15d65e21e18c2f5313c0678745ab37eaaf518477
   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                $SVN_URL
   8                $GIT_SVN_INDEX $GIT_SVN
   9                $GIT_DIR $GIT_SVN_DIR $REVDB
  10                $_follow_parent $sha1 $sha1_short $_revision
  11                $_cp_remote $_upgrade $_rmdir $_q $_cp_similarity
  12                $_find_copies_harder $_l $_authors %users/;
  13$AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
  14$VERSION = '@@GIT_VERSION@@';
  15
  16$ENV{GIT_DIR} ||= '.git';
  17$Git::SVN::default_repo_id = $ENV{GIT_SVN_ID} || 'git-svn';
  18
  19my $LC_ALL = $ENV{LC_ALL};
  20$Git::SVN::Log::TZ = $ENV{TZ};
  21# make sure the svn binary gives consistent output between locales and TZs:
  22$ENV{TZ} = 'UTC';
  23$ENV{LC_ALL} = 'C';
  24$| = 1; # unbuffer STDOUT
  25
  26sub fatal (@) { print STDERR @_; exit 1 }
  27require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
  28require SVN::Ra;
  29require SVN::Delta;
  30if ($SVN::Core::VERSION lt '1.1.0') {
  31        fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)\n";
  32}
  33push @Git::SVN::Ra::ISA, 'SVN::Ra';
  34push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
  35push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
  36use Carp qw/croak/;
  37use IO::File qw//;
  38use File::Basename qw/dirname basename/;
  39use File::Path qw/mkpath/;
  40use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev pass_through/;
  41use IPC::Open3;
  42use Git;
  43
  44BEGIN {
  45        my $s;
  46        foreach (qw/command command_oneline command_noisy command_output_pipe
  47                    command_input_pipe command_close_pipe/) {
  48                $s .= "*SVN::Git::Editor::$_ = *SVN::Git::Fetcher::$_ = ".
  49                      "*Git::SVN::Migration::$_ = ".
  50                      "*Git::SVN::Log::$_ = *Git::SVN::$_ = *$_ = *Git::$_; ";
  51        }
  52        eval $s;
  53}
  54
  55my ($SVN);
  56
  57my $_optimize_commits = 1 unless $ENV{GIT_SVN_NO_OPTIMIZE_COMMITS};
  58$sha1 = qr/[a-f\d]{40}/;
  59$sha1_short = qr/[a-f\d]{4,40}/;
  60my ($_stdin, $_help, $_edit,
  61        $_repack, $_repack_nr, $_repack_flags,
  62        $_message, $_file, $_no_metadata,
  63        $_template, $_shared,
  64        $_version, $_upgrade,
  65        $_merge, $_strategy, $_dry_run,
  66        $_prefix);
  67
  68my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
  69                    'config-dir=s' => \$Git::SVN::Ra::config_dir,
  70                    'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache );
  71my %fc_opts = ( 'follow-parent|follow' => \$_follow_parent,
  72                'authors-file|A=s' => \$_authors,
  73                'repack:i' => \$_repack,
  74                'no-metadata' => \$_no_metadata,
  75                'quiet|q' => \$_q,
  76                'repack-flags|repack-args|repack-opts=s' => \$_repack_flags,
  77                %remote_opts );
  78
  79my ($_trunk, $_tags, $_branches);
  80my %multi_opts = ( 'trunk|T=s' => \$_trunk,
  81                'tags|t=s' => \$_tags,
  82                'branches|b=s' => \$_branches );
  83my %init_opts = ( 'template=s' => \$_template, 'shared' => \$_shared );
  84my %cmt_opts = ( 'edit|e' => \$_edit,
  85                'rmdir' => \$_rmdir,
  86                'find-copies-harder' => \$_find_copies_harder,
  87                'l=i' => \$_l,
  88                'copy-similarity|C=i'=> \$_cp_similarity
  89);
  90
  91my %cmd = (
  92        fetch => [ \&cmd_fetch, "Download new revisions from SVN",
  93                        { 'revision|r=s' => \$_revision, %fc_opts } ],
  94        init => [ \&cmd_init, "Initialize a repo for tracking" .
  95                          " (requires URL argument)",
  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        rebuild => [ \&cmd_rebuild, "Rebuild git-svn metadata (after git clone)",
 109                        { 'copy-remote|remote=s' => \$_cp_remote,
 110                          'upgrade' => \$_upgrade } ],
 111        'multi-init' => [ \&cmd_multi_init,
 112                        'Initialize multiple trees (like git-svnimport)',
 113                        { %multi_opts, %init_opts, %remote_opts,
 114                         'revision|r=i' => \$_revision,
 115                         'prefix=s' => \$_prefix,
 116                        } ],
 117        'multi-fetch' => [ \&cmd_multi_fetch,
 118                        'Fetch multiple trees (like git-svnimport)',
 119                        \%fc_opts ],
 120        'migrate' => [ sub { },
 121                       # no-op, we automatically run this anyways,
 122                       'Migrate configuration/metadata/layout from
 123                        previous versions of git-svn',
 124                        \%remote_opts ],
 125        'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
 126                        { 'limit=i' => \$Git::SVN::Log::limit,
 127                          'revision|r=s' => \$_revision,
 128                          'verbose|v' => \$Git::SVN::Log::verbose,
 129                          'incremental' => \$Git::SVN::Log::incremental,
 130                          'oneline' => \$Git::SVN::Log::oneline,
 131                          'show-commit' => \$Git::SVN::Log::show_commit,
 132                          'non-recursive' => \$Git::SVN::Log::non_recursive,
 133                          'authors-file|A=s' => \$_authors,
 134                          'color' => \$Git::SVN::Log::color,
 135                          'pager=s' => \$Git::SVN::Log::pager,
 136                        } ],
 137        'commit-diff' => [ \&cmd_commit_diff,
 138                           'Commit a diff between two trees',
 139                        { 'message|m=s' => \$_message,
 140                          'file|F=s' => \$_file,
 141                          'revision|r=s' => \$_revision,
 142                        %cmt_opts } ],
 143);
 144
 145my $cmd;
 146for (my $i = 0; $i < @ARGV; $i++) {
 147        if (defined $cmd{$ARGV[$i]}) {
 148                $cmd = $ARGV[$i];
 149                splice @ARGV, $i, 1;
 150                last;
 151        }
 152};
 153
 154my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
 155
 156read_repo_config(\%opts);
 157my $rv = GetOptions(%opts, 'help|H|h' => \$_help,
 158                                'version|V' => \$_version,
 159                                'minimize-connections' =>
 160                                  \$Git::SVN::Migration::_minimize,
 161                                'id|i=s' => \$Git::SVN::default_repo_id);
 162exit 1 if (!$rv && $cmd ne 'log');
 163
 164usage(0) if $_help;
 165version() if $_version;
 166usage(1) unless defined $cmd;
 167load_authors() if $_authors;
 168unless ($cmd =~ /^(?:init|rebuild|multi-init|commit-diff)$/) {
 169        Git::SVN::Migration::migration_check();
 170}
 171$cmd{$cmd}->[0]->(@ARGV);
 172exit 0;
 173
 174####################### primary functions ######################
 175sub usage {
 176        my $exit = shift || 0;
 177        my $fd = $exit ? \*STDERR : \*STDOUT;
 178        print $fd <<"";
 179git-svn - bidirectional operations between a single Subversion tree and git
 180Usage: $0 <command> [options] [arguments]\n
 181
 182        print $fd "Available commands:\n" unless $cmd;
 183
 184        foreach (sort keys %cmd) {
 185                next if $cmd && $cmd ne $_;
 186                print $fd '  ',pack('A17',$_),$cmd{$_}->[1],"\n";
 187                foreach (keys %{$cmd{$_}->[2]}) {
 188                        # prints out arguments as they should be passed:
 189                        my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
 190                        print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
 191                                                        "--$_" : "-$_" }
 192                                                split /\|/,$_)," $x\n";
 193                }
 194        }
 195        print $fd <<"";
 196\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
 197arbitrary identifier if you're tracking multiple SVN branches/repositories in
 198one git repository and want to keep them separate.  See git-svn(1) for more
 199information.
 200
 201        exit $exit;
 202}
 203
 204sub version {
 205        print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
 206        exit 0;
 207}
 208
 209sub cmd_rebuild {
 210        my $url = shift;
 211        my $gs = $url ? Git::SVN->init($url)
 212                      : eval { Git::SVN->new };
 213        $gs ||= Git::SVN->_new;
 214        if (!verify_ref($gs->refname.'^0')) {
 215                $gs->copy_remote_ref;
 216        }
 217
 218        my ($rev_list, $ctx) = command_output_pipe("rev-list", $gs->refname);
 219        my $latest;
 220        my $svn_uuid;
 221        while (<$rev_list>) {
 222                chomp;
 223                my $c = $_;
 224                fatal "Non-SHA1: $c\n" unless $c =~ /^$sha1$/o;
 225                my ($url, $rev, $uuid) = cmt_metadata($c);
 226
 227                # ignore merges (from set-tree)
 228                next if (!defined $rev || !$uuid);
 229
 230                # if we merged or otherwise started elsewhere, this is
 231                # how we break out of it
 232                if ((defined $svn_uuid && ($uuid ne $svn_uuid)) ||
 233                    ($gs->{url} && $url && ($url ne $gs->{url}))) {
 234                        next;
 235                }
 236
 237                unless (defined $latest) {
 238                        if (!$gs->{url} && !$url) {
 239                                fatal "SVN repository location required\n";
 240                        }
 241                        $gs = Git::SVN->init($url);
 242                        $latest = $rev;
 243                }
 244                $gs->rev_db_set($rev, $c);
 245                print "r$rev = $c\n";
 246        }
 247        command_close_pipe($rev_list, $ctx);
 248}
 249
 250sub do_git_init_db {
 251        unless (-d $ENV{GIT_DIR}) {
 252                my @init_db = ('init');
 253                push @init_db, "--template=$_template" if defined $_template;
 254                push @init_db, "--shared" if defined $_shared;
 255                command_noisy(@init_db);
 256        }
 257}
 258
 259sub cmd_init {
 260        my $url = shift or die "SVN repository location required " .
 261                                "as a command-line argument\n";
 262        if (my $repo_path = shift) {
 263                unless (-d $repo_path) {
 264                        mkpath([$repo_path]);
 265                }
 266                chdir $repo_path or croak $!;
 267                $ENV{GIT_DIR} = $repo_path . "/.git";
 268        }
 269        do_git_init_db();
 270
 271        Git::SVN->init($url);
 272}
 273
 274sub cmd_fetch {
 275        my $gs = Git::SVN->new;
 276        $gs->fetch(@_);
 277        if ($gs->{last_commit} && !verify_ref('refs/heads/master^0')) {
 278                command_noisy(qw(update-ref refs/heads/master),
 279                              $gs->{last_commit});
 280        }
 281}
 282
 283sub cmd_set_tree {
 284        my (@commits) = @_;
 285        if ($_stdin || !@commits) {
 286                print "Reading from stdin...\n";
 287                @commits = ();
 288                while (<STDIN>) {
 289                        if (/\b($sha1_short)\b/o) {
 290                                unshift @commits, $1;
 291                        }
 292                }
 293        }
 294        my @revs;
 295        foreach my $c (@commits) {
 296                my @tmp = command('rev-parse',$c);
 297                if (scalar @tmp == 1) {
 298                        push @revs, $tmp[0];
 299                } elsif (scalar @tmp > 1) {
 300                        push @revs, reverse(command('rev-list',@tmp));
 301                } else {
 302                        fatal "Failed to rev-parse $c\n";
 303                }
 304        }
 305        my $gs = Git::SVN->new;
 306        my ($r_last, $cmt_last) = $gs->last_rev_commit;
 307        $gs->fetch;
 308        if ($r_last != $gs->{last_rev}) {
 309                fatal "There are new revisions that were fetched ",
 310                      "and need to be merged (or acknowledged) ",
 311                      "before committing.\nlast rev: $r_last\n",
 312                      " current: $gs->{last_rev}\n";
 313        }
 314        $gs->set_tree($_) foreach @revs;
 315        print "Done committing ",scalar @revs," revisions to SVN\n";
 316}
 317
 318sub cmd_dcommit {
 319        my $head = shift;
 320        my $gs = Git::SVN->new;
 321        $head ||= 'HEAD';
 322        my @refs = command(qw/rev-list --no-merges/, $gs->refname."..$head");
 323        my $last_rev;
 324        foreach my $d (reverse @refs) {
 325                if (!verify_ref("$d~1")) {
 326                        fatal "Commit $d\n",
 327                              "has no parent commit, and therefore ",
 328                              "nothing to diff against.\n",
 329                              "You should be working from a repository ",
 330                              "originally created by git-svn\n";
 331                }
 332                unless (defined $last_rev) {
 333                        (undef, $last_rev, undef) = cmt_metadata("$d~1");
 334                        unless (defined $last_rev) {
 335                                fatal "Unable to extract revision information ",
 336                                      "from commit $d~1\n";
 337                        }
 338                }
 339                if ($_dry_run) {
 340                        print "diff-tree $d~1 $d\n";
 341                } else {
 342                        my $log = get_commit_entry($d)->{log};
 343                        my $ra = $gs->ra;
 344                        my $pool = SVN::Pool->new;
 345                        my %ed_opts = ( r => $last_rev,
 346                                        ra => $ra->dup,
 347                                        svn_path => $ra->{svn_path} );
 348                        my $ed = SVN::Git::Editor->new(\%ed_opts,
 349                                         $ra->get_commit_editor($log,
 350                                         sub { print "Committed r$_[0]\n";
 351                                               $last_rev = $_[0]; }),
 352                                         $pool);
 353                        my $mods = $ed->apply_diff("$d~1", $d);
 354                        if (@$mods == 0) {
 355                                print "No changes\n$d~1 == $d\n";
 356                        }
 357                }
 358        }
 359        return if $_dry_run;
 360        $gs->fetch;
 361        # we always want to rebase against the current HEAD, not any
 362        # head that was passed to us
 363        my @diff = command('diff-tree', 'HEAD', $gs->refname, '--');
 364        my @finish;
 365        if (@diff) {
 366                @finish = qw/rebase/;
 367                push @finish, qw/--merge/ if $_merge;
 368                push @finish, "--strategy=$_strategy" if $_strategy;
 369                print STDERR "W: HEAD and ", $gs->refname, " differ, ",
 370                             "using @finish:\n", "@diff";
 371        } else {
 372                print "No changes between current HEAD and ",
 373                      $gs->refname, "\nResetting to the latest ",
 374                      $gs->refname, "\n";
 375                @finish = qw/reset --mixed/;
 376        }
 377        command_noisy(@finish, $gs->refname);
 378}
 379
 380sub cmd_show_ignore {
 381        my $gs = Git::SVN->new;
 382        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 383        $gs->traverse_ignore(\*STDOUT, '', $r);
 384}
 385
 386sub cmd_multi_init {
 387        my $url = shift;
 388        unless (defined $_trunk || defined $_branches || defined $_tags) {
 389                usage(1);
 390        }
 391        do_git_init_db();
 392        $_prefix = '' unless defined $_prefix;
 393        $url =~ s#/+$## if defined $url;
 394        if (defined $_trunk) {
 395                my $trunk_ref = $_prefix . 'trunk';
 396                # try both old-style and new-style lookups:
 397                my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
 398                unless ($gs_trunk) {
 399                        my ($trunk_url, $trunk_path) =
 400                                              complete_svn_url($url, $_trunk);
 401                        $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
 402                                                   undef, $trunk_ref);
 403                }
 404        }
 405        return unless defined $_branches || defined $_tags;
 406        my $ra = $url ? Git::SVN::Ra->new($url) : undef;
 407        complete_url_ls_init($ra, $_branches, '--branches/-b', $_prefix);
 408        complete_url_ls_init($ra, $_tags, '--tags/-t', $_prefix . 'tags/');
 409}
 410
 411sub cmd_multi_fetch {
 412        my @gs;
 413        foreach (command(qw/config -l/)) {
 414                next unless m!^svn-remote\.(.+)\.fetch=
 415                              \s*(.*)\s*:\s*refs/remotes/(.+)\s*$!x;
 416                my ($repo_id, $path, $ref_id) = ($1, $2, $3);
 417                push @gs, Git::SVN->new($ref_id, $repo_id, $path);
 418        }
 419        foreach (@gs) {
 420                $_->fetch;
 421        }
 422}
 423
 424# this command is special because it requires no metadata
 425sub cmd_commit_diff {
 426        my ($ta, $tb, $url) = @_;
 427        my $usage = "Usage: $0 commit-diff -r<revision> ".
 428                    "<tree-ish> <tree-ish> [<URL>]\n";
 429        fatal($usage) if (!defined $ta || !defined $tb);
 430        if (!defined $url) {
 431                my $gs = eval { Git::SVN->new };
 432                if (!$gs) {
 433                        fatal("Needed URL or usable git-svn --id in ",
 434                              "the command-line\n", $usage);
 435                }
 436                $url = $gs->{url};
 437        }
 438        unless (defined $_revision) {
 439                fatal("-r|--revision is a required argument\n", $usage);
 440        }
 441        if (defined $_message && defined $_file) {
 442                fatal("Both --message/-m and --file/-F specified ",
 443                      "for the commit message.\n",
 444                      "I have no idea what you mean\n");
 445        }
 446        if (defined $_file) {
 447                $_message = file_to_s($_file);
 448        } else {
 449                $_message ||= get_commit_entry($tb)->{log};
 450        }
 451        my $ra ||= Git::SVN::Ra->new($url);
 452        my $r = $_revision;
 453        if ($r eq 'HEAD') {
 454                $r = $ra->get_latest_revnum;
 455        } elsif ($r !~ /^\d+$/) {
 456                die "revision argument: $r not understood by git-svn\n";
 457        }
 458        my $pool = SVN::Pool->new;
 459        my %ed_opts = ( r => $r,
 460                        ra => $ra->dup,
 461                        svn_path => $ra->{svn_path} );
 462        my $ed = SVN::Git::Editor->new(\%ed_opts,
 463                                       $ra->get_commit_editor($_message,
 464                                         sub { print "Committed r$_[0]\n" }),
 465                                       $pool);
 466        my $mods = $ed->apply_diff($ta, $tb);
 467        if (@$mods == 0) {
 468                print "No changes\n$ta == $tb\n";
 469        }
 470        $pool->clear;
 471}
 472
 473########################### utility functions #########################
 474
 475sub complete_svn_url {
 476        my ($url, $path) = @_;
 477        $path =~ s#/+$##;
 478        if ($path !~ m#^[a-z\+]+://#) {
 479                if (!defined $url || $url !~ m#^[a-z\+]+://#) {
 480                        fatal("E: '$path' is not a complete URL ",
 481                              "and a separate URL is not specified\n");
 482                }
 483                return ($url, $path);
 484        }
 485        return ($path, '');
 486}
 487
 488sub complete_url_ls_init {
 489        my ($ra, $repo_path, $switch, $pfx) = @_;
 490        unless ($repo_path) {
 491                print STDERR "W: $switch not specified\n";
 492                return;
 493        }
 494        $repo_path =~ s#/+$##;
 495        if ($repo_path =~ m#^[a-z\+]+://#) {
 496                $ra = Git::SVN::Ra->new($repo_path);
 497                $repo_path = '';
 498        } else {
 499                $repo_path =~ s#^/+##;
 500                unless ($ra) {
 501                        fatal("E: '$repo_path' is not a complete URL ",
 502                              "and a separate URL is not specified\n");
 503                }
 504        }
 505        my $r = defined $_revision ? $_revision : $ra->get_latest_revnum;
 506        my ($dirent, undef, undef) = $ra->get_dir($repo_path, $r);
 507        my $url = $ra->{url};
 508        foreach my $d (sort keys %$dirent) {
 509                next if ($dirent->{$d}->kind != $SVN::Node::dir);
 510                my $path =  "$repo_path/$d";
 511                my $ref = "$pfx$d";
 512                my $gs = eval { Git::SVN->new($ref) };
 513                # don't try to init already existing refs
 514                unless ($gs) {
 515                        print "init $url/$path => $ref\n";
 516                        Git::SVN->init($url, $path, undef, $ref);
 517                }
 518        }
 519}
 520
 521sub verify_ref {
 522        my ($ref) = @_;
 523        eval { command_oneline([ 'rev-parse', '--verify', $ref ],
 524                               { STDERR => 0 }); };
 525}
 526
 527sub get_tree_from_treeish {
 528        my ($treeish) = @_;
 529        # $treeish can be a symbolic ref, too:
 530        my $type = command_oneline(qw/cat-file -t/, $treeish);
 531        my $expected;
 532        while ($type eq 'tag') {
 533                ($treeish, $type) = command(qw/cat-file tag/, $treeish);
 534        }
 535        if ($type eq 'commit') {
 536                $expected = (grep /^tree /, command(qw/cat-file commit/,
 537                                                    $treeish))[0];
 538                ($expected) = ($expected =~ /^tree ($sha1)$/o);
 539                die "Unable to get tree from $treeish\n" unless $expected;
 540        } elsif ($type eq 'tree') {
 541                $expected = $treeish;
 542        } else {
 543                die "$treeish is a $type, expected tree, tag or commit\n";
 544        }
 545        return $expected;
 546}
 547
 548sub get_commit_entry {
 549        my ($treeish) = shift;
 550        my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
 551        my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
 552        my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
 553        open my $log_fh, '>', $commit_editmsg or croak $!;
 554
 555        my $type = command_oneline(qw/cat-file -t/, $treeish);
 556        if ($type eq 'commit' || $type eq 'tag') {
 557                my ($msg_fh, $ctx) = command_output_pipe('cat-file',
 558                                                         $type, $treeish);
 559                my $in_msg = 0;
 560                while (<$msg_fh>) {
 561                        if (!$in_msg) {
 562                                $in_msg = 1 if (/^\s*$/);
 563                        } elsif (/^git-svn-id: /) {
 564                                # skip this for now, we regenerate the
 565                                # correct one on re-fetch anyways
 566                                # TODO: set *:merge properties or like...
 567                        } else {
 568                                print $log_fh $_ or croak $!;
 569                        }
 570                }
 571                command_close_pipe($msg_fh, $ctx);
 572        }
 573        close $log_fh or croak $!;
 574
 575        if ($_edit || ($type eq 'tree')) {
 576                my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
 577                # TODO: strip out spaces, comments, like git-commit.sh
 578                system($editor, $commit_editmsg);
 579        }
 580        rename $commit_editmsg, $commit_msg or croak $!;
 581        open $log_fh, '<', $commit_msg or croak $!;
 582        { local $/; chomp($log_entry{log} = <$log_fh>); }
 583        close $log_fh or croak $!;
 584        unlink $commit_msg;
 585        \%log_entry;
 586}
 587
 588sub s_to_file {
 589        my ($str, $file, $mode) = @_;
 590        open my $fd,'>',$file or croak $!;
 591        print $fd $str,"\n" or croak $!;
 592        close $fd or croak $!;
 593        chmod ($mode &~ umask, $file) if (defined $mode);
 594}
 595
 596sub file_to_s {
 597        my $file = shift;
 598        open my $fd,'<',$file or croak "$!: file: $file\n";
 599        local $/;
 600        my $ret = <$fd>;
 601        close $fd or croak $!;
 602        $ret =~ s/\s*$//s;
 603        return $ret;
 604}
 605
 606# '<svn username> = real-name <email address>' mapping based on git-svnimport:
 607sub load_authors {
 608        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
 609        my $log = $cmd eq 'log';
 610        while (<$authors>) {
 611                chomp;
 612                next unless /^(\S+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
 613                my ($user, $name, $email) = ($1, $2, $3);
 614                if ($log) {
 615                        $Git::SVN::Log::rusers{"$name <$email>"} = $user;
 616                } else {
 617                        $users{$user} = [$name, $email];
 618                }
 619        }
 620        close $authors or croak $!;
 621}
 622
 623# convert GetOpt::Long specs for use by git-config
 624sub read_repo_config {
 625        return unless -d $ENV{GIT_DIR};
 626        my $opts = shift;
 627        foreach my $o (keys %$opts) {
 628                my $v = $opts->{$o};
 629                my ($key) = ($o =~ /^([a-z\-]+)/);
 630                $key =~ s/-//g;
 631                my $arg = 'git-config';
 632                $arg .= ' --int' if ($o =~ /[:=]i$/);
 633                $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
 634                if (ref $v eq 'ARRAY') {
 635                        chomp(my @tmp = `$arg --get-all svn.$key`);
 636                        @$v = @tmp if @tmp;
 637                } else {
 638                        chomp(my $tmp = `$arg --get svn.$key`);
 639                        if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
 640                                $$v = $tmp;
 641                        }
 642                }
 643        }
 644}
 645
 646sub extract_metadata {
 647        my $id = shift or return (undef, undef, undef);
 648        my ($url, $rev, $uuid) = ($id =~ /^git-svn-id:\s(\S+?)\@(\d+)
 649                                                        \s([a-f\d\-]+)$/x);
 650        if (!defined $rev || !$uuid || !$url) {
 651                # some of the original repositories I made had
 652                # identifiers like this:
 653                ($rev, $uuid) = ($id =~/^git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
 654        }
 655        return ($url, $rev, $uuid);
 656}
 657
 658sub cmt_metadata {
 659        return extract_metadata((grep(/^git-svn-id: /,
 660                command(qw/cat-file commit/, shift)))[-1]);
 661}
 662
 663sub get_commit_time {
 664        my $cmt = shift;
 665        my $fh = command_output_pipe(qw/rev-list --pretty=raw -n1/, $cmt);
 666        while (<$fh>) {
 667                /^committer\s(?:.+) (\d+) ([\-\+]?\d+)$/ or next;
 668                my ($s, $tz) = ($1, $2);
 669                if ($tz =~ s/^\+//) {
 670                        $s += tz_to_s_offset($tz);
 671                } elsif ($tz =~ s/^\-//) {
 672                        $s -= tz_to_s_offset($tz);
 673                }
 674                close $fh;
 675                return $s;
 676        }
 677        die "Can't get commit time for commit: $cmt\n";
 678}
 679
 680sub tz_to_s_offset {
 681        my ($tz) = @_;
 682        $tz =~ s/(\d\d)$//;
 683        return ($1 * 60) + ($tz * 3600);
 684}
 685
 686package Git::SVN;
 687use strict;
 688use warnings;
 689use vars qw/$default_repo_id/;
 690use Carp qw/croak/;
 691use File::Path qw/mkpath/;
 692use IPC::Open3;
 693
 694# properties that we do not log:
 695my %SKIP_PROP;
 696BEGIN {
 697        %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
 698                                        svn:special svn:executable
 699                                        svn:entry:committed-rev
 700                                        svn:entry:last-author
 701                                        svn:entry:uuid
 702                                        svn:entry:committed-date/;
 703}
 704
 705sub read_all_remotes {
 706        my $r = {};
 707        foreach (grep { s/^svn-remote\.// } command(qw/repo-config -l/)) {
 708                if (m!^(.+)\.fetch=\s*(.*)\s*:\s*refs/remotes/(.+)\s*$!) {
 709                        $r->{$1}->{fetch}->{$2} = $3;
 710                } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
 711                        $r->{$1}->{url} = $2;
 712                }
 713        }
 714        $r;
 715}
 716
 717# we allow more chars than remotes2config.sh...
 718sub sanitize_remote_name {
 719        my ($name) = @_;
 720        $name =~ tr{A-Za-z0-9:,/+-}{.}c;
 721        $name;
 722}
 723
 724sub init {
 725        my ($class, $url, $path, $repo_id, $ref_id) = @_;
 726        my $self = _new($class, $repo_id, $ref_id, $path);
 727        mkpath([$self->{dir}]);
 728        if (defined $url) {
 729                $url =~ s!/+$!!; # strip trailing slash
 730                my $orig_url = eval {
 731                        command_oneline('config', '--get',
 732                                        "svn-remote.$repo_id.url")
 733                };
 734                if ($orig_url) {
 735                        if ($orig_url ne $url) {
 736                                die "svn-remote.$repo_id.url already set: ",
 737                                    "$orig_url\nwanted to set to: $url\n";
 738                        }
 739                } else {
 740                        command_noisy('config',
 741                                      "svn-remote.$repo_id.url", $url);
 742                }
 743                command_noisy('config', '--add',
 744                              "svn-remote.$repo_id.fetch",
 745                              "$path:".$self->refname);
 746        }
 747        $self->{url} = $url;
 748        unless (-f $self->{db_path}) {
 749                open my $fh, '>>', $self->{db_path} or croak $!;
 750                close $fh or croak $!;
 751        }
 752        $self;
 753}
 754
 755sub find_ref {
 756        my ($ref_id) = @_;
 757        foreach (command(qw/config -l/)) {
 758                next unless m!^svn-remote\.(.+)\.fetch=
 759                              \s*(.*)\s*:\s*refs/remotes/(.+)\s*$!x;
 760                my ($repo_id, $path, $ref) = ($1, $2, $3);
 761                if ($ref eq $ref_id) {
 762                        $path = '' if ($path =~ m#^\./?#);
 763                        return ($repo_id, $path);
 764                }
 765        }
 766        (undef, undef, undef);
 767}
 768
 769sub new {
 770        my ($class, $ref_id, $repo_id, $path) = @_;
 771        if (defined $ref_id && !defined $repo_id && !defined $path) {
 772                ($repo_id, $path) = find_ref($ref_id);
 773                if (!defined $repo_id) {
 774                        die "Could not find a \"svn-remote.*.fetch\" key ",
 775                            "in the repository configuration matching: ",
 776                            "refs/remotes/$ref_id\n";
 777                }
 778        }
 779        my $self = _new($class, $repo_id, $ref_id, $path);
 780        $self->{url} = command_oneline('config', '--get',
 781                                       "svn-remote.$repo_id.url") or
 782                  die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
 783        $self;
 784}
 785
 786sub refname { "refs/remotes/$_[0]->{ref_id}" }
 787
 788sub ra {
 789        my ($self) = shift;
 790        $self->{ra} ||= Git::SVN::Ra->new($self->{url});
 791}
 792
 793sub copy_remote_ref {
 794        my ($self) = @_;
 795        my $origin = $::_cp_remote ? $::_cp_remote : 'origin';
 796        my $ref = $self->refname;
 797        if (command('ls-remote', $origin, $ref)) {
 798                command_noisy('fetch', $origin, "$ref:$ref");
 799        } elsif ($::_cp_remote && !$::_upgrade) {
 800                die "Unable to find remote reference: $ref on $origin\n";
 801        }
 802}
 803
 804sub traverse_ignore {
 805        my ($self, $fh, $path, $r) = @_;
 806        $path =~ s#^/+##g;
 807        my ($dirent, undef, $props) = $self->ra->get_dir($path, $r);
 808        my $p = $path;
 809        $p =~ s#^\Q$self->{ra}->{svn_path}\E/##;
 810        print $fh length $p ? "\n# $p\n" : "\n# /\n";
 811        if (my $s = $props->{'svn:ignore'}) {
 812                $s =~ s/[\r\n]+/\n/g;
 813                chomp $s;
 814                if (length $p == 0) {
 815                        $s =~ s#\n#\n/$p#g;
 816                        print $fh "/$s\n";
 817                } else {
 818                        $s =~ s#\n#\n/$p/#g;
 819                        print $fh "/$p/$s\n";
 820                }
 821        }
 822        foreach (sort keys %$dirent) {
 823                next if $dirent->{$_}->kind != $SVN::Node::dir;
 824                $self->traverse_ignore($fh, "$path/$_", $r);
 825        }
 826}
 827
 828# returns the newest SVN revision number and newest commit SHA1
 829sub last_rev_commit {
 830        my ($self) = @_;
 831        if (defined $self->{last_rev} && defined $self->{last_commit}) {
 832                return ($self->{last_rev}, $self->{last_commit});
 833        }
 834        my $c = ::verify_ref($self->refname.'^0');
 835        if ($c) {
 836                my $rev = (::cmt_metadata($c))[1];
 837                if (defined $rev) {
 838                        ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
 839                        return ($rev, $c);
 840                }
 841        }
 842        my $offset = -41; # from tail
 843        my $rl;
 844        open my $fh, '<', $self->{db_path} or
 845                                 croak "$self->{db_path} not readable: $!\n";
 846        seek $fh, $offset, 2;
 847        $rl = readline $fh;
 848        defined $rl or return (undef, undef);
 849        chomp $rl;
 850        while ($c ne $rl && tell $fh != 0) {
 851                $offset -= 41;
 852                seek $fh, $offset, 2;
 853                $rl = readline $fh;
 854                defined $rl or return (undef, undef);
 855                chomp $rl;
 856        }
 857        my $rev = tell $fh;
 858        croak $! if ($rev < 0);
 859        $rev =  ($rev - 41) / 41;
 860        close $fh or croak $!;
 861        ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
 862        return ($rev, $c);
 863}
 864
 865sub parse_revision {
 866        my ($self, $base) = @_;
 867        my $head = $self->ra->get_latest_revnum;
 868        if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
 869                return ($base + 1, $head) if (defined $base);
 870                return (0, $head);
 871        }
 872        return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
 873        return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
 874        if ($::_revision =~ /^BASE:(\d+)$/) {
 875                return ($base + 1, $1) if (defined $base);
 876                return (0, $head);
 877        }
 878        return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
 879        die "revision argument: $::_revision not understood by git-svn\n",
 880                "Try using the command-line svn client instead\n";
 881}
 882
 883sub tmp_index_do {
 884        my ($self, $sub) = @_;
 885        my $old_index = $ENV{GIT_INDEX_FILE};
 886        $ENV{GIT_INDEX_FILE} = $self->{index};
 887        my @ret = &$sub;
 888        if ($old_index) {
 889                $ENV{GIT_INDEX_FILE} = $old_index;
 890        } else {
 891                delete $ENV{GIT_INDEX_FILE};
 892        }
 893        wantarray ? @ret : $ret[0];
 894}
 895
 896sub assert_index_clean {
 897        my ($self, $treeish) = @_;
 898
 899        $self->tmp_index_do(sub {
 900                command_noisy('read-tree', $treeish) unless -e $self->{index};
 901                my $x = command_oneline('write-tree');
 902                my ($y) = (command(qw/cat-file commit/, $treeish) =~
 903                           /^tree ($::sha1)/mo);
 904                if ($y ne $x) {
 905                        unlink $self->{index} or croak $!;
 906                        command_noisy('read-tree', $treeish);
 907                }
 908                $x = command_oneline('write-tree');
 909                if ($y ne $x) {
 910                        ::fatal "trees ($treeish) $y != $x\n",
 911                                "Something is seriously wrong...\n";
 912                }
 913        });
 914}
 915
 916sub get_commit_parents {
 917        my ($self, $log_entry, @parents) = @_;
 918        my (%seen, @ret, @tmp);
 919        # commit parents can be conditionally bound to a particular
 920        # svn revision via: "svn_revno=commit_sha1", filter them out here:
 921        foreach my $p (@parents) {
 922                next unless defined $p;
 923                if ($p =~ /^(\d+)=($::sha1_short)$/o) {
 924                        push @tmp, $2 if $1 == $log_entry->{revision};
 925                } else {
 926                        push @tmp, $p if $p =~ /^$::sha1_short$/o;
 927                }
 928        }
 929        if (my $cur = ::verify_ref($self->refname.'^0')) {
 930                push @tmp, $cur;
 931        }
 932        push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
 933        while (my $p = shift @tmp) {
 934                next if $seen{$p};
 935                $seen{$p} = 1;
 936                push @ret, $p;
 937                # MAXPARENT is defined to 16 in commit-tree.c:
 938                last if @ret >= 16;
 939        }
 940        if (@tmp) {
 941                die "r$log_entry->{revision}: No room for parents:\n\t",
 942                    join("\n\t", @tmp), "\n";
 943        }
 944        @ret;
 945}
 946
 947sub full_url {
 948        my ($self) = @_;
 949        $self->ra->{url} . (length $self->{path} ? '/' . $self->{path} : '');
 950}
 951
 952sub do_git_commit {
 953        my ($self, $log_entry, @parents) = @_;
 954        if (my $c = $self->rev_db_get($log_entry->{revision})) {
 955                croak "$log_entry->{revision} = $c already exists! ",
 956                      "Why are we refetching it?\n";
 957        }
 958        my $author = $log_entry->{author};
 959        my ($name, $email) = (defined $::users{$author} ? @{$::users{$author}}
 960                           : ($author, "$author\@".$self->ra->uuid));
 961        $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} = $name;
 962        $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} = $email;
 963        $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
 964
 965        my $tree = $log_entry->{tree};
 966        if (!defined $tree) {
 967                $tree = $self->tmp_index_do(sub {
 968                                            command_oneline('write-tree') });
 969        }
 970        die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
 971
 972        my @exec = ('git-commit-tree', $tree);
 973        foreach ($self->get_commit_parents($log_entry, @parents)) {
 974                push @exec, '-p', $_;
 975        }
 976        defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
 977                                                                   or croak $!;
 978        print $msg_fh $log_entry->{log} or croak $!;
 979        print $msg_fh "\ngit-svn-id: ", $self->full_url, '@',
 980                      $log_entry->{revision}, ' ',
 981                      $self->ra->uuid, "\n" or croak $!;
 982        $msg_fh->flush == 0 or croak $!;
 983        close $msg_fh or croak $!;
 984        chomp(my $commit = do { local $/; <$out_fh> });
 985        close $out_fh or croak $!;
 986        waitpid $pid, 0;
 987        croak $? if $?;
 988        if ($commit !~ /^$::sha1$/o) {
 989                die "Failed to commit, invalid sha1: $commit\n";
 990        }
 991
 992        command_noisy('update-ref',$self->refname, $commit);
 993        $self->rev_db_set($log_entry->{revision}, $commit);
 994
 995        $self->{last_rev} = $log_entry->{revision};
 996        $self->{last_commit} = $commit;
 997        print "r$log_entry->{revision} = $commit\n";
 998        return $commit;
 999}
1000
1001sub do_fetch {
1002        my ($self, $paths, $rev) = @_;
1003        my $ed = SVN::Git::Fetcher->new($self);
1004        my ($last_rev, @parents);
1005        if ($self->{last_commit}) {
1006                $last_rev = $self->{last_rev};
1007                $ed->{c} = $self->{last_commit};
1008                @parents = ($self->{last_commit});
1009        } else {
1010                $last_rev = $rev;
1011        }
1012        unless ($self->ra->gs_do_update($last_rev, $rev,
1013                                        $self->{path}, 1, $ed)) {
1014                die "SVN connection failed somewhere...\n";
1015        }
1016        $self->make_log_entry($rev, \@parents, $ed);
1017}
1018
1019sub write_untracked {
1020        my ($self, $rev, $fh, $untracked) = @_;
1021        my $h;
1022        print $fh "r$rev\n" or croak $!;
1023        $h = $untracked->{empty};
1024        foreach (sort keys %$h) {
1025                my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
1026                print $fh "  $act: ", uri_encode($_), "\n" or croak $!;
1027                warn "W: $act: $_\n";
1028        }
1029        foreach my $t (qw/dir_prop file_prop/) {
1030                $h = $untracked->{$t} or next;
1031                foreach my $path (sort keys %$h) {
1032                        my $ppath = $path eq '' ? '.' : $path;
1033                        foreach my $prop (sort keys %{$h->{$path}}) {
1034                                next if $SKIP_PROP{$prop};
1035                                my $v = $h->{$path}->{$prop};
1036                                if (defined $v) {
1037                                        print $fh "  +$t: ",
1038                                                  uri_encode($ppath), ' ',
1039                                                  uri_encode($prop), ' ',
1040                                                  uri_encode($v), "\n"
1041                                                  or croak $!;
1042                                } else {
1043                                        print $fh "  -$t: ",
1044                                                  uri_encode($ppath), ' ',
1045                                                  uri_encode($prop), "\n"
1046                                                  or croak $!;
1047                                }
1048                        }
1049                }
1050        }
1051        foreach my $t (qw/absent_file absent_directory/) {
1052                $h = $untracked->{$t} or next;
1053                foreach my $parent (sort keys %$h) {
1054                        foreach my $path (sort @{$h->{$parent}}) {
1055                                print $fh "  $t: ",
1056                                      uri_encode("$parent/$path"), "\n"
1057                                      or croak $!;
1058                                warn "W: $t: $parent/$path ",
1059                                     "Insufficient permissions?\n";
1060                        }
1061                }
1062        }
1063}
1064
1065sub parse_svn_date {
1066        my $date = shift || return '+0000 1970-01-01 00:00:00';
1067        my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
1068                                            (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x) or
1069                                         croak "Unable to parse date: $date\n";
1070        "+0000 $Y-$m-$d $H:$M:$S";
1071}
1072
1073sub check_author {
1074        my ($author) = @_;
1075        if (!defined $author || length $author == 0) {
1076                $author = '(no author)';
1077        }
1078        if (defined $::_authors && ! defined $::users{$author}) {
1079                die "Author: $author not defined in $::_authors file\n";
1080        }
1081        $author;
1082}
1083
1084sub make_log_entry {
1085        my ($self, $rev, $parents, $untracked) = @_;
1086        my $rp = $self->ra->rev_proplist($rev);
1087        my %log_entry = ( parents => $parents || [], revision => $rev,
1088                          revprops => $rp, log => '');
1089        open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
1090        $self->write_untracked($rev, $un, $untracked);
1091        foreach (sort keys %$rp) {
1092                my $v = $rp->{$_};
1093                if (/^svn:(author|date|log)$/) {
1094                        $log_entry{$1} = $v;
1095                } else {
1096                        print $un "  rev_prop: ", uri_encode($_), ' ',
1097                                  uri_encode($v), "\n";
1098                }
1099        }
1100        close $un or croak $!;
1101        $log_entry{date} = parse_svn_date($log_entry{date});
1102        $log_entry{author} = check_author($log_entry{author});
1103        $log_entry{log} .= "\n";
1104        \%log_entry;
1105}
1106
1107sub fetch {
1108        my ($self, @parents) = @_;
1109        my ($last_rev, $last_commit) = $self->last_rev_commit;
1110        my ($base, $head) = $self->parse_revision($last_rev);
1111        return if ($base > $head);
1112        if (defined $last_commit) {
1113                $self->assert_index_clean($last_commit);
1114        }
1115        my $inc = 1000;
1116        my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
1117        my $err_handler = $SVN::Error::handler;
1118        $SVN::Error::handler = \&skip_unknown_revs;
1119        while (1) {
1120                my @revs;
1121                $self->ra->get_log([''], $min, $max, 0, 1, 1, sub {
1122                        my ($paths, $rev, $author, $date, $log) = @_;
1123                        push @revs, $rev });
1124                foreach (@revs) {
1125                        my $log_entry = $self->do_fetch(undef, $_);
1126                        $self->do_git_commit($log_entry, @parents);
1127                }
1128                last if $max >= $head;
1129                $min = $max + 1;
1130                $max += $inc;
1131                $max = $head if ($max > $head);
1132        }
1133        $SVN::Error::handler = $err_handler;
1134}
1135
1136sub set_tree_cb {
1137        my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
1138        # TODO: enable and test optimized commits:
1139        if (0 && $rev == ($self->{last_rev} + 1)) {
1140                $log_entry->{revision} = $rev;
1141                $log_entry->{author} = $author;
1142                $self->do_git_commit($log_entry, "$rev=$tree");
1143        } else {
1144                $self->fetch("$rev=$tree");
1145        }
1146}
1147
1148sub set_tree {
1149        my ($self, $tree) = (shift, shift);
1150        my $log_entry = ::get_commit_entry($tree);
1151        unless ($self->{last_rev}) {
1152                fatal("Must have an existing revision to commit\n");
1153        }
1154        my $pool = SVN::Pool->new;
1155        my $ed = SVN::Git::Editor->new({ r => $self->{last_rev},
1156                                         ra => $self->ra->dup,
1157                                         svn_path => $self->ra->{svn_path}
1158                                       },
1159                                       $self->ra->get_commit_editor(
1160                                         $log_entry->{log}, sub {
1161                                           $self->set_tree_cb($log_entry,
1162                                                              $tree, @_);
1163                                       }),
1164                                       $pool);
1165        my $mods = $ed->apply_diff($self->{last_commit}, $tree);
1166        if (@$mods == 0) {
1167                print "No changes\nr$self->{last_rev} = $tree\n";
1168        }
1169        $pool->clear;
1170}
1171
1172sub skip_unknown_revs {
1173        my ($err) = @_;
1174        my $errno = $err->apr_err();
1175        # Maybe the branch we're tracking didn't
1176        # exist when the repo started, so it's
1177        # not an error if it doesn't, just continue
1178        #
1179        # Wonderfully consistent library, eh?
1180        # 160013 - svn:// and file://
1181        # 175002 - http(s)://
1182        # 175007 - http(s):// (this repo required authorization, too...)
1183        #   More codes may be discovered later...
1184        if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
1185                return;
1186        }
1187        croak "Error from SVN, ($errno): ", $err->expanded_message,"\n";
1188}
1189
1190# rev_db:
1191# Tie::File seems to be prone to offset errors if revisions get sparse,
1192# it's not that fast, either.  Tie::File is also not in Perl 5.6.  So
1193# one of my favorite modules is out :<  Next up would be one of the DBM
1194# modules, but I'm not sure which is most portable...  So I'll just
1195# go with something that's plain-text, but still capable of
1196# being randomly accessed.  So here's my ultra-simple fixed-width
1197# database.  All records are 40 characters + "\n", so it's easy to seek
1198# to a revision: (41 * rev) is the byte offset.
1199# A record of 40 0s denotes an empty revision.
1200# And yes, it's still pretty fast (faster than Tie::File).
1201
1202sub rev_db_set {
1203        my ($self, $rev, $commit) = @_;
1204        length $commit == 40 or croak "arg3 must be a full SHA1 hexsum\n";
1205        open my $fh, '+<', $self->{db_path} or croak $!;
1206        my $offset = $rev * 41;
1207        # assume that append is the common case:
1208        seek $fh, 0, 2 or croak $!;
1209        my $pos = tell $fh;
1210        if ($pos < $offset) {
1211                print $fh (('0' x 40),"\n") x (($offset - $pos) / 41)
1212                  or croak $!;
1213        }
1214        seek $fh, $offset, 0 or croak $!;
1215        print $fh $commit,"\n" or croak $!;
1216        close $fh or croak $!;
1217}
1218
1219sub rev_db_get {
1220        my ($self, $rev) = @_;
1221        my $ret;
1222        my $offset = $rev * 41;
1223        open my $fh, '<', $self->{db_path} or croak $!;
1224        if (seek $fh, $offset, 0) {
1225                $ret = readline $fh;
1226                if (defined $ret) {
1227                        chomp $ret;
1228                        $ret = undef if ($ret =~ /^0{40}$/);
1229                }
1230        }
1231        close $fh or croak $!;
1232        $ret;
1233}
1234
1235sub _new {
1236        my ($class, $repo_id, $ref_id, $path) = @_;
1237        unless (defined $repo_id && length $repo_id) {
1238                $repo_id = $Git::SVN::default_repo_id;
1239        }
1240        unless (defined $ref_id && length $ref_id) {
1241                $_[2] = $ref_id = $repo_id;
1242        }
1243        $_[1] = $repo_id = sanitize_remote_name($repo_id);
1244        my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
1245        $_[3] = $path = '' unless (defined $path);
1246        bless { ref_id => $ref_id, dir => $dir, index => "$dir/index",
1247                path => $path,
1248                db_path => "$dir/.rev_db", repo_id => $repo_id }, $class;
1249}
1250
1251sub uri_encode {
1252        my ($f) = @_;
1253        $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
1254        $f
1255}
1256
1257package Git::SVN::Prompt;
1258use strict;
1259use warnings;
1260require SVN::Core;
1261use vars qw/$_no_auth_cache $_username/;
1262
1263sub simple {
1264        my ($cred, $realm, $default_username, $may_save, $pool) = @_;
1265        $may_save = undef if $_no_auth_cache;
1266        $default_username = $_username if defined $_username;
1267        if (defined $default_username && length $default_username) {
1268                if (defined $realm && length $realm) {
1269                        print STDERR "Authentication realm: $realm\n";
1270                        STDERR->flush;
1271                }
1272                $cred->username($default_username);
1273        } else {
1274                username($cred, $realm, $may_save, $pool);
1275        }
1276        $cred->password(_read_password("Password for '" .
1277                                       $cred->username . "': ", $realm));
1278        $cred->may_save($may_save);
1279        $SVN::_Core::SVN_NO_ERROR;
1280}
1281
1282sub ssl_server_trust {
1283        my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
1284        $may_save = undef if $_no_auth_cache;
1285        print STDERR "Error validating server certificate for '$realm':\n";
1286        if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
1287                print STDERR " - The certificate is not issued by a trusted ",
1288                      "authority. Use the\n",
1289                      "   fingerprint to validate the certificate manually!\n";
1290        }
1291        if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
1292                print STDERR " - The certificate hostname does not match.\n";
1293        }
1294        if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
1295                print STDERR " - The certificate is not yet valid.\n";
1296        }
1297        if ($failures & $SVN::Auth::SSL::EXPIRED) {
1298                print STDERR " - The certificate has expired.\n";
1299        }
1300        if ($failures & $SVN::Auth::SSL::OTHER) {
1301                print STDERR " - The certificate has an unknown error.\n";
1302        }
1303        printf STDERR
1304                "Certificate information:\n".
1305                " - Hostname: %s\n".
1306                " - Valid: from %s until %s\n".
1307                " - Issuer: %s\n".
1308                " - Fingerprint: %s\n",
1309                map $cert_info->$_, qw(hostname valid_from valid_until
1310                                       issuer_dname fingerprint);
1311        my $choice;
1312prompt:
1313        print STDERR $may_save ?
1314              "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
1315              "(R)eject or accept (t)emporarily? ";
1316        STDERR->flush;
1317        $choice = lc(substr(<STDIN> || 'R', 0, 1));
1318        if ($choice =~ /^t$/i) {
1319                $cred->may_save(undef);
1320        } elsif ($choice =~ /^r$/i) {
1321                return -1;
1322        } elsif ($may_save && $choice =~ /^p$/i) {
1323                $cred->may_save($may_save);
1324        } else {
1325                goto prompt;
1326        }
1327        $cred->accepted_failures($failures);
1328        $SVN::_Core::SVN_NO_ERROR;
1329}
1330
1331sub ssl_client_cert {
1332        my ($cred, $realm, $may_save, $pool) = @_;
1333        $may_save = undef if $_no_auth_cache;
1334        print STDERR "Client certificate filename: ";
1335        STDERR->flush;
1336        chomp(my $filename = <STDIN>);
1337        $cred->cert_file($filename);
1338        $cred->may_save($may_save);
1339        $SVN::_Core::SVN_NO_ERROR;
1340}
1341
1342sub ssl_client_cert_pw {
1343        my ($cred, $realm, $may_save, $pool) = @_;
1344        $may_save = undef if $_no_auth_cache;
1345        $cred->password(_read_password("Password: ", $realm));
1346        $cred->may_save($may_save);
1347        $SVN::_Core::SVN_NO_ERROR;
1348}
1349
1350sub username {
1351        my ($cred, $realm, $may_save, $pool) = @_;
1352        $may_save = undef if $_no_auth_cache;
1353        if (defined $realm && length $realm) {
1354                print STDERR "Authentication realm: $realm\n";
1355        }
1356        my $username;
1357        if (defined $_username) {
1358                $username = $_username;
1359        } else {
1360                print STDERR "Username: ";
1361                STDERR->flush;
1362                chomp($username = <STDIN>);
1363        }
1364        $cred->username($username);
1365        $cred->may_save($may_save);
1366        $SVN::_Core::SVN_NO_ERROR;
1367}
1368
1369sub _read_password {
1370        my ($prompt, $realm) = @_;
1371        print STDERR $prompt;
1372        STDERR->flush;
1373        require Term::ReadKey;
1374        Term::ReadKey::ReadMode('noecho');
1375        my $password = '';
1376        while (defined(my $key = Term::ReadKey::ReadKey(0))) {
1377                last if $key =~ /[\012\015]/; # \n\r
1378                $password .= $key;
1379        }
1380        Term::ReadKey::ReadMode('restore');
1381        print STDERR "\n";
1382        STDERR->flush;
1383        $password;
1384}
1385
1386package main;
1387
1388sub uri_encode {
1389        my ($f) = @_;
1390        $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
1391        $f
1392}
1393
1394sub uri_decode {
1395        my ($f) = @_;
1396        $f =~ tr/+/ /;
1397        $f =~ s/%([A-F0-9]{2})/chr hex($1)/ge;
1398        $f
1399}
1400
1401sub revisions_eq {
1402        my ($path, $r0, $r1) = @_;
1403        return 1 if $r0 == $r1;
1404        my $nr = 0;
1405        # should be OK to use Pool here (r1 - r0) should be small
1406        $SVN->get_log([$path], $r0, $r1, 0, 0, 1, sub {$nr++});
1407        return 0 if ($nr > 1);
1408        return 1;
1409}
1410
1411sub libsvn_find_parent_branch {
1412        my ($paths, $rev, $author, $date, $log) = @_;
1413        my $svn_path = '/'.$SVN->{svn_path};
1414
1415        # look for a parent from another branch:
1416        my $i = $paths->{$svn_path} or return;
1417        my $branch_from = $i->copyfrom_path or return;
1418        my $r = $i->copyfrom_rev;
1419        print STDERR  "Found possible branch point: ",
1420                                "$branch_from => $svn_path, $r\n";
1421        $branch_from =~ s#^/##;
1422        my $l_map = {};
1423        read_url_paths_all($l_map, '', "$GIT_DIR/svn");
1424        my $url = $SVN->{repos_root};
1425        defined $l_map->{$url} or return;
1426        my $id = $l_map->{$url}->{$branch_from};
1427        if (!defined $id && $_follow_parent) {
1428                print STDERR "Following parent: $branch_from\@$r\n";
1429                # auto create a new branch and follow it
1430                $id = basename($branch_from);
1431                $id .= '@'.$r if -r "$GIT_DIR/svn/$id";
1432                while (-r "$GIT_DIR/svn/$id") {
1433                        # just grow a tail if we're not unique enough :x
1434                        $id .= '-';
1435                }
1436        }
1437        return unless defined $id;
1438
1439        my ($r0, $parent) = find_rev_before($r,$id,1);
1440        if ($_follow_parent && (!defined $r0 || !defined $parent)) {
1441                defined(my $pid = fork) or croak $!;
1442                if (!$pid) {
1443                        $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
1444                        init_vars();
1445                        $SVN_URL = "$url/$branch_from";
1446                        $SVN = undef;
1447                        setup_git_svn();
1448                        # we can't assume SVN_URL exists at r+1:
1449                        $_revision = "0:$r";
1450                        fetch_lib();
1451                        exit 0;
1452                }
1453                waitpid $pid, 0;
1454                croak $? if $?;
1455                ($r0, $parent) = find_rev_before($r,$id,1);
1456        }
1457        return unless (defined $r0 && defined $parent);
1458        if (revisions_eq($branch_from, $r0, $r)) {
1459                unlink $GIT_SVN_INDEX;
1460                print STDERR "Found branch parent: ($GIT_SVN) $parent\n";
1461                command_noisy('read-tree', $parent);
1462                unless ($SVN->can_do_switch) {
1463                        return _libsvn_new_tree($paths, $rev, $author, $date,
1464                                                $log, [$parent]);
1465                }
1466                # do_switch works with svn/trunk >= r22312, but that is not
1467                # included with SVN 1.4.2 (the latest version at the moment),
1468                # so we can't rely on it.
1469                my $ra = Git::SVN::Ra->new("$url/$branch_from");
1470                my $ed = SVN::Git::Fetcher->new({c => $parent, q => $_q });
1471                $ra->gs_do_switch($r0, $rev, '', 1, $SVN->{url}, $ed) or
1472                                   die "SVN connection failed somewhere...\n";
1473                return libsvn_log_entry($rev, $author, $date, $log, [$parent]);
1474        }
1475        print STDERR "Nope, branch point not imported or unknown\n";
1476        return undef;
1477}
1478
1479sub _libsvn_new_tree {
1480        my ($paths, $rev, $author, $date, $log, $parents) = @_;
1481        my $ed = SVN::Git::Fetcher->new({q => $_q});
1482        unless ($SVN->gs_do_update($rev, $rev, '', 1, $ed)) {
1483                die "SVN connection failed somewhere...\n";
1484        }
1485        libsvn_log_entry($rev, $author, $date, $log, $parents, $ed);
1486}
1487
1488{
1489        my $kill_stupid_warnings = $SVN::Node::none.$SVN::Node::file.
1490                                $SVN::Node::dir.$SVN::Node::unknown.
1491                                $SVN::Node::none.$SVN::Node::file.
1492                                $SVN::Node::dir.$SVN::Node::unknown.
1493                                $SVN::Auth::SSL::CNMISMATCH.
1494                                $SVN::Auth::SSL::NOTYETVALID.
1495                                $SVN::Auth::SSL::EXPIRED.
1496                                $SVN::Auth::SSL::UNKNOWNCA.
1497                                $SVN::Auth::SSL::OTHER;
1498}
1499
1500package SVN::Git::Fetcher;
1501use vars qw/@ISA/;
1502use strict;
1503use warnings;
1504use Carp qw/croak/;
1505use IO::File qw//;
1506
1507# file baton members: path, mode_a, mode_b, pool, fh, blob, base
1508sub new {
1509        my ($class, $git_svn) = @_;
1510        my $self = SVN::Delta::Editor->new;
1511        bless $self, $class;
1512        $self->{c} = $git_svn->{last_commit} if exists $git_svn->{last_commit};
1513        if (length $git_svn->{path}) {
1514                $self->{path_strip} = qr/\Q$git_svn->{path}\E\/?/;
1515        }
1516        $self->{empty} = {};
1517        $self->{dir_prop} = {};
1518        $self->{file_prop} = {};
1519        $self->{absent_dir} = {};
1520        $self->{absent_file} = {};
1521        ($self->{gui}, $self->{ctx}) = $git_svn->tmp_index_do(
1522               sub { command_input_pipe(qw/update-index -z --index-info/) } );
1523        require Digest::MD5;
1524        $self;
1525}
1526
1527sub open_root {
1528        { path => '' };
1529}
1530
1531sub open_directory {
1532        my ($self, $path, $pb, $rev) = @_;
1533        { path => $path };
1534}
1535
1536sub git_path {
1537        my ($self, $path) = @_;
1538        $path =~ s!$self->{path_strip}!! if $self->{path_strip};
1539        $path;
1540}
1541
1542sub delete_entry {
1543        my ($self, $path, $rev, $pb) = @_;
1544        my $gui = $self->{gui};
1545
1546        my $gpath = $self->git_path($path);
1547        # remove entire directories.
1548        if (command('ls-tree', $self->{c}, '--', $gpath) =~ /^040000 tree/) {
1549                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
1550                                                     -r --name-only -z/,
1551                                                     $self->{c}, '--', $gpath);
1552                local $/ = "\0";
1553                while (<$ls>) {
1554                        print $gui '0 ',0 x 40,"\t",$_ or croak $!;
1555                        print "\tD\t$_\n" unless $self->{q};
1556                }
1557                print "\tD\t$gpath/\n" unless $self->{q};
1558                command_close_pipe($ls, $ctx);
1559                $self->{empty}->{$path} = 0
1560        } else {
1561                print $gui '0 ',0 x 40,"\t",$gpath,"\0" or croak $!;
1562                print "\tD\t$gpath\n" unless $self->{q};
1563        }
1564        undef;
1565}
1566
1567sub open_file {
1568        my ($self, $path, $pb, $rev) = @_;
1569        my $gpath = $self->git_path($path);
1570        my ($mode, $blob) = (command('ls-tree', $self->{c}, '--', $gpath)
1571                             =~ /^(\d{6}) blob ([a-f\d]{40})\t/);
1572        unless (defined $mode && defined $blob) {
1573                die "$path was not found in commit $self->{c} (r$rev)\n";
1574        }
1575        { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
1576          pool => SVN::Pool->new, action => 'M' };
1577}
1578
1579sub add_file {
1580        my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
1581        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
1582        delete $self->{empty}->{$dir};
1583        { path => $path, mode_a => 100644, mode_b => 100644,
1584          pool => SVN::Pool->new, action => 'A' };
1585}
1586
1587sub add_directory {
1588        my ($self, $path, $cp_path, $cp_rev) = @_;
1589        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
1590        delete $self->{empty}->{$dir};
1591        $self->{empty}->{$path} = 1;
1592        { path => $path };
1593}
1594
1595sub change_dir_prop {
1596        my ($self, $db, $prop, $value) = @_;
1597        $self->{dir_prop}->{$db->{path}} ||= {};
1598        $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
1599        undef;
1600}
1601
1602sub absent_directory {
1603        my ($self, $path, $pb) = @_;
1604        $self->{absent_dir}->{$pb->{path}} ||= [];
1605        push @{$self->{absent_dir}->{$pb->{path}}}, $path;
1606        undef;
1607}
1608
1609sub absent_file {
1610        my ($self, $path, $pb) = @_;
1611        $self->{absent_file}->{$pb->{path}} ||= [];
1612        push @{$self->{absent_file}->{$pb->{path}}}, $path;
1613        undef;
1614}
1615
1616sub change_file_prop {
1617        my ($self, $fb, $prop, $value) = @_;
1618        if ($prop eq 'svn:executable') {
1619                if ($fb->{mode_b} != 120000) {
1620                        $fb->{mode_b} = defined $value ? 100755 : 100644;
1621                }
1622        } elsif ($prop eq 'svn:special') {
1623                $fb->{mode_b} = defined $value ? 120000 : 100644;
1624        } else {
1625                $self->{file_prop}->{$fb->{path}} ||= {};
1626                $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
1627        }
1628        undef;
1629}
1630
1631sub apply_textdelta {
1632        my ($self, $fb, $exp) = @_;
1633        my $fh = IO::File->new_tmpfile;
1634        $fh->autoflush(1);
1635        # $fh gets auto-closed() by SVN::TxDelta::apply(),
1636        # (but $base does not,) so dup() it for reading in close_file
1637        open my $dup, '<&', $fh or croak $!;
1638        my $base = IO::File->new_tmpfile;
1639        $base->autoflush(1);
1640        if ($fb->{blob}) {
1641                defined (my $pid = fork) or croak $!;
1642                if (!$pid) {
1643                        open STDOUT, '>&', $base or croak $!;
1644                        print STDOUT 'link ' if ($fb->{mode_a} == 120000);
1645                        exec qw/git-cat-file blob/, $fb->{blob} or croak $!;
1646                }
1647                waitpid $pid, 0;
1648                croak $? if $?;
1649
1650                if (defined $exp) {
1651                        seek $base, 0, 0 or croak $!;
1652                        my $md5 = Digest::MD5->new;
1653                        $md5->addfile($base);
1654                        my $got = $md5->hexdigest;
1655                        die "Checksum mismatch: $fb->{path} $fb->{blob}\n",
1656                            "expected: $exp\n",
1657                            "     got: $got\n" if ($got ne $exp);
1658                }
1659        }
1660        seek $base, 0, 0 or croak $!;
1661        $fb->{fh} = $dup;
1662        $fb->{base} = $base;
1663        [ SVN::TxDelta::apply($base, $fh, undef, $fb->{path}, $fb->{pool}) ];
1664}
1665
1666sub close_file {
1667        my ($self, $fb, $exp) = @_;
1668        my $hash;
1669        my $path = $self->git_path($fb->{path});
1670        if (my $fh = $fb->{fh}) {
1671                seek($fh, 0, 0) or croak $!;
1672                my $md5 = Digest::MD5->new;
1673                $md5->addfile($fh);
1674                my $got = $md5->hexdigest;
1675                die "Checksum mismatch: $path\n",
1676                    "expected: $exp\n    got: $got\n" if ($got ne $exp);
1677                seek($fh, 0, 0) or croak $!;
1678                if ($fb->{mode_b} == 120000) {
1679                        read($fh, my $buf, 5) == 5 or croak $!;
1680                        $buf eq 'link ' or die "$path has mode 120000",
1681                                               "but is not a link\n";
1682                }
1683                defined(my $pid = open my $out,'-|') or die "Can't fork: $!\n";
1684                if (!$pid) {
1685                        open STDIN, '<&', $fh or croak $!;
1686                        exec qw/git-hash-object -w --stdin/ or croak $!;
1687                }
1688                chomp($hash = do { local $/; <$out> });
1689                close $out or croak $!;
1690                close $fh or croak $!;
1691                $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
1692                close $fb->{base} or croak $!;
1693        } else {
1694                $hash = $fb->{blob} or die "no blob information\n";
1695        }
1696        $fb->{pool}->clear;
1697        my $gui = $self->{gui};
1698        print $gui "$fb->{mode_b} $hash\t$path\0" or croak $!;
1699        print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $self->{q};
1700        undef;
1701}
1702
1703sub abort_edit {
1704        my $self = shift;
1705        eval { command_close_pipe($self->{gui}, $self->{ctx}) };
1706        $self->SUPER::abort_edit(@_);
1707}
1708
1709sub close_edit {
1710        my $self = shift;
1711        command_close_pipe($self->{gui}, $self->{ctx});
1712        $self->{git_commit_ok} = 1;
1713        $self->SUPER::close_edit(@_);
1714}
1715
1716package SVN::Git::Editor;
1717use vars qw/@ISA/;
1718use strict;
1719use warnings;
1720use Carp qw/croak/;
1721use IO::File;
1722
1723sub new {
1724        my $class = shift;
1725        my $git_svn = shift;
1726        my $self = SVN::Delta::Editor->new(@_);
1727        bless $self, $class;
1728        foreach (qw/svn_path r ra/) {
1729                die "$_ required!\n" unless (defined $git_svn->{$_});
1730                $self->{$_} = $git_svn->{$_};
1731        }
1732        $self->{pool} = SVN::Pool->new;
1733        $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
1734        $self->{rm} = { };
1735        require Digest::MD5;
1736        return $self;
1737}
1738
1739sub split_path {
1740        return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
1741}
1742
1743sub repo_path {
1744        (defined $_[1] && length $_[1]) ? $_[1] : ''
1745}
1746
1747sub url_path {
1748        my ($self, $path) = @_;
1749        $self->{ra}->{url} . '/' . $self->repo_path($path);
1750}
1751
1752sub rmdirs {
1753        my ($self, $tree_b) = @_;
1754        my $rm = $self->{rm};
1755        delete $rm->{''}; # we never delete the url we're tracking
1756        return unless %$rm;
1757
1758        foreach (keys %$rm) {
1759                my @d = split m#/#, $_;
1760                my $c = shift @d;
1761                $rm->{$c} = 1;
1762                while (@d) {
1763                        $c .= '/' . shift @d;
1764                        $rm->{$c} = 1;
1765                }
1766        }
1767        delete $rm->{$self->{svn_path}};
1768        delete $rm->{''}; # we never delete the url we're tracking
1769        return unless %$rm;
1770
1771        my ($fh, $ctx) = command_output_pipe(
1772                                   qw/ls-tree --name-only -r -z/, $tree_b);
1773        local $/ = "\0";
1774        while (<$fh>) {
1775                chomp;
1776                my @dn = split m#/#, $_;
1777                while (pop @dn) {
1778                        delete $rm->{join '/', @dn};
1779                }
1780                unless (%$rm) {
1781                        close $fh;
1782                        return;
1783                }
1784        }
1785        command_close_pipe($fh, $ctx);
1786
1787        my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
1788        foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
1789                $self->close_directory($bat->{$d}, $p);
1790                my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
1791                print "\tD+\t$d/\n" unless $::_q;
1792                $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
1793                delete $bat->{$d};
1794        }
1795}
1796
1797sub open_or_add_dir {
1798        my ($self, $full_path, $baton) = @_;
1799        my $t = $self->{ra}->check_path($full_path, $self->{r});
1800        if ($t == $SVN::Node::none) {
1801                return $self->add_directory($full_path, $baton,
1802                                                undef, -1, $self->{pool});
1803        } elsif ($t == $SVN::Node::dir) {
1804                return $self->open_directory($full_path, $baton,
1805                                                $self->{r}, $self->{pool});
1806        }
1807        print STDERR "$full_path already exists in repository at ",
1808                "r$self->{r} and it is not a directory (",
1809                ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
1810        exit 1;
1811}
1812
1813sub ensure_path {
1814        my ($self, $path) = @_;
1815        my $bat = $self->{bat};
1816        $path = $self->repo_path($path);
1817        return $bat->{''} unless (length $path);
1818        my @p = split m#/+#, $path;
1819        my $c = shift @p;
1820        $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
1821        while (@p) {
1822                my $c0 = $c;
1823                $c .= '/' . shift @p;
1824                $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
1825        }
1826        return $bat->{$c};
1827}
1828
1829sub A {
1830        my ($self, $m) = @_;
1831        my ($dir, $file) = split_path($m->{file_b});
1832        my $pbat = $self->ensure_path($dir);
1833        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
1834                                        undef, -1);
1835        print "\tA\t$m->{file_b}\n" unless $::_q;
1836        $self->chg_file($fbat, $m);
1837        $self->close_file($fbat,undef,$self->{pool});
1838}
1839
1840sub C {
1841        my ($self, $m) = @_;
1842        my ($dir, $file) = split_path($m->{file_b});
1843        my $pbat = $self->ensure_path($dir);
1844        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
1845                                $self->url_path($m->{file_a}), $self->{r});
1846        print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
1847        $self->chg_file($fbat, $m);
1848        $self->close_file($fbat,undef,$self->{pool});
1849}
1850
1851sub delete_entry {
1852        my ($self, $path, $pbat) = @_;
1853        my $rpath = $self->repo_path($path);
1854        my ($dir, $file) = split_path($rpath);
1855        $self->{rm}->{$dir} = 1;
1856        $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
1857}
1858
1859sub R {
1860        my ($self, $m) = @_;
1861        my ($dir, $file) = split_path($m->{file_b});
1862        my $pbat = $self->ensure_path($dir);
1863        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
1864                                $self->url_path($m->{file_a}), $self->{r});
1865        print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
1866        $self->chg_file($fbat, $m);
1867        $self->close_file($fbat,undef,$self->{pool});
1868
1869        ($dir, $file) = split_path($m->{file_a});
1870        $pbat = $self->ensure_path($dir);
1871        $self->delete_entry($m->{file_a}, $pbat);
1872}
1873
1874sub M {
1875        my ($self, $m) = @_;
1876        my ($dir, $file) = split_path($m->{file_b});
1877        my $pbat = $self->ensure_path($dir);
1878        my $fbat = $self->open_file($self->repo_path($m->{file_b}),
1879                                $pbat,$self->{r},$self->{pool});
1880        print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
1881        $self->chg_file($fbat, $m);
1882        $self->close_file($fbat,undef,$self->{pool});
1883}
1884
1885sub T { shift->M(@_) }
1886
1887sub change_file_prop {
1888        my ($self, $fbat, $pname, $pval) = @_;
1889        $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
1890}
1891
1892sub chg_file {
1893        my ($self, $fbat, $m) = @_;
1894        if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
1895                $self->change_file_prop($fbat,'svn:executable','*');
1896        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
1897                $self->change_file_prop($fbat,'svn:executable',undef);
1898        }
1899        my $fh = IO::File->new_tmpfile or croak $!;
1900        if ($m->{mode_b} =~ /^120/) {
1901                print $fh 'link ' or croak $!;
1902                $self->change_file_prop($fbat,'svn:special','*');
1903        } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
1904                $self->change_file_prop($fbat,'svn:special',undef);
1905        }
1906        defined(my $pid = fork) or croak $!;
1907        if (!$pid) {
1908                open STDOUT, '>&', $fh or croak $!;
1909                exec qw/git-cat-file blob/, $m->{sha1_b} or croak $!;
1910        }
1911        waitpid $pid, 0;
1912        croak $? if $?;
1913        $fh->flush == 0 or croak $!;
1914        seek $fh, 0, 0 or croak $!;
1915
1916        my $md5 = Digest::MD5->new;
1917        $md5->addfile($fh) or croak $!;
1918        seek $fh, 0, 0 or croak $!;
1919
1920        my $exp = $md5->hexdigest;
1921        my $pool = SVN::Pool->new;
1922        my $atd = $self->apply_textdelta($fbat, undef, $pool);
1923        my $got = SVN::TxDelta::send_stream($fh, @$atd, $pool);
1924        die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
1925        $pool->clear;
1926
1927        close $fh or croak $!;
1928}
1929
1930sub D {
1931        my ($self, $m) = @_;
1932        my ($dir, $file) = split_path($m->{file_b});
1933        my $pbat = $self->ensure_path($dir);
1934        print "\tD\t$m->{file_b}\n" unless $::_q;
1935        $self->delete_entry($m->{file_b}, $pbat);
1936}
1937
1938sub close_edit {
1939        my ($self) = @_;
1940        my ($p,$bat) = ($self->{pool}, $self->{bat});
1941        foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
1942                $self->close_directory($bat->{$_}, $p);
1943        }
1944        $self->SUPER::close_edit($p);
1945        $p->clear;
1946}
1947
1948sub abort_edit {
1949        my ($self) = @_;
1950        $self->SUPER::abort_edit($self->{pool});
1951        $self->{pool}->clear;
1952}
1953
1954# this drives the editor
1955sub apply_diff {
1956        my ($self, $tree_a, $tree_b) = @_;
1957        my @diff_tree = qw(diff-tree -z -r);
1958        if ($::_cp_similarity) {
1959                push @diff_tree, "-C$::_cp_similarity";
1960        } else {
1961                push @diff_tree, '-C';
1962        }
1963        push @diff_tree, '--find-copies-harder' if $::_find_copies_harder;
1964        push @diff_tree, "-l$::_l" if defined $::_l;
1965        push @diff_tree, $tree_a, $tree_b;
1966        my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
1967        my $nl = $/;
1968        local $/ = "\0";
1969        my $state = 'meta';
1970        my @mods;
1971        while (<$diff_fh>) {
1972                chomp $_; # this gets rid of the trailing "\0"
1973                if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
1974                                        $::sha1\s($::sha1)\s
1975                                        ([MTCRAD])\d*$/xo) {
1976                        push @mods, {   mode_a => $1, mode_b => $2,
1977                                        sha1_b => $3, chg => $4 };
1978                        if ($4 =~ /^(?:C|R)$/) {
1979                                $state = 'file_a';
1980                        } else {
1981                                $state = 'file_b';
1982                        }
1983                } elsif ($state eq 'file_a') {
1984                        my $x = $mods[$#mods] or croak "Empty array\n";
1985                        if ($x->{chg} !~ /^(?:C|R)$/) {
1986                                croak "Error parsing $_, $x->{chg}\n";
1987                        }
1988                        $x->{file_a} = $_;
1989                        $state = 'file_b';
1990                } elsif ($state eq 'file_b') {
1991                        my $x = $mods[$#mods] or croak "Empty array\n";
1992                        if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
1993                                croak "Error parsing $_, $x->{chg}\n";
1994                        }
1995                        if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
1996                                croak "Error parsing $_, $x->{chg}\n";
1997                        }
1998                        $x->{file_b} = $_;
1999                        $state = 'meta';
2000                } else {
2001                        croak "Error parsing $_\n";
2002                }
2003        }
2004        command_close_pipe($diff_fh, $ctx);
2005        $/ = $nl;
2006
2007        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
2008        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @mods) {
2009                my $f = $m->{chg};
2010                if (defined $o{$f}) {
2011                        $self->$f($m);
2012                } else {
2013                        fatal("Invalid change type: $f\n");
2014                }
2015        }
2016        $self->rmdirs($tree_b) if $::_rmdir;
2017        if (@mods == 0) {
2018                $self->abort_edit;
2019        } else {
2020                $self->close_edit;
2021        }
2022        \@mods;
2023}
2024
2025package Git::SVN::Ra;
2026use vars qw/@ISA $config_dir/;
2027use strict;
2028use warnings;
2029my ($can_do_switch);
2030my %RA;
2031
2032BEGIN {
2033        # enforce temporary pool usage for some simple functions
2034        my $e;
2035        foreach (qw/get_latest_revnum rev_proplist get_file
2036                    check_path get_dir get_uuid get_repos_root/) {
2037                $e .= "sub $_ {
2038                        my \$self = shift;
2039                        my \$pool = SVN::Pool->new;
2040                        my \@ret = \$self->SUPER::$_(\@_,\$pool);
2041                        \$pool->clear;
2042                        wantarray ? \@ret : \$ret[0]; }\n";
2043        }
2044        eval $e;
2045}
2046
2047sub new {
2048        my ($class, $url) = @_;
2049        $url =~ s!/+$!!;
2050        return $RA{$url} if $RA{$url};
2051
2052        SVN::_Core::svn_config_ensure($config_dir, undef);
2053        my ($baton, $callbacks) = SVN::Core::auth_open_helper([
2054            SVN::Client::get_simple_provider(),
2055            SVN::Client::get_ssl_server_trust_file_provider(),
2056            SVN::Client::get_simple_prompt_provider(
2057              \&Git::SVN::Prompt::simple, 2),
2058            SVN::Client::get_ssl_client_cert_prompt_provider(
2059              \&Git::SVN::Prompt::ssl_client_cert, 2),
2060            SVN::Client::get_ssl_client_cert_pw_prompt_provider(
2061              \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
2062            SVN::Client::get_username_provider(),
2063            SVN::Client::get_ssl_server_trust_prompt_provider(
2064              \&Git::SVN::Prompt::ssl_server_trust),
2065            SVN::Client::get_username_prompt_provider(
2066              \&Git::SVN::Prompt::username, 2),
2067          ]);
2068        my $config = SVN::Core::config_get_config($config_dir);
2069        my $self = SVN::Ra->new(url => $url, auth => $baton,
2070                              config => $config,
2071                              pool => SVN::Pool->new,
2072                              auth_provider_callbacks => $callbacks);
2073        $self->{svn_path} = $url;
2074        $self->{repos_root} = $self->get_repos_root;
2075        $self->{svn_path} =~ s#^\Q$self->{repos_root}\E/*##;
2076        $RA{$url} = bless $self, $class;
2077}
2078
2079sub DESTROY {
2080        # do not call the real DESTROY since we store ourselves in %RA
2081}
2082
2083sub dup {
2084        my ($self) = @_;
2085        my $dup = SVN::Ra->new(pool => SVN::Pool->new,
2086                                map { $_ => $self->{$_} } qw/config url
2087                     auth auth_provider_callbacks repos_root svn_path/);
2088        bless $dup, ref $self;
2089}
2090
2091sub get_log {
2092        my ($self, @args) = @_;
2093        my $pool = SVN::Pool->new;
2094        $args[4]-- if $args[4] && ! $::_follow_parent;
2095        splice(@args, 3, 1) if ($SVN::Core::VERSION le '1.2.0');
2096        my $ret = $self->SUPER::get_log(@args, $pool);
2097        $pool->clear;
2098        $ret;
2099}
2100
2101sub get_commit_editor {
2102        my ($self, $log, $cb, $pool) = @_;
2103        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
2104        $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
2105}
2106
2107sub uuid {
2108        my ($self) = @_;
2109        $self->{uuid} ||= $self->get_uuid;
2110}
2111
2112sub gs_do_update {
2113        my ($self, $rev_a, $rev_b, $path, $recurse, $editor) = @_;
2114        my $pool = SVN::Pool->new;
2115        my $reporter = $self->do_update($rev_b, $path, $recurse,
2116                                        $editor, $pool);
2117        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
2118        my $new = ($rev_a == $rev_b);
2119        $reporter->set_path('', $rev_a, $new, @lock, $pool);
2120        $reporter->finish_report($pool);
2121        $pool->clear;
2122        $editor->{git_commit_ok};
2123}
2124
2125sub gs_do_switch {
2126        my ($self, $rev_a, $rev_b, $path, $recurse, $url_b, $editor) = @_;
2127        my $pool = SVN::Pool->new;
2128        my $reporter = $self->do_switch($rev_b, $path, $recurse,
2129                                        $url_b, $editor, $pool);
2130        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
2131        $reporter->set_path($path, $rev_a, 0, @lock, $pool);
2132        $reporter->finish_report($pool);
2133        $pool->clear;
2134        $editor->{git_commit_ok};
2135}
2136
2137sub can_do_switch {
2138        my $self = shift;
2139        unless (defined $can_do_switch) {
2140                my $pool = SVN::Pool->new;
2141                my $rep = eval {
2142                        $self->do_switch(1, '', 0, $self->{url},
2143                                         SVN::Delta::Editor->new, $pool);
2144                };
2145                if ($@) {
2146                        $can_do_switch = 0;
2147                } else {
2148                        $rep->abort_report($pool);
2149                        $can_do_switch = 1;
2150                }
2151                $pool->clear;
2152        }
2153        $can_do_switch;
2154}
2155
2156package Git::SVN::Log;
2157use strict;
2158use warnings;
2159use POSIX qw/strftime/;
2160use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
2161            %rusers $show_commit $incremental/;
2162my $l_fmt;
2163
2164sub cmt_showable {
2165        my ($c) = @_;
2166        return 1 if defined $c->{r};
2167        if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
2168                                $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
2169                my @log = command(qw/cat-file commit/, $c->{c});
2170                shift @log while ($log[0] ne "\n");
2171                shift @log;
2172                @{$c->{l}} = grep !/^git-svn-id: /, @log;
2173
2174                (undef, $c->{r}, undef) = ::extract_metadata(
2175                                (grep(/^git-svn-id: /, @log))[-1]);
2176        }
2177        return defined $c->{r};
2178}
2179
2180sub log_use_color {
2181        return 1 if $color;
2182        my ($dc, $dcvar);
2183        $dcvar = 'color.diff';
2184        $dc = `git-config --get $dcvar`;
2185        if ($dc eq '') {
2186                # nothing at all; fallback to "diff.color"
2187                $dcvar = 'diff.color';
2188                $dc = `git-config --get $dcvar`;
2189        }
2190        chomp($dc);
2191        if ($dc eq 'auto') {
2192                my $pc;
2193                $pc = `git-config --get color.pager`;
2194                if ($pc eq '') {
2195                        # does not have it -- fallback to pager.color
2196                        $pc = `git-config --bool --get pager.color`;
2197                }
2198                else {
2199                        $pc = `git-config --bool --get color.pager`;
2200                        if ($?) {
2201                                $pc = 'false';
2202                        }
2203                }
2204                chomp($pc);
2205                if (-t *STDOUT || (defined $pager && $pc eq 'true')) {
2206                        return ($ENV{TERM} && $ENV{TERM} ne 'dumb');
2207                }
2208                return 0;
2209        }
2210        return 0 if $dc eq 'never';
2211        return 1 if $dc eq 'always';
2212        chomp($dc = `git-config --bool --get $dcvar`);
2213        return ($dc eq 'true');
2214}
2215
2216sub git_svn_log_cmd {
2217        my ($r_min, $r_max) = @_;
2218        my $gs = Git::SVN->_new;
2219        my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
2220                   $gs->refname);
2221        push @cmd, '-r' unless $non_recursive;
2222        push @cmd, qw/--raw --name-status/ if $verbose;
2223        push @cmd, '--color' if log_use_color();
2224        return @cmd unless defined $r_max;
2225        if ($r_max == $r_min) {
2226                push @cmd, '--max-count=1';
2227                if (my $c = $gs->rev_db_get($r_max)) {
2228                        push @cmd, $c;
2229                }
2230        } else {
2231                my ($c_min, $c_max);
2232                $c_max = $gs->rev_db_get($r_max);
2233                $c_min = $gs->rev_db_get($r_min);
2234                if (defined $c_min && defined $c_max) {
2235                        if ($r_max > $r_max) {
2236                                push @cmd, "$c_min..$c_max";
2237                        } else {
2238                                push @cmd, "$c_max..$c_min";
2239                        }
2240                } elsif ($r_max > $r_min) {
2241                        push @cmd, $c_max;
2242                } else {
2243                        push @cmd, $c_min;
2244                }
2245        }
2246        return @cmd;
2247}
2248
2249# adapted from pager.c
2250sub config_pager {
2251        $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
2252        if (!defined $pager) {
2253                $pager = 'less';
2254        } elsif (length $pager == 0 || $pager eq 'cat') {
2255                $pager = undef;
2256        }
2257}
2258
2259sub run_pager {
2260        return unless -t *STDOUT;
2261        pipe my $rfd, my $wfd or return;
2262        defined(my $pid = fork) or ::fatal "Can't fork: $!\n";
2263        if (!$pid) {
2264                open STDOUT, '>&', $wfd or
2265                                     ::fatal "Can't redirect to stdout: $!\n";
2266                return;
2267        }
2268        open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!\n";
2269        $ENV{LESS} ||= 'FRSX';
2270        exec $pager or ::fatal "Can't run pager: $! ($pager)\n";
2271}
2272
2273sub get_author_info {
2274        my ($dest, $author, $t, $tz) = @_;
2275        $author =~ s/(?:^\s*|\s*$)//g;
2276        $dest->{a_raw} = $author;
2277        my $au;
2278        if ($::_authors) {
2279                $au = $rusers{$author} || undef;
2280        }
2281        if (!$au) {
2282                ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
2283        }
2284        $dest->{t} = $t;
2285        $dest->{tz} = $tz;
2286        $dest->{a} = $au;
2287        # Date::Parse isn't in the standard Perl distro :(
2288        if ($tz =~ s/^\+//) {
2289                $t += ::tz_to_s_offset($tz);
2290        } elsif ($tz =~ s/^\-//) {
2291                $t -= ::tz_to_s_offset($tz);
2292        }
2293        $dest->{t_utc} = $t;
2294}
2295
2296sub process_commit {
2297        my ($c, $r_min, $r_max, $defer) = @_;
2298        if (defined $r_min && defined $r_max) {
2299                if ($r_min == $c->{r} && $r_min == $r_max) {
2300                        show_commit($c);
2301                        return 0;
2302                }
2303                return 1 if $r_min == $r_max;
2304                if ($r_min < $r_max) {
2305                        # we need to reverse the print order
2306                        return 0 if (defined $limit && --$limit < 0);
2307                        push @$defer, $c;
2308                        return 1;
2309                }
2310                if ($r_min != $r_max) {
2311                        return 1 if ($r_min < $c->{r});
2312                        return 1 if ($r_max > $c->{r});
2313                }
2314        }
2315        return 0 if (defined $limit && --$limit < 0);
2316        show_commit($c);
2317        return 1;
2318}
2319
2320sub show_commit {
2321        my $c = shift;
2322        if ($oneline) {
2323                my $x = "\n";
2324                if (my $l = $c->{l}) {
2325                        while ($l->[0] =~ /^\s*$/) { shift @$l }
2326                        $x = $l->[0];
2327                }
2328                $l_fmt ||= 'A' . length($c->{r});
2329                print 'r',pack($l_fmt, $c->{r}),' | ';
2330                print "$c->{c} | " if $show_commit;
2331                print $x;
2332        } else {
2333                show_commit_normal($c);
2334        }
2335}
2336
2337sub show_commit_changed_paths {
2338        my ($c) = @_;
2339        return unless $c->{changed};
2340        print "Changed paths:\n", @{$c->{changed}};
2341}
2342
2343sub show_commit_normal {
2344        my ($c) = @_;
2345        print '-' x72, "\nr$c->{r} | ";
2346        print "$c->{c} | " if $show_commit;
2347        print "$c->{a} | ", strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)",
2348                                 localtime($c->{t_utc})), ' | ';
2349        my $nr_line = 0;
2350
2351        if (my $l = $c->{l}) {
2352                while ($l->[$#$l] eq "\n" && $#$l > 0
2353                                          && $l->[($#$l - 1)] eq "\n") {
2354                        pop @$l;
2355                }
2356                $nr_line = scalar @$l;
2357                if (!$nr_line) {
2358                        print "1 line\n\n\n";
2359                } else {
2360                        if ($nr_line == 1) {
2361                                $nr_line = '1 line';
2362                        } else {
2363                                $nr_line .= ' lines';
2364                        }
2365                        print $nr_line, "\n";
2366                        show_commit_changed_paths($c);
2367                        print "\n";
2368                        print $_ foreach @$l;
2369                }
2370        } else {
2371                print "1 line\n";
2372                show_commit_changed_paths($c);
2373                print "\n";
2374
2375        }
2376        foreach my $x (qw/raw diff/) {
2377                if ($c->{$x}) {
2378                        print "\n";
2379                        print $_ foreach @{$c->{$x}}
2380                }
2381        }
2382}
2383
2384sub cmd_show_log {
2385        my (@args) = @_;
2386        my ($r_min, $r_max);
2387        my $r_last = -1; # prevent dupes
2388        if (defined $TZ) {
2389                $ENV{TZ} = $TZ;
2390        } else {
2391                delete $ENV{TZ};
2392        }
2393        if (defined $::_revision) {
2394                if ($::_revision =~ /^(\d+):(\d+)$/) {
2395                        ($r_min, $r_max) = ($1, $2);
2396                } elsif ($::_revision =~ /^\d+$/) {
2397                        $r_min = $r_max = $::_revision;
2398                } else {
2399                        ::fatal "-r$::_revision is not supported, use ",
2400                                "standard \'git log\' arguments instead\n";
2401                }
2402        }
2403
2404        config_pager();
2405        @args = (git_svn_log_cmd($r_min, $r_max), @args);
2406        my $log = command_output_pipe(@args);
2407        run_pager();
2408        my (@k, $c, $d);
2409        my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
2410        while (<$log>) {
2411                if (/^${esc_color}commit ($::sha1_short)/o) {
2412                        my $cmt = $1;
2413                        if ($c && cmt_showable($c) && $c->{r} != $r_last) {
2414                                $r_last = $c->{r};
2415                                process_commit($c, $r_min, $r_max, \@k) or
2416                                                                goto out;
2417                        }
2418                        $d = undef;
2419                        $c = { c => $cmt };
2420                } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
2421                        get_author_info($c, $1, $2, $3);
2422                } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
2423                        # ignore
2424                } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
2425                        push @{$c->{raw}}, $_;
2426                } elsif (/^${esc_color}[ACRMDT]\t/) {
2427                        # we could add $SVN->{svn_path} here, but that requires
2428                        # remote access at the moment (repo_path_split)...
2429                        s#^(${esc_color})([ACRMDT])\t#$1   $2 #o;
2430                        push @{$c->{changed}}, $_;
2431                } elsif (/^${esc_color}diff /o) {
2432                        $d = 1;
2433                        push @{$c->{diff}}, $_;
2434                } elsif ($d) {
2435                        push @{$c->{diff}}, $_;
2436                } elsif (/^${esc_color}    (git-svn-id:.+)$/o) {
2437                        ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
2438                } elsif (s/^${esc_color}    //o) {
2439                        push @{$c->{l}}, $_;
2440                }
2441        }
2442        if ($c && defined $c->{r} && $c->{r} != $r_last) {
2443                $r_last = $c->{r};
2444                process_commit($c, $r_min, $r_max, \@k);
2445        }
2446        if (@k) {
2447                my $swap = $r_max;
2448                $r_max = $r_min;
2449                $r_min = $swap;
2450                process_commit($_, $r_min, $r_max) foreach reverse @k;
2451        }
2452out:
2453        close $log;
2454        print '-' x72,"\n" unless $incremental || $oneline;
2455}
2456
2457package Git::SVN::Migration;
2458# these version numbers do NOT correspond to actual version numbers
2459# of git nor git-svn.  They are just relative.
2460#
2461# v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
2462#
2463# v1 layout: .git/$id/info/url, refs/remotes/$id
2464#
2465# v2 layout: .git/svn/$id/info/url, refs/remotes/$id
2466#
2467# v3 layout: .git/svn/$id, refs/remotes/$id
2468#            - info/url may remain for backwards compatibility
2469#            - this is what we migrate up to this layout automatically,
2470#            - this will be used by git svn init on single branches
2471#
2472# v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
2473#            - this is only created for newly multi-init-ed
2474#              repositories.  Similar in spirit to the
2475#              --use-separate-remotes option in git-clone (now default)
2476#            - we do not automatically migrate to this (following
2477#              the example set by core git)
2478use strict;
2479use warnings;
2480use Carp qw/croak/;
2481use File::Path qw/mkpath/;
2482use File::Basename qw/dirname basename/;
2483use vars qw/$_minimize/;
2484
2485sub migrate_from_v0 {
2486        my $git_dir = $ENV{GIT_DIR};
2487        return undef unless -d $git_dir;
2488        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
2489        my $migrated = 0;
2490        while (<$fh>) {
2491                chomp;
2492                my ($id, $orig_ref) = ($_, $_);
2493                next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
2494                next unless -f "$git_dir/$id/info/url";
2495                my $new_ref = "refs/remotes/$id";
2496                if (::verify_ref("$new_ref^0")) {
2497                        print STDERR "W: $orig_ref is probably an old ",
2498                                     "branch used by an ancient version of ",
2499                                     "git-svn.\n",
2500                                     "However, $new_ref also exists.\n",
2501                                     "We will not be able ",
2502                                     "to use this branch until this ",
2503                                     "ambiguity is resolved.\n";
2504                        next;
2505                }
2506                print STDERR "Migrating from v0 layout...\n" if !$migrated;
2507                print STDERR "Renaming ref: $orig_ref => $new_ref\n";
2508                command_noisy('update-ref', $new_ref, $orig_ref);
2509                command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
2510                $migrated++;
2511        }
2512        command_close_pipe($fh, $ctx);
2513        print STDERR "Done migrating from v0 layout...\n" if $migrated;
2514        $migrated;
2515}
2516
2517sub migrate_from_v1 {
2518        my $git_dir = $ENV{GIT_DIR};
2519        my $migrated = 0;
2520        return $migrated unless -d $git_dir;
2521        my $svn_dir = "$git_dir/svn";
2522
2523        # just in case somebody used 'svn' as their $id at some point...
2524        return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
2525
2526        print STDERR "Migrating from a git-svn v1 layout...\n";
2527        mkpath([$svn_dir]);
2528        print STDERR "Data from a previous version of git-svn exists, but\n\t",
2529                     "$svn_dir\n\t(required for this version ",
2530                     "($::VERSION) of git-svn) does not. exist\n";
2531        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
2532        while (<$fh>) {
2533                my $x = $_;
2534                next unless $x =~ s#^refs/remotes/##;
2535                chomp $x;
2536                next unless -f "$git_dir/$x/info/url";
2537                my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
2538                next unless $u;
2539                my $dn = dirname("$git_dir/svn/$x");
2540                mkpath([$dn]) unless -d $dn;
2541                if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
2542                        mkpath(["$git_dir/svn/svn"]);
2543                        print STDERR " - $git_dir/$x/info => ",
2544                                        "$git_dir/svn/$x/info\n";
2545                        rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
2546                               croak "$!: $x";
2547                        # don't worry too much about these, they probably
2548                        # don't exist with repos this old (save for index,
2549                        # and we can easily regenerate that)
2550                        foreach my $f (qw/unhandled.log index .rev_db/) {
2551                                rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
2552                        }
2553                } else {
2554                        print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
2555                        rename "$git_dir/$x", "$git_dir/svn/$x" or
2556                               croak "$!: $x";
2557                }
2558                $migrated++;
2559        }
2560        command_close_pipe($fh, $ctx);
2561        print STDERR "Done migrating from a git-svn v1 layout\n";
2562        $migrated;
2563}
2564
2565sub read_old_urls {
2566        my ($l_map, $pfx, $path) = @_;
2567        my @dir;
2568        foreach (<$path/*>) {
2569                if (-r "$_/info/url") {
2570                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
2571                        my $ref_id = $pfx . basename $_;
2572                        my $url = ::file_to_s("$_/info/url");
2573                        $l_map->{$ref_id} = $url;
2574                } elsif (-d $_) {
2575                        push @dir, $_;
2576                }
2577        }
2578        foreach (@dir) {
2579                my $x = $_;
2580                $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
2581                read_old_urls($l_map, $x, $_);
2582        }
2583}
2584
2585sub migrate_from_v2 {
2586        my @cfg = command(qw/config -l/);
2587        return if grep /^svn-remote\..+\.url=/, @cfg;
2588        my %l_map;
2589        read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
2590        my $migrated = 0;
2591
2592        foreach my $ref_id (sort keys %l_map) {
2593                Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
2594                $migrated++;
2595        }
2596        $migrated;
2597}
2598
2599sub minimize_connections {
2600        my $r = Git::SVN::read_all_remotes();
2601        my $new_urls = {};
2602        my $root_repos = {};
2603        foreach my $repo_id (keys %$r) {
2604                my $url = $r->{$repo_id}->{url} or next;
2605                my $fetch = $r->{$repo_id}->{fetch} or next;
2606                my $ra = Git::SVN::Ra->new($url);
2607
2608                # skip existing cases where we already connect to the root
2609                if (($ra->{url} eq $ra->{repos_root}) ||
2610                    (Git::SVN::sanitize_remote_name($ra->{repos_root}) eq
2611                     $repo_id)) {
2612                        $root_repos->{$ra->{url}} = $repo_id;
2613                        next;
2614                }
2615
2616                my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
2617                my $root_path = $ra->{url};
2618                $root_path =~ s#^\Q$ra->{repos_root}\E/*##;
2619                foreach my $path (keys %$fetch) {
2620                        my $ref_id = $fetch->{$path};
2621                        my $gs = Git::SVN->new($ref_id, $repo_id, $path);
2622
2623                        # make sure we can read when connecting to
2624                        # a higher level of a repository
2625                        my ($last_rev, undef) = $gs->last_rev_commit;
2626                        if (!defined $last_rev) {
2627                                $last_rev = eval {
2628                                        $root_ra->get_latest_revnum;
2629                                };
2630                                next if $@;
2631                        }
2632                        my $new = $root_path;
2633                        $new .= length $path ? "/$path" : '';
2634                        eval {
2635                                $root_ra->get_log([$new], $last_rev, $last_rev,
2636                                                  0, 0, 1, sub { });
2637                        };
2638                        next if $@;
2639                        $new_urls->{$ra->{repos_root}}->{$new} =
2640                                { ref_id => $ref_id,
2641                                  old_repo_id => $repo_id,
2642                                  old_path => $path };
2643                }
2644        }
2645
2646        my @emptied;
2647        foreach my $url (keys %$new_urls) {
2648                # see if we can re-use an existing [svn-remote "repo_id"]
2649                # instead of creating a(n ugly) new section:
2650                my $repo_id = $root_repos->{$url} ||
2651                              Git::SVN::sanitize_remote_name($url);
2652
2653                my $fetch = $new_urls->{$url};
2654                foreach my $path (keys %$fetch) {
2655                        my $x = $fetch->{$path};
2656                        Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
2657                        my $pfx = "svn-remote.$x->{old_repo_id}";
2658
2659                        my $old_fetch = quotemeta("$x->{old_path}:".
2660                                                  "refs/remotes/$x->{ref_id}");
2661                        command_noisy(qw/repo-config --unset/,
2662                                      "$pfx.fetch", '^'. $old_fetch . '$');
2663                        delete $r->{$x->{old_repo_id}}->
2664                               {fetch}->{$x->{old_path}};
2665                        if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
2666                                command_noisy(qw/repo-config --unset/,
2667                                              "$pfx.url");
2668                                push @emptied, $x->{old_repo_id}
2669                        }
2670                }
2671        }
2672        if (@emptied) {
2673                my $file = $ENV{GIT_CONFIG} || $ENV{GIT_CONFIG_LOCAL} ||
2674                           "$ENV{GIT_DIR}/config";
2675                print STDERR <<EOF;
2676The following [svn-remote] sections in your config file ($file) are empty
2677and can be safely removed:
2678EOF
2679                print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
2680        }
2681}
2682
2683sub migration_check {
2684        migrate_from_v0();
2685        migrate_from_v1();
2686        migrate_from_v2();
2687        minimize_connections() if $_minimize;
2688}
2689
2690__END__
2691
2692Data structures:
2693
2694$log_entry hashref as returned by libsvn_log_entry()
2695{
2696        log => 'whitespace-formatted log entry
2697',                                              # trailing newline is preserved
2698        revision => '8',                        # integer
2699        date => '2004-02-24T17:01:44.108345Z',  # commit date
2700        author => 'committer name'
2701};
2702
2703@mods = array of diff-index line hashes, each element represents one line
2704        of diff-index output
2705
2706diff-index line ($m hash)
2707{
2708        mode_a => first column of diff-index output, no leading ':',
2709        mode_b => second column of diff-index output,
2710        sha1_b => sha1sum of the final blob,
2711        chg => change type [MCRADT],
2712        file_a => original file name of a file (iff chg is 'C' or 'R')
2713        file_b => new/current file name of a file (any chg)
2714}
2715;
2716
2717# retval of read_url_paths{,_all}();
2718$l_map = {
2719        # repository root url
2720        'https://svn.musicpd.org' => {
2721                # repository path               # GIT_SVN_ID
2722                'mpd/trunk'             =>      'trunk',
2723                'mpd/tags/0.11.5'       =>      'tags/0.11.5',
2724        },
2725}
2726
2727Notes:
2728        I don't trust the each() function on unless I created %hash myself
2729        because the internal iterator may not have started at base.