        $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
2694}
2695
2696sub gs_do_update {
2697        my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
2698        my $new = ($rev_a == $rev_b);
2699        my $path = $gs->{path};
2700
2701        my $pool = SVN::Pool->new;
2702        $editor->set_path_strip($path);
2703        my (@pc) = split m#/#, $path;
2704        my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
2705                                        1, $editor, $pool);
2706        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
2707
2708        # Since we can't rely on svn_ra_reparent being available, we'll
2709        # just have to do some magic with set_path to make it so
2710        # we only want a partial path.
2711        my $sp = '';
2712        my $final = join('/', @pc);
2713        while (@pc) {
2714                $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
2715                $sp .= '/' if length $sp;
2716                $sp .= shift @pc;
2717        }
2718        die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
2719
2720        $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
2721
2722        $reporter->finish_report($pool);
2723        $pool->clear;
2724        $editor->{git_commit_ok};
2725}
2726
2727# this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
2728# svn_ra_reparent didn't work before 1.4)
2729sub gs_do_switch {
2730        my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
2731        my $path = $gs->{path};
2732        my $pool = SVN::Pool->new;
2733
2734        my $full_url = $self->{url};
2735        my $old_url = $full_url;
2736        $full_url .= "/$path" if length $path;
2737        my ($ra, $reparented);
2738        if ($old_url ne $full_url) {
2739                if ($old_url !~ m#^svn(\+ssh)?://#) {
2740                        SVN::_Ra::svn_ra_reparent($self->{session}, $full_url,
2741                                                  $pool);
2742                        $self->{url} = $full_url;
2743                        $reparented = 1;
2744                } else {
2745                        $ra = Git::SVN::Ra->new($full_url);
2746                }
2747        }
2748        $ra ||= $self;
2749        my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
2750        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
2751        $reporter->set_path('', $rev_a, 0, @lock, $pool);
2752        $reporter->finish_report($pool);
2753
2754        if ($reparented) {
2755                SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
2756                $self->{url} = $old_url;
2757        }
2758
2759        $pool->clear;
2760        $editor->{git_commit_ok};
2761}
2762
2763sub gs_fetch_loop_common {
2764        my ($self, $base, $head, $gsv, $globs) = @_;
2765        return if ($base > $head);
2766        my $inc = 1000;
2767        my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
2768        my %common;
2769        my $common_max = scalar @$gsv;
2770
2771        foreach my $gs (@$gsv) {
2772                if (my $last_commit = $gs->last_commit) {
2773                        $gs->assert_index_clean($last_commit);
2774                }
2775                my @tmp = split m#/#, $gs->{path};
2776                my $p = '';
2777                foreach (@tmp) {
2778                        $p .= length($p) ? "/$_" : $_;
2779                        $common{$p} ||= 0;
2780                        $common{$p}++;
2781                }
2782        }
2783        $globs ||= [];
2784        $common_max += scalar @$globs;
2785        foreach my $glob (@$globs) {
2786                my @tmp = split m#/#, $glob->{path}->{left};
2787                my $p = '';
2788                foreach (@tmp) {
2789                        $p .= length($p) ? "/$_" : $_;
2790                        $common{$p} ||= 0;
2791                        $common{$p}++;
2792                }
2793        }
2794
2795        my $longest_path = '';
2796        foreach (sort {length $b <=> length $a} keys %common) {
2797                if ($common{$_} == $common_max) {
2798                        $longest_path = $_;
2799                        last;
2800                }
2801        }
2802        while (1) {
2803                my %revs;
2804                my $err;
2805                my $err_handler = $SVN::Error::handler;
2806                $SVN::Error::handler = sub {
2807                        ($err) = @_;
2808                        skip_unknown_revs($err);
2809                };
2810                sub _cb {
2811                        my ($paths, $r, $author, $date, $log) = @_;
2812                        [ dup_changed_paths($paths),
2813                          { author => $author, date => $date, log => $log } ];
2814                }
2815                $self->get_log([$longest_path], $min, $max, 0, 1, 1,
2816                               sub { $revs{$_[1]} = _cb(@_) });
2817                if ($err && $max >= $head) {
2818                        print STDERR "Path '$longest_path' ",
2819                                     "was probably deleted:\n",
2820                                     $err->expanded_message,
2821                                     "\nWill attempt to follow ",
2822                                     "revisions r$min .. r$max ",
2823                                     "committed before the deletion\n";
2824                        my $hi = $max;
2825                        while (--$hi >= $min) {
2826                                my $ok;
2827                                $self->get_log([$longest_path], $min, $hi,
2828                                               0, 1, 1, sub {
2829                                               $ok ||= $_[1];
2830                                               $revs{$_[1]} = _cb(@_) });
2831                                if ($ok) {
2832                                        print STDERR "r$min .. r$ok OK\n";
2833                                        last;
2834                                }
2835                        }
2836                }
2837                $SVN::Error::handler = $err_handler;
2838
2839                my %exists = map { $_->{path} => $_ } @$gsv;
2840                foreach my $r (sort {$a <=> $b} keys %revs) {
2841                        my ($paths, $logged) = @{$revs{$r}};
2842
2843                        foreach my $gs ($self->match_globs(\%exists, $paths,
2844                                                           $globs, $r)) {
2845                                if ($gs->rev_db_max >= $r) {
2846                                        next;
2847                                }
2848                                next unless $gs->match_paths($paths, $r);
2849                                $gs->{logged_rev_props} = $logged;
2850                                my $log_entry = $gs->do_fetch($paths, $r);
2851                                if ($log_entry) {
2852                                        $gs->do_git_commit($log_entry);
2853                                }
2854                        }
2855                        foreach my $g (@$globs) {
2856                                my $k = "svn-remote.$g->{remote}." .
2857                                        "$g->{t}-maxRev";
2858                                Git::SVN::tmp_config($k, $r);
2859                        }
2860                }
2861                # pre-fill the .rev_db since it'll eventually get filled in
2862                # with '0' x40 if something new gets committed
2863                foreach my $gs (@$gsv) {
2864                        next if defined $gs->rev_db_get($max);
2865                        $gs->rev_db_set($max, 0 x40);
2866                }
2867                foreach my $g (@$globs) {
2868                        my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
2869                        Git::SVN::tmp_config($k, $max);
2870                }
2871                last if $max >= $head;
2872                $min = $max + 1;
2873                $max += $inc;
2874                $max = $head if ($max > $head);
2875        }
2876}
2877
2878sub match_globs {
2879        my ($self, $exists, $paths, $globs, $r) = @_;
2880
2881        sub get_dir_check {
2882                my ($self, $exists, $g, $r) = @_;
2883                my @x = eval { $self->get_dir($g->{path}->{left}, $r) };
2884                return unless scalar @x == 3;
2885                my $dirents = $x[0];
2886                foreach my $de (keys %$dirents) {
2887                        next if $dirents->{$de}->kind != $SVN::Node::dir;
2888                        my $p = $g->{path}->full_path($de);
2889                        next if $exists->{$p};
2890                        next if (length $g->{path}->{right} &&
2891                                 ($self->check_path($p, $r) !=
2892                                  $SVN::Node::dir));
2893                        $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
2894                                         $g->{ref}->full_path($de), 1);
2895                }
2896        }
2897        foreach my $g (@$globs) {
2898                if (my $path = $paths->{"/$g->{path}->{left}"}) {
2899                        if ($path->{action} =~ /^[AR]$/) {
2900                                get_dir_check($self, $exists, $g, $r);
2901                        }
2902                }
2903                foreach (keys %$paths) {
2904                        if (/$g->{path}->{left_regex}/) {
2905                                next if $paths->{$_}->{action} !~ /^[AR]$/;
2906                                get_dir_check($self, $exists, $g, $r);
2907                        }
2908                        next unless /$g->{path}->{regex}/;
2909                        my $p = $1;
2910                        my $pathname = $g->{path}->full_path($p);
2911                        next if $exists->{$pathname};
2912                        $exists->{$pathname} = Git::SVN->init(
2913                                              $self->{url}, $pathname, undef,
2914                                              $g->{ref}->full_path($p), 1);
2915                }
2916                my $c = '';
2917                foreach (split m#/#, $g->{path}->{left}) {
2918                        $c .= "/$_";
2919                        next unless ($paths->{$c} &&
2920                                     ($paths->{$c}->{action} =~ /^[AR]$/));
2921                        get_dir_check($self, $exists, $g, $r);
2922                }
2923        }
2924        values %$exists;
2925}
2926
2927sub minimize_url {
2928        my ($self) = @_;
2929        return $self->{url} if ($self->{url} eq $self->{repos_root});
2930        my $url = $self->{repos_root};
2931        my @components = split(m!/!, $self->{svn_path});
2932        my $c = '';
2933        do {
2934                $url .= "/$c" if length $c;
2935                eval { (ref $self)->new($url)->get_latest_revnum };
2936        } while ($@ && ($c = shift @components));
2937        $url;
2938}
2939
2940sub can_do_switch {
2941        my $self = shift;
2942        unless (defined $can_do_switch) {
2943                my $pool = SVN::Pool->new;
2944                my $rep = eval {
2945                        $self->do_switch(1, '', 0, $self->{url},
2946                                         SVN::Delta::Editor->new, $pool);
2947                };
2948                if ($@) {
2949                        $can_do_switch = 0;
2950                } else {
2951                        $rep->abort_report($pool);
2952                        $can_do_switch = 1;
2953                }
2954                $pool->clear;
2955        }
2956        $can_do_switch;
2957}
2958
2959sub skip_unknown_revs {
2960        my ($err) = @_;
2961        my $errno = $err->apr_err();
2962        # Maybe the branch we're tracking didn't
2963        # exist when the repo started, so it's
2964        # not an error if it doesn't, just continue
2965        #
2966        # Wonderfully consistent library, eh?
2967        # 160013 - svn:// and file://
2968        # 175002 - http(s)://
2969        # 175007 - http(s):// (this repo required authorization, too...)
2970        #   More codes may be discovered later...
2971        if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
2972                return;
2973        }
2974        die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
2975}
2976
2977# svn_log_changed_path_t objects passed to get_log are likely to be
2978# overwritten even if only the refs are copied to an external variable,
2979# so we should dup the structures in their entirety.  Using an externally
2980# passed pool (instead of our temporary and quickly cleared pool in
2981# Git::SVN::Ra) does not help matters at all...
2982sub dup_changed_paths {
2983        my ($paths) = @_;
2984        return undef unless $paths;
2985        my %ret;
2986        foreach my $p (keys %$paths) {
2987                my $i = $paths->{$p};
2988                my %s = map { $_ => $i->$_ }
2989                              qw/copyfrom_path copyfrom_rev action/;
2990                $ret{$p} = \%s;
2991        }
2992        \%ret;
2993}
2994
2995package Git::SVN::Log;
2996use strict;
2997use warnings;
2998use POSIX qw/strftime/;
2999use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
3000            %rusers $show_commit $incremental/;
3001my $l_fmt;
3002
3003sub cmt_showable {
3004        my ($c) = @_;
3005        return 1 if defined $c->{r};
3006        if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
3007                                $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
3008                my @log = command(qw/cat-file commit/, $c->{c});
3009                shift @log while ($log[0] ne "\n");
3010                shift @log;
3011                @{$c->{l}} = grep !/^git-svn-id: /, @log;
3012
3013                (undef, $c->{r}, undef) = ::extract_metadata(
3014                                (grep(/^git-svn-id: /, @log))[-1]);
3015        }
3016        return defined $c->{r};
3017}
3018
3019sub log_use_color {
3020        return 1 if $color;
3021        my ($dc, $dcvar);
3022        $dcvar = 'color.diff';
3023        $dc = `git-config --get $dcvar`;
3024        if ($dc eq '') {
3025                # nothing at all; fallback to "diff.color"
3026                $dcvar = 'diff.color';
3027                $dc = `git-config --get $dcvar`;
3028        }
3029        chomp($dc);
3030        if ($dc eq 'auto') {
3031                my $pc;
3032                $pc = `git-config --get color.pager`;
3033                if ($pc eq '') {
3034                        # does not have it -- fallback to pager.color
3035                        $pc = `git-config --bool --get pager.color`;
3036                }
3037                else {
3038                        $pc = `git-config --bool --get color.pager`;
3039                        if ($?) {
3040                                $pc = 'false';
3041                        }
3042                }
3043                chomp($pc);
3044                if (-t *STDOUT || (defined $pager && $pc eq 'true')) {
3045                        return ($ENV{TERM} && $ENV{TERM} ne 'dumb');
3046                }
3047                return 0;
3048        }
3049        return 0 if $dc eq 'never';
3050        return 1 if $dc eq 'always';
3051        chomp($dc = `git-config --bool --get $dcvar`);
3052        return ($dc eq 'true');
3053}
3054
3055sub git_svn_log_cmd {
3056        my ($r_min, $r_max, @args) = @_;
3057        my $head = 'HEAD';
3058        foreach my $x (@args) {
3059                last if $x eq '--';
3060                next unless ::verify_ref("$x^0");
3061                $head = $x;
3062                last;
3063        }
3064
3065        my $url;
3066        my ($fh, $ctx) = command_output_pipe('rev-list', $head);
3067        while (<$fh>) {
3068                chomp;
3069                $url = (::cmt_metadata($_))[0];
3070                last if defined $url;
3071        }
3072        close $fh; # break the pipe
3073
3074        my $gs = Git::SVN->find_by_url($url) || Git::SVN->_new;
3075        my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
3076                   $gs->refname);
3077        push @cmd, '-r' unless $non_recursive;
3078        push @cmd, qw/--raw --name-status/ if $verbose;
3079        push @cmd, '--color' if log_use_color();
3080        return @cmd unless defined $r_max;
3081        if ($r_max == $r_min) {
3082                push @cmd, '--max-count=1';
3083                if (my $c = $gs->rev_db_get($r_max)) {
3084                        push @cmd, $c;
3085                }
3086        } else {
3087                my ($c_min, $c_max);
3088                $c_max = $gs->rev_db_get($r_max);
3089                $c_min = $gs->rev_db_get($r_min);
3090                if (defined $c_min && defined $c_max) {
3091                        if ($r_max > $r_max) {
3092                                push @cmd, "$c_min..$c_max";
3093                        } else {
3094                                push @cmd, "$c_max..$c_min";
3095                        }
3096                } elsif ($r_max > $r_min) {
3097                        push @cmd, $c_max;
3098                } else {
3099                        push @cmd, $c_min;
3100                }
3101        }
3102        return @cmd;
3103}
3104
3105# adapted from pager.c
3106sub config_pager {
3107        $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
3108        if (!defined $pager) {
3109                $pager = 'less';
3110        } elsif (length $pager == 0 || $pager eq 'cat') {
3111                $pager = undef;
3112        }
3113}
3114
3115sub run_pager {
3116        return unless -t *STDOUT;
3117        pipe my $rfd, my $wfd or return;
3118        defined(my $pid = fork) or ::fatal "Can't fork: $!\n";
3119        if (!$pid) {
3120                open STDOUT, '>&', $wfd or
3121                                     ::fatal "Can't redirect to stdout: $!\n";
3122                return;
3123        }
3124        open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!\n";
3125        $ENV{LESS} ||= 'FRSX';
3126        exec $pager or ::fatal "Can't run pager: $! ($pager)\n";
3127}
3128
3129sub tz_to_s_offset {
3130        my ($tz) = @_;
3131        $tz =~ s/(\d\d)$//;
3132        return ($1 * 60) + ($tz * 3600);
3133}
3134
3135sub get_author_info {
3136        my ($dest, $author, $t, $tz) = @_;
3137        $author =~ s/(?:^\s*|\s*$)//g;
3138        $dest->{a_raw} = $author;
3139        my $au;
3140        if ($::_authors) {
3141                $au = $rusers{$author} || undef;
3142        }
3143        if (!$au) {
3144                ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
3145        }
3146        $dest->{t} = $t;
3147        $dest->{tz} = $tz;
3148        $dest->{a} = $au;
3149        # Date::Parse isn't in the standard Perl distro :(
3150        if ($tz =~ s/^\+//) {
3151                $t += tz_to_s_offset($tz);
3152        } elsif ($tz =~ s/^\-//) {
3153                $t -= tz_to_s_offset($tz);
3154        }
3155        $dest->{t_utc} = $t;
3156}
3157
3158sub process_commit {
3159        my ($c, $r_min, $r_max, $defer) = @_;
3160        if (defined $r_min && defined $r_max) {
3161                if ($r_min == $c->{r} && $r_min == $r_max) {
3162                        show_commit($c);
3163                        return 0;
3164                }
3165                return 1 if $r_min == $r_max;
3166                if ($r_min < $r_max) {
3167                        # we need to reverse the print order
3168                        return 0 if (defined $limit && --$limit < 0);
3169                        push @$defer, $c;
3170                        return 1;
3171                }
3172                if ($r_min != $r_max) {
3173                        return 1 if ($r_min < $c->{r});
3174                        return 1 if ($r_max > $c->{r});
3175                }
3176        }
3177        return 0 if (defined $limit && --$limit < 0);
3178        show_commit($c);
3179        return 1;
3180}
3181
3182sub show_commit {
3183        my $c = shift;
3184        if ($oneline) {
3185                my $x = "\n";
3186                if (my $l = $c->{l}) {
3187                        while ($l->[0] =~ /^\s*$/) { shift @$l }
3188                        $x = $l->[0];
3189                }
3190                $l_fmt ||= 'A' . length($c->{r});
3191                print 'r',pack($l_fmt, $c->{r}),' | ';
3192                print "$c->{c} | " if $show_commit;
3193                print $x;
3194        } else {
3195                show_commit_normal($c);
3196        }
3197}
3198
3199sub show_commit_changed_paths {
3200        my ($c) = @_;
3201        return unless $c->{changed};
3202        print "Changed paths:\n", @{$c->{changed}};
3203}
3204
3205sub show_commit_normal {
3206        my ($c) = @_;
3207        print '-' x72, "\nr$c->{r} | ";
3208        print "$c->{c} | " if $show_commit;
3209        print "$c->{a} | ", strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)",
3210                                 localtime($c->{t_utc})), ' | ';
3211        my $nr_line = 0;
3212
3213        if (my $l = $c->{l}) {
3214                while ($l->[$#$l] eq "\n" && $#$l > 0
3215                                          && $l->[($#$l - 1)] eq "\n") {
3216                        pop @$l;
3217                }
3218                $nr_line = scalar @$l;
3219                if (!$nr_line) {
3220                        print "1 line\n\n\n";
3221                } else {
3222                        if ($nr_line == 1) {
3223                                $nr_line = '1 line';
3224                        } else {
3225                                $nr_line .= ' lines';
3226                        }
3227                        print $nr_line, "\n";
3228                        show_commit_changed_paths($c);
3229                        print "\n";
3230                        print $_ foreach @$l;
3231                }
3232        } else {
3233                print "1 line\n";
3234                show_commit_changed_paths($c);
3235                print "\n";
3236
3237        }
3238        foreach my $x (qw/raw diff/) {
3239                if ($c->{$x}) {
3240                        print "\n";
3241                        print $_ foreach @{$c->{$x}}
3242                }
3243        }
3244}
3245
3246sub cmd_show_log {
3247        my (@args) = @_;
3248        my ($r_min, $r_max);
3249        my $r_last = -1; # prevent dupes
3250        if (defined $TZ) {
3251                $ENV{TZ} = $TZ;
3252        } else {
3253                delete $ENV{TZ};
3254        }
3255        if (defined $::_revision) {
3256                if ($::_revision =~ /^(\d+):(\d+)$/) {
3257                        ($r_min, $r_max) = ($1, $2);
3258                } elsif ($::_revision =~ /^\d+$/) {
3259                        $r_min = $r_max = $::_revision;
3260                } else {
3261                        ::fatal "-r$::_revision is not supported, use ",
3262                                "standard \'git log\' arguments instead\n";
3263                }
3264        }
3265
3266        config_pager();
3267        @args = (git_svn_log_cmd($r_min, $r_max, @args), @args);
3268        my $log = command_output_pipe(@args);
3269        run_pager();
3270        my (@k, $c, $d);
3271        my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
3272        while (<$log>) {
3273                if (/^${esc_color}commit ($::sha1_short)/o) {
3274                        my $cmt = $1;
3275                        if ($c && cmt_showable($c) && $c->{r} != $r_last) {
3276                                $r_last = $c->{r};
3277                                process_commit($c, $r_min, $r_max, \@k) or
3278                                                                goto out;
3279                        }
3280                        $d = undef;
3281                        $c = { c => $cmt };
3282                } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
3283                        get_author_info($c, $1, $2, $3);
3284                } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
3285                        # ignore
3286                } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
3287                        push @{$c->{raw}}, $_;
3288                } elsif (/^${esc_color}[ACRMDT]\t/) {
3289                        # we could add $SVN->{svn_path} here, but that requires
3290                        # remote access at the moment (repo_path_split)...
3291                        s#^(${esc_color})([ACRMDT])\t#$1   $2 #o;
3292                        push @{$c->{changed}}, $_;
3293                } elsif (/^${esc_color}diff /o) {
3294                        $d = 1;
3295                        push @{$c->{diff}}, $_;
3296                } elsif ($d) {
3297                        push @{$c->{diff}}, $_;
3298                } elsif (/^${esc_color}    (git-svn-id:.+)$/o) {
3299                        ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
3300                } elsif (s/^${esc_color}    //o) {
3301                        push @{$c->{l}}, $_;
3302                }
3303        }
3304        if ($c && defined $c->{r} && $c->{r} != $r_last) {
3305                $r_last = $c->{r};
3306                process_commit($c, $r_min, $r_max, \@k);
3307        }
3308        if (@k) {
3309                my $swap = $r_max;
3310                $r_max = $r_min;
3311                $r_min = $swap;
3312                process_commit($_, $r_min, $r_max) foreach reverse @k;
3313        }
3314out:
3315        close $log;
3316        print '-' x72,"\n" unless $incremental || $oneline;
3317}
3318
3319package Git::SVN::Migration;
3320# these version numbers do NOT correspond to actual version numbers
3321# of git nor git-svn.  They are just relative.
3322#
3323# v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
3324#
3325# v1 layout: .git/$id/info/url, refs/remotes/$id
3326#
3327# v2 layout: .git/svn/$id/info/url, refs/remotes/$id
3328#
3329# v3 layout: .git/svn/$id, refs/remotes/$id
3330#            - info/url may remain for backwards compatibility
3331#            - this is what we migrate up to this layout automatically,
3332#            - this will be used by git svn init on single branches
3333# v3.1 layout (auto migrated):
3334#            - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
3335#              for backwards compatibility
3336#
3337# v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
3338#            - this is only created for newly multi-init-ed
3339#              repositories.  Similar in spirit to the
3340#              --use-separate-remotes option in git-clone (now default)
3341#            - we do not automatically migrate to this (following
3342#              the example set by core git)
3343use strict;
3344use warnings;
3345use Carp qw/croak/;
3346use File::Path qw/mkpath/;
3347use File::Basename qw/dirname basename/;
3348use vars qw/$_minimize/;
3349
3350sub migrate_from_v0 {
3351        my $git_dir = $ENV{GIT_DIR};
3352        return undef unless -d $git_dir;
3353        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
3354        my $migrated = 0;
3355        while (<$fh>) {
3356                chomp;
3357                my ($id, $orig_ref) = ($_, $_);
3358                next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
3359                next unless -f "$git_dir/$id/info/url";
3360                my $new_ref = "refs/remotes/$id";
3361                if (::verify_ref("$new_ref^0")) {
3362                        print STDERR "W: $orig_ref is probably an old ",
3363                                     "branch used by an ancient version of ",
3364                                     "git-svn.\n",
3365                                     "However, $new_ref also exists.\n",
3366                                     "We will not be able ",
3367                                     "to use this branch until this ",
3368                                     "ambiguity is resolved.\n";
3369                        next;
3370                }
3371                print STDERR "Migrating from v0 layout...\n" if !$migrated;
3372                print STDERR "Renaming ref: $orig_ref => $new_ref\n";
3373                command_noisy('update-ref', $new_ref, $orig_ref);
3374                command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
3375                $migrated++;
3376        }
3377        command_close_pipe($fh, $ctx);
3378        print STDERR "Done migrating from v0 layout...\n" if $migrated;
3379        $migrated;
3380}
3381
3382sub migrate_from_v1 {
3383        my $git_dir = $ENV{GIT_DIR};
3384        my $migrated = 0;
3385        return $migrated unless -d $git_dir;
3386        my $svn_dir = "$git_dir/svn";
3387
3388        # just in case somebody used 'svn' as their $id at some point...
3389        return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
3390
3391        print STDERR "Migrating from a git-svn v1 layout...\n";
3392        mkpath([$svn_dir]);
3393        print STDERR "Data from a previous version of git-svn exists, but\n\t",
3394                     "$svn_dir\n\t(required for this version ",
3395                     "($::VERSION) of git-svn) does not. exist\n";
3396        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
3397        while (<$fh>) {
3398                my $x = $_;
3399                next unless $x =~ s#^refs/remotes/##;
3400                chomp $x;
3401                next unless -f "$git_dir/$x/info/url";
3402                my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
3403                next unless $u;
3404                my $dn = dirname("$git_dir/svn/$x");
3405                mkpath([$dn]) unless -d $dn;
3406                if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
3407                        mkpath(["$git_dir/svn/svn"]);
3408                        print STDERR " - $git_dir/$x/info => ",
3409                                        "$git_dir/svn/$x/info\n";
3410                        rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
3411                               croak "$!: $x";
3412                        # don't worry too much about these, they probably
3413                        # don't exist with repos this old (save for index,
3414                        # and we can easily regenerate that)
3415                        foreach my $f (qw/unhandled.log index .rev_db/) {
3416                                rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
3417                        }
3418                } else {
3419                        print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
3420                        rename "$git_dir/$x", "$git_dir/svn/$x" or
3421                               croak "$!: $x";
3422                }
3423                $migrated++;
3424        }
3425        command_close_pipe($fh, $ctx);
3426        print STDERR "Done migrating from a git-svn v1 layout\n";
3427        $migrated;
3428}
3429
3430sub read_old_urls {
3431        my ($l_map, $pfx, $path) = @_;
3432        my @dir;
3433        foreach (<$path/*>) {
3434                if (-r "$_/info/url") {
3435                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
3436                        my $ref_id = $pfx . basename $_;
3437                        my $url = ::file_to_s("$_/info/url");
3438                        $l_map->{$ref_id} = $url;
3439                } elsif (-d $_) {
3440                        push @dir, $_;
3441                }
3442        }
3443        foreach (@dir) {
3444                my $x = $_;
3445                $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
3446                read_old_urls($l_map, $x, $_);
3447        }
3448}
3449
3450sub migrate_from_v2 {
3451        my @cfg = command(qw/config -l/);
3452        return if grep /^svn-remote\..+\.url=/, @cfg;
3453        my %l_map;
3454        read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
3455        my $migrated = 0;
3456
3457        foreach my $ref_id (sort keys %l_map) {
3458                eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
3459                if ($@) {
3460                        Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
3461                }
3462                $migrated++;
3463        }
3464        $migrated;
3465}
3466
3467sub minimize_connections {
3468        my $r = Git::SVN::read_all_remotes();
3469        my $new_urls = {};
3470        my $root_repos = {};
3471        foreach my $repo_id (keys %$r) {
3472                my $url = $r->{$repo_id}->{url} or next;
3473                my $fetch = $r->{$repo_id}->{fetch} or next;
3474                my $ra = Git::SVN::Ra->new($url);
3475
3476                # skip existing cases where we already connect to the root
3477                if (($ra->{url} eq $ra->{repos_root}) ||
3478                    (Git::SVN::sanitize_remote_name($ra->{repos_root}) eq
3479                     $repo_id)) {
3480                        $root_repos->{$ra->{url}} = $repo_id;
3481                        next;
3482                }
3483
3484                my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
3485                my $root_path = $ra->{url};
3486                $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
3487                foreach my $path (keys %$fetch) {
3488                        my $ref_id = $fetch->{$path};
3489                        my $gs = Git::SVN->new($ref_id, $repo_id, $path);
3490
3491                        # make sure we can read when connecting to
3492                        # a higher level of a repository
3493                        my ($last_rev, undef) = $gs->last_rev_commit;
3494                        if (!defined $last_rev) {
3495                                $last_rev = eval {
3496                                        $root_ra->get_latest_revnum;
3497                                };
3498                                next if $@;
3499                        }
3500                        my $new = $root_path;
3501                        $new .= length $path ? "/$path" : '';
3502                        eval {
3503                                $root_ra->get_log([$new], $last_rev, $last_rev,
3504                                                  0, 0, 1, sub { });
3505                        };
3506                        next if $@;
3507                        $new_urls->{$ra->{repos_root}}->{$new} =
3508                                { ref_id => $ref_id,
3509                                  old_repo_id => $repo_id,
3510                                  old_path => $path };
3511                }
3512        }
3513
3514        my @emptied;
3515        foreach my $url (keys %$new_urls) {
3516                # see if we can re-use an existing [svn-remote "repo_id"]
3517                # instead of creating a(n ugly) new section:
3518                my $repo_id = $root_repos->{$url} ||
3519                              Git::SVN::sanitize_remote_name($url);
3520
3521                my $fetch = $new_urls->{$url};
3522                foreach my $path (keys %$fetch) {
3523                        my $x = $fetch->{$path};
3524                        Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
3525                        my $pfx = "svn-remote.$x->{old_repo_id}";
3526
3527                        my $old_fetch = quotemeta("$x->{old_path}:".
3528                                                  "refs/remotes/$x->{ref_id}");
3529                        command_noisy(qw/config --unset/,
3530                                      "$pfx.fetch", '^'. $old_fetch . '$');
3531                        delete $r->{$x->{old_repo_id}}->
3532                               {fetch}->{$x->{old_path}};
3533                        if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
3534                                command_noisy(qw/config --unset/,
3535                                              "$pfx.url");
3536                                push @emptied, $x->{old_repo_id}
3537                        }
3538                }
3539        }
3540        if (@emptied) {
3541                my $file = $ENV{GIT_CONFIG} || $ENV{GIT_CONFIG_LOCAL} ||
3542                           "$ENV{GIT_DIR}/config";
3543                print STDERR <<EOF;
3544The following [svn-remote] sections in your config file ($file) are empty
3545and can be safely removed:
3546EOF
3547                print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
3548        }
3549}
3550
3551sub migration_check {
3552        migrate_from_v0();
3553        migrate_from_v1();
3554        migrate_from_v2();
3555        minimize_connections() if $_minimize;
3556}
3557
3558package Git::IndexInfo;
3559use strict;
3560use warnings;
3561use Git qw/command_input_pipe command_close_pipe/;
3562
3563sub new {
3564        my ($class) = @_;
3565        my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
3566        bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
3567}
3568
3569sub remove {
3570        my ($self, $path) = @_;
3571        if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
3572                return ++$self->{nr};
3573        }
3574        undef;
3575}
3576
3577sub update {
3578        my ($self, $mode, $hash, $path) = @_;
3579        if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
3580                return ++$self->{nr};
3581        }
3582        undef;
3583}
3584
3585sub DESTROY {
3586        my ($self) = @_;
3587        command_close_pipe($self->{gui}, $self->{ctx});
3588}
3589
3590package Git::SVN::GlobSpec;
3591use strict;
3592use warnings;
3593
3594sub new {
3595        my ($class, $glob) = @_;
3596        my $re = $glob;
3597        $re =~ s!/+$!!g; # no need for trailing slashes
3598        my $nr = ($re =~ s!^(.*)\*(.*)$!\(\[^/\]+\)!g);
3599        my ($left, $right) = ($1, $2);
3600        if ($nr > 1) {
3601                die "Only one '*' wildcard expansion ",
3602                    "is supported (got $nr): '$glob'\n";
3603        } elsif ($nr == 0) {
3604                die "One '*' is needed for glob: '$glob'\n";
3605        }
3606        $re = quotemeta($left) . $re . quotemeta($right);
3607        if (length $left && !($left =~ s!/+$!!g)) {
3608                die "Missing trailing '/' on left side of: '$glob' ($left)\n";
3609        }
3610        if (length $right && !($right =~ s!^/+!!g)) {
3611                die "Missing leading '/' on right side of: '$glob' ($right)\n";
3612        }
3613        my $left_re = qr/^\/\Q$left\E(\/|$)/;
3614        bless { left => $left, right => $right, left_regex => $left_re,
3615                regex => qr/$re/, glob => $glob }, $class;
3616}
3617
3618sub full_path {
3619        my ($self, $path) = @_;
3620        return (length $self->{left} ? "$self->{left}/" : '') .
3621               $path . (length $self->{right} ? "/$self->{right}" : '');
3622}
3623
3624__END__
3625
3626Data structures:
3627
3628
3629$remotes = { # returned by read_all_remotes()
3630        'svn' => {
3631                # svn-remote.svn.url=https://svn.musicpd.org
3632                url => 'https://svn.musicpd.org',
3633                # svn-remote.svn.fetch=mpd/trunk:trunk
3634                fetch => {
3635                        'mpd/trunk' => 'trunk',
3636                },
3637                # svn-remote.svn.tags=mpd/tags/*:tags/*
3638                tags => {
3639                        path => {
3640                                left => 'mpd/tags',
3641                                right => '',
3642                                regex => qr!mpd/tags/([^/]+)$!,
3643                                glob => 'tags/*',
3644                        },
3645                        ref => {
3646                                left => 'tags',
3647                                right => '',
3648                                regex => qr!tags/([^/]+)$!,
3649                                glob => 'tags/*',
3650                        },
3651                }
3652        }
3653};
3654
3655$log_entry hashref as returned by libsvn_log_entry()
3656{
3657        log => 'whitespace-formatted log entry
3658',                                              # trailing newline is preserved
3659        revision => '8',                        # integer
3660        date => '2004-02-24T17:01:44.108345Z',  # commit date
3661        author => 'committer name'
3662};
3663
3664
3665# this is generated by generate_diff();
3666@mods = array of diff-index line hashes, each element represents one line
3667        of diff-index output
3668
3669diff-index line ($m hash)
3670{
3671        mode_a => first column of diff-index output, no leading ':',
3672        mode_b => second column of diff-index output,
3673        sha1_b => sha1sum of the final blob,
3674        chg => change type [MCRADT],
3675        file_a => original file name of a file (iff chg is 'C' or 'R')
3676        file_b => new/current file name of a file (any chg)
3677}
3678;
3679
3680# retval of read_url_paths{,_all}();
3681$l_map = {
3682        # repository root url
3683        'https://svn.musicpd.org' => {
3684                # repository path               # GIT_SVN_ID
3685                'mpd/trunk'             =>      'trunk',
3686                'mpd/tags/0.11.5'       =>      'tags/0.11.5',
3687        },
3688}
3689
3690Notes:
3691        I don't trust the each() function on unless I created %hash myself
3692        because the internal iterator may not have started at base.