1d448e75dab8e494afb1f5c3eede924bf95e9592
   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$Git::SVN::default_repo_id = 'git-svn';
  18$Git::SVN::default_ref_id = $ENV{GIT_SVN_ID} || 'git-svn';
  19
  20my $LC_ALL = $ENV{LC_ALL};
  21$Git::SVN::Log::TZ = $ENV{TZ};
  22# make sure the svn binary gives consistent output between locales and TZs:
  23$ENV{TZ} = 'UTC';
  24$ENV{LC_ALL} = 'C';
  25$| = 1; # unbuffer STDOUT
  26
  27sub fatal (@) { print STDERR @_; exit 1 }
  28require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
  29require SVN::Ra;
  30require SVN::Delta;
  31if ($SVN::Core::VERSION lt '1.1.0') {
  32        fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)\n";
  33}
  34push @Git::SVN::Ra::ISA, 'SVN::Ra';
  35push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
  36push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
  37use Carp qw/croak/;
  38use IO::File qw//;
  39use File::Basename qw/dirname basename/;
  40use File::Path qw/mkpath/;
  41use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev pass_through/;
  42use IPC::Open3;
  43use Git;
  44
  45BEGIN {
  46        my $s;
  47        foreach (qw/command command_oneline command_noisy command_output_pipe
  48                    command_input_pipe command_close_pipe/) {
  49                $s .= "*SVN::Git::Editor::$_ = *SVN::Git::Fetcher::$_ = ".
  50                      "*Git::SVN::Migration::$_ = ".
  51                      "*Git::SVN::Log::$_ = *Git::SVN::$_ = *$_ = *Git::$_; ";
  52        }
  53        eval $s;
  54}
  55
  56my ($SVN);
  57
  58my $_optimize_commits = 1 unless $ENV{GIT_SVN_NO_OPTIMIZE_COMMITS};
  59$sha1 = qr/[a-f\d]{40}/;
  60$sha1_short = qr/[a-f\d]{4,40}/;
  61my ($_stdin, $_help, $_edit,
  62        $_repack, $_repack_nr, $_repack_flags,
  63        $_message, $_file, $_no_metadata,
  64        $_template, $_shared,
  65        $_version, $_upgrade,
  66        $_merge, $_strategy, $_dry_run,
  67        $_prefix);
  68
  69my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
  70                    'config-dir=s' => \$Git::SVN::Ra::config_dir,
  71                    'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache );
  72my %fc_opts = ( 'follow-parent|follow' => \$_follow_parent,
  73                'authors-file|A=s' => \$_authors,
  74                'repack:i' => \$_repack,
  75                'no-metadata' => \$_no_metadata,
  76                'quiet|q' => \$_q,
  77                'repack-flags|repack-args|repack-opts=s' => \$_repack_flags,
  78                %remote_opts );
  79
  80my ($_trunk, $_tags, $_branches);
  81my %multi_opts = ( 'trunk|T=s' => \$_trunk,
  82                'tags|t=s' => \$_tags,
  83                'branches|b=s' => \$_branches );
  84my %init_opts = ( 'template=s' => \$_template, 'shared' => \$_shared );
  85my %cmt_opts = ( 'edit|e' => \$_edit,
  86                'rmdir' => \$_rmdir,
  87                'find-copies-harder' => \$_find_copies_harder,
  88                'l=i' => \$_l,
  89                'copy-similarity|C=i'=> \$_cp_similarity
  90);
  91
  92my %cmd = (
  93        fetch => [ \&cmd_fetch, "Download new revisions from SVN",
  94                        { 'revision|r=s' => \$_revision, %fc_opts } ],
  95        init => [ \&cmd_init, "Initialize a repo for tracking" .
  96                          " (requires URL argument)",
  97                          \%init_opts ],
  98        dcommit => [ \&cmd_dcommit,
  99                     'Commit several diffs to merge with upstream',
 100                        { 'merge|m|M' => \$_merge,
 101                          'strategy|s=s' => \$_strategy,
 102                          'dry-run|n' => \$_dry_run,
 103                        %cmt_opts, %fc_opts } ],
 104        'set-tree' => [ \&cmd_set_tree,
 105                        "Set an SVN repository to a git tree-ish",
 106                        { 'stdin|' => \$_stdin, %cmt_opts, %fc_opts, } ],
 107        'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
 108                        { 'revision|r=i' => \$_revision } ],
 109        rebuild => [ \&cmd_rebuild, "Rebuild git-svn metadata (after git clone)",
 110                        { 'copy-remote|remote=s' => \$_cp_remote,
 111                          'upgrade' => \$_upgrade } ],
 112        'multi-init' => [ \&cmd_multi_init,
 113                        'Initialize multiple trees (like git-svnimport)',
 114                        { %multi_opts, %init_opts, %remote_opts,
 115                         'revision|r=i' => \$_revision,
 116                         'prefix=s' => \$_prefix,
 117                        } ],
 118        'multi-fetch' => [ \&cmd_multi_fetch,
 119                        'Fetch multiple trees (like git-svnimport)',
 120                        \%fc_opts ],
 121        'migrate' => [ sub { },
 122                       # no-op, we automatically run this anyways,
 123                       'Migrate configuration/metadata/layout from
 124                        previous versions of git-svn',
 125                        \%remote_opts ],
 126        'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
 127                        { 'limit=i' => \$Git::SVN::Log::limit,
 128                          'revision|r=s' => \$_revision,
 129                          'verbose|v' => \$Git::SVN::Log::verbose,
 130                          'incremental' => \$Git::SVN::Log::incremental,
 131                          'oneline' => \$Git::SVN::Log::oneline,
 132                          'show-commit' => \$Git::SVN::Log::show_commit,
 133                          'non-recursive' => \$Git::SVN::Log::non_recursive,
 134                          'authors-file|A=s' => \$_authors,
 135                          'color' => \$Git::SVN::Log::color,
 136                          'pager=s' => \$Git::SVN::Log::pager,
 137                        } ],
 138        'commit-diff' => [ \&cmd_commit_diff,
 139                           'Commit a diff between two trees',
 140                        { 'message|m=s' => \$_message,
 141                          'file|F=s' => \$_file,
 142                          'revision|r=s' => \$_revision,
 143                        %cmt_opts } ],
 144);
 145
 146my $cmd;
 147for (my $i = 0; $i < @ARGV; $i++) {
 148        if (defined $cmd{$ARGV[$i]}) {
 149                $cmd = $ARGV[$i];
 150                splice @ARGV, $i, 1;
 151                last;
 152        }
 153};
 154
 155my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
 156
 157read_repo_config(\%opts);
 158my $rv = GetOptions(%opts, 'help|H|h' => \$_help,
 159                                'version|V' => \$_version,
 160                                'minimize-connections' =>
 161                                  \$Git::SVN::Migration::_minimize,
 162                                'id|i=s' => \$Git::SVN::default_ref_id);
 163exit 1 if (!$rv && $cmd ne 'log');
 164
 165usage(0) if $_help;
 166version() if $_version;
 167usage(1) unless defined $cmd;
 168load_authors() if $_authors;
 169unless ($cmd =~ /^(?:init|rebuild|multi-init|commit-diff)$/) {
 170        Git::SVN::Migration::migration_check();
 171}
 172eval {
 173        Git::SVN::verify_remotes_sanity();
 174        $cmd{$cmd}->[0]->(@ARGV);
 175};
 176fatal $@ if $@;
 177exit 0;
 178
 179####################### primary functions ######################
 180sub usage {
 181        my $exit = shift || 0;
 182        my $fd = $exit ? \*STDERR : \*STDOUT;
 183        print $fd <<"";
 184git-svn - bidirectional operations between a single Subversion tree and git
 185Usage: $0 <command> [options] [arguments]\n
 186
 187        print $fd "Available commands:\n" unless $cmd;
 188
 189        foreach (sort keys %cmd) {
 190                next if $cmd && $cmd ne $_;
 191                print $fd '  ',pack('A17',$_),$cmd{$_}->[1],"\n";
 192                foreach (keys %{$cmd{$_}->[2]}) {
 193                        # prints out arguments as they should be passed:
 194                        my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
 195                        print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
 196                                                        "--$_" : "-$_" }
 197                                                split /\|/,$_)," $x\n";
 198                }
 199        }
 200        print $fd <<"";
 201\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
 202arbitrary identifier if you're tracking multiple SVN branches/repositories in
 203one git repository and want to keep them separate.  See git-svn(1) for more
 204information.
 205
 206        exit $exit;
 207}
 208
 209sub version {
 210        print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
 211        exit 0;
 212}
 213
 214sub cmd_rebuild {
 215        my $url = shift;
 216        my $gs = $url ? Git::SVN->init($url)
 217                      : eval { Git::SVN->new };
 218        $gs ||= Git::SVN->_new;
 219        if (!verify_ref($gs->refname.'^0')) {
 220                $gs->copy_remote_ref;
 221        }
 222
 223        my ($rev_list, $ctx) = command_output_pipe("rev-list", $gs->refname);
 224        my $latest;
 225        my $svn_uuid;
 226        while (<$rev_list>) {
 227                chomp;
 228                my $c = $_;
 229                fatal "Non-SHA1: $c\n" unless $c =~ /^$sha1$/o;
 230                my ($url, $rev, $uuid) = cmt_metadata($c);
 231
 232                # ignore merges (from set-tree)
 233                next if (!defined $rev || !$uuid);
 234
 235                # if we merged or otherwise started elsewhere, this is
 236                # how we break out of it
 237                if ((defined $svn_uuid && ($uuid ne $svn_uuid)) ||
 238                    ($gs->{url} && $url && ($url ne $gs->{url}))) {
 239                        next;
 240                }
 241
 242                unless (defined $latest) {
 243                        if (!$gs->{url} && !$url) {
 244                                fatal "SVN repository location required\n";
 245                        }
 246                        $gs = Git::SVN->init($url);
 247                        $latest = $rev;
 248                }
 249                $gs->rev_db_set($rev, $c);
 250                print "r$rev = $c\n";
 251        }
 252        command_close_pipe($rev_list, $ctx);
 253}
 254
 255sub do_git_init_db {
 256        unless (-d $ENV{GIT_DIR}) {
 257                my @init_db = ('init');
 258                push @init_db, "--template=$_template" if defined $_template;
 259                push @init_db, "--shared" if defined $_shared;
 260                command_noisy(@init_db);
 261        }
 262}
 263
 264sub cmd_init {
 265        my $url = shift or die "SVN repository location required " .
 266                                "as a command-line argument\n";
 267        if (my $repo_path = shift) {
 268                unless (-d $repo_path) {
 269                        mkpath([$repo_path]);
 270                }
 271                chdir $repo_path or croak $!;
 272                $ENV{GIT_DIR} = $repo_path . "/.git";
 273        }
 274        do_git_init_db();
 275
 276        Git::SVN->init($url);
 277}
 278
 279sub cmd_fetch {
 280        if (@_) {
 281                die "Additional fetch arguments are no longer supported.\n",
 282                    "Use --follow-parent if you have moved/copied directories
 283                    instead.\n";
 284        }
 285        my $gs = Git::SVN->new;
 286        $gs->fetch(parse_revision_argument());
 287        if ($gs->{last_commit} && !verify_ref('refs/heads/master^0')) {
 288                command_noisy(qw(update-ref refs/heads/master),
 289                              $gs->{last_commit});
 290        }
 291}
 292
 293sub cmd_set_tree {
 294        my (@commits) = @_;
 295        if ($_stdin || !@commits) {
 296                print "Reading from stdin...\n";
 297                @commits = ();
 298                while (<STDIN>) {
 299                        if (/\b($sha1_short)\b/o) {
 300                                unshift @commits, $1;
 301                        }
 302                }
 303        }
 304        my @revs;
 305        foreach my $c (@commits) {
 306                my @tmp = command('rev-parse',$c);
 307                if (scalar @tmp == 1) {
 308                        push @revs, $tmp[0];
 309                } elsif (scalar @tmp > 1) {
 310                        push @revs, reverse(command('rev-list',@tmp));
 311                } else {
 312                        fatal "Failed to rev-parse $c\n";
 313                }
 314        }
 315        my $gs = Git::SVN->new;
 316        my ($r_last, $cmt_last) = $gs->last_rev_commit;
 317        $gs->fetch;
 318        if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
 319                fatal "There are new revisions that were fetched ",
 320                      "and need to be merged (or acknowledged) ",
 321                      "before committing.\nlast rev: $r_last\n",
 322                      " current: $gs->{last_rev}\n";
 323        }
 324        $gs->set_tree($_) foreach @revs;
 325        print "Done committing ",scalar @revs," revisions to SVN\n";
 326}
 327
 328sub cmd_dcommit {
 329        my $head = shift;
 330        my $gs = Git::SVN->new;
 331        $head ||= 'HEAD';
 332        my @refs = command(qw/rev-list --no-merges/, $gs->refname."..$head");
 333        my $last_rev;
 334        foreach my $d (reverse @refs) {
 335                if (!verify_ref("$d~1")) {
 336                        fatal "Commit $d\n",
 337                              "has no parent commit, and therefore ",
 338                              "nothing to diff against.\n",
 339                              "You should be working from a repository ",
 340                              "originally created by git-svn\n";
 341                }
 342                unless (defined $last_rev) {
 343                        (undef, $last_rev, undef) = cmt_metadata("$d~1");
 344                        unless (defined $last_rev) {
 345                                fatal "Unable to extract revision information ",
 346                                      "from commit $d~1\n";
 347                        }
 348                }
 349                if ($_dry_run) {
 350                        print "diff-tree $d~1 $d\n";
 351                } else {
 352                        my $log = get_commit_entry($d)->{log};
 353                        my $ra = $gs->ra;
 354                        my $pool = SVN::Pool->new;
 355                        my %ed_opts = ( r => $last_rev,
 356                                        ra => $ra->dup,
 357                                        svn_path => $gs->{path} );
 358                        my $ed = SVN::Git::Editor->new(\%ed_opts,
 359                                         $ra->get_commit_editor($log,
 360                                         sub { print "Committed r$_[0]\n";
 361                                               $last_rev = $_[0]; }),
 362                                         $pool);
 363                        my $mods = $ed->apply_diff("$d~1", $d);
 364                        if (@$mods == 0) {
 365                                print "No changes\n$d~1 == $d\n";
 366                        }
 367                }
 368        }
 369        return if $_dry_run;
 370        $gs->fetch;
 371        # we always want to rebase against the current HEAD, not any
 372        # head that was passed to us
 373        my @diff = command('diff-tree', 'HEAD', $gs->refname, '--');
 374        my @finish;
 375        if (@diff) {
 376                @finish = qw/rebase/;
 377                push @finish, qw/--merge/ if $_merge;
 378                push @finish, "--strategy=$_strategy" if $_strategy;
 379                print STDERR "W: HEAD and ", $gs->refname, " differ, ",
 380                             "using @finish:\n", "@diff";
 381        } else {
 382                print "No changes between current HEAD and ",
 383                      $gs->refname, "\nResetting to the latest ",
 384                      $gs->refname, "\n";
 385                @finish = qw/reset --mixed/;
 386        }
 387        command_noisy(@finish, $gs->refname);
 388}
 389
 390sub cmd_show_ignore {
 391        my $gs = Git::SVN->new;
 392        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 393        $gs->traverse_ignore(\*STDOUT, '', $r);
 394}
 395
 396sub cmd_multi_init {
 397        my $url = shift;
 398        unless (defined $_trunk || defined $_branches || defined $_tags) {
 399                usage(1);
 400        }
 401        do_git_init_db();
 402        $_prefix = '' unless defined $_prefix;
 403        $url =~ s#/+$## if defined $url;
 404        if (defined $_trunk) {
 405                my $trunk_ref = $_prefix . 'trunk';
 406                # try both old-style and new-style lookups:
 407                my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
 408                unless ($gs_trunk) {
 409                        my ($trunk_url, $trunk_path) =
 410                                              complete_svn_url($url, $_trunk);
 411                        $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
 412                                                   undef, $trunk_ref);
 413                }
 414        }
 415        return unless defined $_branches || defined $_tags;
 416        my $ra = $url ? Git::SVN::Ra->new($url) : undef;
 417        complete_url_ls_init($ra, $_branches, '--branches/-b', $_prefix);
 418        complete_url_ls_init($ra, $_tags, '--tags/-t', $_prefix . 'tags/');
 419}
 420
 421sub cmd_multi_fetch {
 422        my @gs;
 423        foreach (command(qw/config -l/)) {
 424                next unless m!^svn-remote\.(.+)\.fetch=
 425                              \s*(.*)\s*:\s*refs/remotes/(.+)\s*$!x;
 426                my ($repo_id, $path, $ref_id) = ($1, $2, $3);
 427                push @gs, Git::SVN->new($ref_id, $repo_id, $path);
 428        }
 429        foreach (@gs) {
 430                $_->fetch;
 431        }
 432}
 433
 434# this command is special because it requires no metadata
 435sub cmd_commit_diff {
 436        my ($ta, $tb, $url) = @_;
 437        my $usage = "Usage: $0 commit-diff -r<revision> ".
 438                    "<tree-ish> <tree-ish> [<URL>]\n";
 439        fatal($usage) if (!defined $ta || !defined $tb);
 440        my $svn_path;
 441        if (!defined $url) {
 442                my $gs = eval { Git::SVN->new };
 443                if (!$gs) {
 444                        fatal("Needed URL or usable git-svn --id in ",
 445                              "the command-line\n", $usage);
 446                }
 447                $url = $gs->{url};
 448                $svn_path = $gs->{path};
 449        }
 450        unless (defined $_revision) {
 451                fatal("-r|--revision is a required argument\n", $usage);
 452        }
 453        if (defined $_message && defined $_file) {
 454                fatal("Both --message/-m and --file/-F specified ",
 455                      "for the commit message.\n",
 456                      "I have no idea what you mean\n");
 457        }
 458        if (defined $_file) {
 459                $_message = file_to_s($_file);
 460        } else {
 461                $_message ||= get_commit_entry($tb)->{log};
 462        }
 463        my $ra ||= Git::SVN::Ra->new($url);
 464        $svn_path ||= $ra->{svn_path};
 465        my $r = $_revision;
 466        if ($r eq 'HEAD') {
 467                $r = $ra->get_latest_revnum;
 468        } elsif ($r !~ /^\d+$/) {
 469                die "revision argument: $r not understood by git-svn\n";
 470        }
 471        my $pool = SVN::Pool->new;
 472        my %ed_opts = ( r => $r,
 473                        ra => $ra->dup,
 474                        svn_path => $svn_path );
 475        my $ed = SVN::Git::Editor->new(\%ed_opts,
 476                                       $ra->get_commit_editor($_message,
 477                                         sub { print "Committed r$_[0]\n" }),
 478                                       $pool);
 479        my $mods = $ed->apply_diff($ta, $tb);
 480        if (@$mods == 0) {
 481                print "No changes\n$ta == $tb\n";
 482        }
 483        $pool->clear;
 484}
 485
 486########################### utility functions #########################
 487
 488sub parse_revision_argument {
 489        if (!defined $_revision || $_revision eq 'BASE:HEAD') {
 490                return (undef, undef);
 491        }
 492        return ($1, $2) if ($_revision =~ /^(\d+):(\d+)$/);
 493        return ($_revision, $_revision) if ($_revision =~ /^\d+$/);
 494        return (undef, $1) if ($_revision =~ /^BASE:(\d+)$/);
 495        return ($1, undef) if ($_revision =~ /^(\d+):HEAD$/);
 496        die "revision argument: $_revision not understood by git-svn\n",
 497            "Try using the command-line svn client instead\n";
 498}
 499
 500sub complete_svn_url {
 501        my ($url, $path) = @_;
 502        $path =~ s#/+$##;
 503        if ($path !~ m#^[a-z\+]+://#) {
 504                if (!defined $url || $url !~ m#^[a-z\+]+://#) {
 505                        fatal("E: '$path' is not a complete URL ",
 506                              "and a separate URL is not specified\n");
 507                }
 508                return ($url, $path);
 509        }
 510        return ($path, '');
 511}
 512
 513sub complete_url_ls_init {
 514        my ($ra, $repo_path, $switch, $pfx) = @_;
 515        unless ($repo_path) {
 516                print STDERR "W: $switch not specified\n";
 517                return;
 518        }
 519        $repo_path =~ s#/+$##;
 520        if ($repo_path =~ m#^[a-z\+]+://#) {
 521                $ra = Git::SVN::Ra->new($repo_path);
 522                $repo_path = '';
 523        } else {
 524                $repo_path =~ s#^/+##;
 525                unless ($ra) {
 526                        fatal("E: '$repo_path' is not a complete URL ",
 527                              "and a separate URL is not specified\n");
 528                }
 529        }
 530        my $r = defined $_revision ? $_revision : $ra->get_latest_revnum;
 531        my ($dirent, undef, undef) = $ra->get_dir($repo_path, $r);
 532        my $url = $ra->{url};
 533        foreach my $d (sort keys %$dirent) {
 534                next if ($dirent->{$d}->kind != $SVN::Node::dir);
 535                my $path =  "$repo_path/$d";
 536                my $ref = "$pfx$d";
 537                my $gs = eval { Git::SVN->new($ref) };
 538                # don't try to init already existing refs
 539                unless ($gs) {
 540                        print "init $url/$path => $ref\n";
 541                        Git::SVN->init($url, $path, undef, $ref);
 542                }
 543        }
 544}
 545
 546sub verify_ref {
 547        my ($ref) = @_;
 548        eval { command_oneline([ 'rev-parse', '--verify', $ref ],
 549                               { STDERR => 0 }); };
 550}
 551
 552sub get_tree_from_treeish {
 553        my ($treeish) = @_;
 554        # $treeish can be a symbolic ref, too:
 555        my $type = command_oneline(qw/cat-file -t/, $treeish);
 556        my $expected;
 557        while ($type eq 'tag') {
 558                ($treeish, $type) = command(qw/cat-file tag/, $treeish);
 559        }
 560        if ($type eq 'commit') {
 561                $expected = (grep /^tree /, command(qw/cat-file commit/,
 562                                                    $treeish))[0];
 563                ($expected) = ($expected =~ /^tree ($sha1)$/o);
 564                die "Unable to get tree from $treeish\n" unless $expected;
 565        } elsif ($type eq 'tree') {
 566                $expected = $treeish;
 567        } else {
 568                die "$treeish is a $type, expected tree, tag or commit\n";
 569        }
 570        return $expected;
 571}
 572
 573sub get_commit_entry {
 574        my ($treeish) = shift;
 575        my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
 576        my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
 577        my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
 578        open my $log_fh, '>', $commit_editmsg or croak $!;
 579
 580        my $type = command_oneline(qw/cat-file -t/, $treeish);
 581        if ($type eq 'commit' || $type eq 'tag') {
 582                my ($msg_fh, $ctx) = command_output_pipe('cat-file',
 583                                                         $type, $treeish);
 584                my $in_msg = 0;
 585                while (<$msg_fh>) {
 586                        if (!$in_msg) {
 587                                $in_msg = 1 if (/^\s*$/);
 588                        } elsif (/^git-svn-id: /) {
 589                                # skip this for now, we regenerate the
 590                                # correct one on re-fetch anyways
 591                                # TODO: set *:merge properties or like...
 592                        } else {
 593                                print $log_fh $_ or croak $!;
 594                        }
 595                }
 596                command_close_pipe($msg_fh, $ctx);
 597        }
 598        close $log_fh or croak $!;
 599
 600        if ($_edit || ($type eq 'tree')) {
 601                my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
 602                # TODO: strip out spaces, comments, like git-commit.sh
 603                system($editor, $commit_editmsg);
 604        }
 605        rename $commit_editmsg, $commit_msg or croak $!;
 606        open $log_fh, '<', $commit_msg or croak $!;
 607        { local $/; chomp($log_entry{log} = <$log_fh>); }
 608        close $log_fh or croak $!;
 609        unlink $commit_msg;
 610        \%log_entry;
 611}
 612
 613sub s_to_file {
 614        my ($str, $file, $mode) = @_;
 615        open my $fd,'>',$file or croak $!;
 616        print $fd $str,"\n" or croak $!;
 617        close $fd or croak $!;
 618        chmod ($mode &~ umask, $file) if (defined $mode);
 619}
 620
 621sub file_to_s {
 622        my $file = shift;
 623        open my $fd,'<',$file or croak "$!: file: $file\n";
 624        local $/;
 625        my $ret = <$fd>;
 626        close $fd or croak $!;
 627        $ret =~ s/\s*$//s;
 628        return $ret;
 629}
 630
 631# '<svn username> = real-name <email address>' mapping based on git-svnimport:
 632sub load_authors {
 633        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
 634        my $log = $cmd eq 'log';
 635        while (<$authors>) {
 636                chomp;
 637                next unless /^(\S+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
 638                my ($user, $name, $email) = ($1, $2, $3);
 639                if ($log) {
 640                        $Git::SVN::Log::rusers{"$name <$email>"} = $user;
 641                } else {
 642                        $users{$user} = [$name, $email];
 643                }
 644        }
 645        close $authors or croak $!;
 646}
 647
 648# convert GetOpt::Long specs for use by git-config
 649sub read_repo_config {
 650        return unless -d $ENV{GIT_DIR};
 651        my $opts = shift;
 652        foreach my $o (keys %$opts) {
 653                my $v = $opts->{$o};
 654                my ($key) = ($o =~ /^([a-z\-]+)/);
 655                $key =~ s/-//g;
 656                my $arg = 'git-config';
 657                $arg .= ' --int' if ($o =~ /[:=]i$/);
 658                $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
 659                if (ref $v eq 'ARRAY') {
 660                        chomp(my @tmp = `$arg --get-all svn.$key`);
 661                        @$v = @tmp if @tmp;
 662                } else {
 663                        chomp(my $tmp = `$arg --get svn.$key`);
 664                        if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
 665                                $$v = $tmp;
 666                        }
 667                }
 668        }
 669}
 670
 671sub extract_metadata {
 672        my $id = shift or return (undef, undef, undef);
 673        my ($url, $rev, $uuid) = ($id =~ /^git-svn-id:\s(\S+?)\@(\d+)
 674                                                        \s([a-f\d\-]+)$/x);
 675        if (!defined $rev || !$uuid || !$url) {
 676                # some of the original repositories I made had
 677                # identifiers like this:
 678                ($rev, $uuid) = ($id =~/^git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
 679        }
 680        return ($url, $rev, $uuid);
 681}
 682
 683sub cmt_metadata {
 684        return extract_metadata((grep(/^git-svn-id: /,
 685                command(qw/cat-file commit/, shift)))[-1]);
 686}
 687
 688sub get_commit_time {
 689        my $cmt = shift;
 690        my $fh = command_output_pipe(qw/rev-list --pretty=raw -n1/, $cmt);
 691        while (<$fh>) {
 692                /^committer\s(?:.+) (\d+) ([\-\+]?\d+)$/ or next;
 693                my ($s, $tz) = ($1, $2);
 694                if ($tz =~ s/^\+//) {
 695                        $s += tz_to_s_offset($tz);
 696                } elsif ($tz =~ s/^\-//) {
 697                        $s -= tz_to_s_offset($tz);
 698                }
 699                close $fh;
 700                return $s;
 701        }
 702        die "Can't get commit time for commit: $cmt\n";
 703}
 704
 705sub tz_to_s_offset {
 706        my ($tz) = @_;
 707        $tz =~ s/(\d\d)$//;
 708        return ($1 * 60) + ($tz * 3600);
 709}
 710
 711package Git::SVN;
 712use strict;
 713use warnings;
 714use vars qw/$default_repo_id $default_ref_id/;
 715use Carp qw/croak/;
 716use File::Path qw/mkpath/;
 717use IPC::Open3;
 718
 719# properties that we do not log:
 720my %SKIP_PROP;
 721BEGIN {
 722        %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
 723                                        svn:special svn:executable
 724                                        svn:entry:committed-rev
 725                                        svn:entry:last-author
 726                                        svn:entry:uuid
 727                                        svn:entry:committed-date/;
 728}
 729
 730sub read_all_remotes {
 731        my $r = {};
 732        foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
 733                if (m!^(.+)\.fetch=\s*(.*)\s*:\s*refs/remotes/(.+)\s*$!) {
 734                        $r->{$1}->{fetch}->{$2} = $3;
 735                } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
 736                        $r->{$1}->{url} = $2;
 737                }
 738        }
 739        $r;
 740}
 741
 742sub verify_remotes_sanity {
 743        return unless -d $ENV{GIT_DIR};
 744        my %seen;
 745        foreach (command(qw/config -l/)) {
 746                if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
 747                        if ($seen{$1}) {
 748                                die "Remote ref refs/remote/$1 is tracked by",
 749                                    "\n  \"$_\"\nand\n  \"$seen{$1}\"\n",
 750                                    "Please resolve this ambiguity in ",
 751                                    "your git configuration file before ",
 752                                    "continuing\n";
 753                        }
 754                        $seen{$1} = $_;
 755                }
 756        }
 757}
 758
 759# we allow more chars than remotes2config.sh...
 760sub sanitize_remote_name {
 761        my ($name) = @_;
 762        $name =~ tr{A-Za-z0-9:,/+-}{.}c;
 763        $name;
 764}
 765
 766sub find_existing_remote {
 767        my ($url, $remotes) = @_;
 768        my $existing;
 769        foreach my $repo_id (keys %$remotes) {
 770                my $u = $remotes->{$repo_id}->{url} or next;
 771                next if $u ne $url;
 772                $existing = $repo_id;
 773                last;
 774        }
 775        $existing;
 776}
 777
 778sub init_remote_config {
 779        my ($self, $url) = @_;
 780        $url =~ s!/+$!!; # strip trailing slash
 781        my $r = read_all_remotes();
 782        my $existing = find_existing_remote($url, $r);
 783        if ($existing) {
 784                print STDERR "Using existing ",
 785                             "[svn-remote \"$existing\"]\n";
 786                $self->{repo_id} = $existing;
 787        } else {
 788                my $min_url = Git::SVN::Ra->new($url)->minimize_url;
 789                $existing = find_existing_remote($min_url, $r);
 790                if ($existing) {
 791                        print STDERR "Using existing ",
 792                                     "[svn-remote \"$existing\"]\n";
 793                        $self->{repo_id} = $existing;
 794                }
 795                if ($min_url ne $url) {
 796                        print STDERR "Using higher level of URL: ",
 797                                     "$url => $min_url\n";
 798                        my $old_path = $self->{path};
 799                        $self->{path} = $url;
 800                        $self->{path} =~ s!^\Q$min_url\E/*!!;
 801                        if (length $old_path) {
 802                                $self->{path} .= "/$old_path";
 803                        }
 804                        $url = $min_url;
 805                }
 806        }
 807        my $orig_url;
 808        if (!$existing) {
 809                # verify that we aren't overwriting anything:
 810                $orig_url = eval {
 811                        command_oneline('config', '--get',
 812                                        "svn-remote.$self->{repo_id}.url")
 813                };
 814                if ($orig_url && ($orig_url ne $url)) {
 815                        die "svn-remote.$self->{repo_id}.url already set: ",
 816                            "$orig_url\nwanted to set to: $url\n";
 817                }
 818        }
 819        my ($xrepo_id, $xpath) = find_ref($self->refname);
 820        if (defined $xpath) {
 821                die "svn-remote.$xrepo_id.fetch already set to track ",
 822                    "$xpath:refs/remotes/", $self->refname, "\n";
 823        }
 824        command_noisy('config',
 825                      "svn-remote.$self->{repo_id}.url", $url);
 826        command_noisy('config', '--add',
 827                      "svn-remote.$self->{repo_id}.fetch",
 828                      "$self->{path}:".$self->refname);
 829        $self->{url} = $url;
 830}
 831
 832sub init {
 833        my ($class, $url, $path, $repo_id, $ref_id) = @_;
 834        my $self = _new($class, $repo_id, $ref_id, $path);
 835        if (defined $url) {
 836                $self->init_remote_config($url);
 837        }
 838        $self;
 839}
 840
 841sub find_ref {
 842        my ($ref_id) = @_;
 843        foreach (command(qw/config -l/)) {
 844                next unless m!^svn-remote\.(.+)\.fetch=
 845                              \s*(.*)\s*:\s*refs/remotes/(.+)\s*$!x;
 846                my ($repo_id, $path, $ref) = ($1, $2, $3);
 847                if ($ref eq $ref_id) {
 848                        $path = '' if ($path =~ m#^\./?#);
 849                        return ($repo_id, $path);
 850                }
 851        }
 852        (undef, undef, undef);
 853}
 854
 855sub new {
 856        my ($class, $ref_id, $repo_id, $path) = @_;
 857        if (defined $ref_id && !defined $repo_id && !defined $path) {
 858                ($repo_id, $path) = find_ref($ref_id);
 859                if (!defined $repo_id) {
 860                        die "Could not find a \"svn-remote.*.fetch\" key ",
 861                            "in the repository configuration matching: ",
 862                            "refs/remotes/$ref_id\n";
 863                }
 864        }
 865        my $self = _new($class, $repo_id, $ref_id, $path);
 866        if (!defined $self->{path} || !length $self->{path}) {
 867                my $fetch = command_oneline('config', '--get',
 868                                            "svn-remote.$repo_id.fetch",
 869                                            ":refs/remotes/$ref_id\$") or
 870                     die "Failed to read \"svn-remote.$repo_id.fetch\" ",
 871                         "\":refs/remotes/$ref_id\$\" in config\n";
 872                ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
 873        }
 874        $self->{url} = command_oneline('config', '--get',
 875                                       "svn-remote.$repo_id.url") or
 876                  die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
 877        $self;
 878}
 879
 880sub refname { "refs/remotes/$_[0]->{ref_id}" }
 881
 882sub ra {
 883        my ($self) = shift;
 884        $self->{ra} ||= Git::SVN::Ra->new($self->{url});
 885}
 886
 887sub rel_path {
 888        my ($self) = @_;
 889        my $repos_root = $self->ra->{repos_root};
 890        return $self->{path} if ($self->{url} eq $repos_root);
 891        my $url = $self->{url} .
 892                  (length $self->{path} ? "/$self->{path}" : $self->{path});
 893        $url =~ s!^\Q$repos_root\E/*!!g;
 894        $url;
 895}
 896
 897sub copy_remote_ref {
 898        my ($self) = @_;
 899        my $origin = $::_cp_remote ? $::_cp_remote : 'origin';
 900        my $ref = $self->refname;
 901        if (command('ls-remote', $origin, $ref)) {
 902                command_noisy('fetch', $origin, "$ref:$ref");
 903        } elsif ($::_cp_remote && !$::_upgrade) {
 904                die "Unable to find remote reference: $ref on $origin\n";
 905        }
 906}
 907
 908sub traverse_ignore {
 909        my ($self, $fh, $path, $r) = @_;
 910        $path =~ s#^/+##g;
 911        my ($dirent, undef, $props) = $self->ra->get_dir($path, $r);
 912        my $p = $path;
 913        $p =~ s#^\Q$self->{ra}->{svn_path}\E/##;
 914        print $fh length $p ? "\n# $p\n" : "\n# /\n";
 915        if (my $s = $props->{'svn:ignore'}) {
 916                $s =~ s/[\r\n]+/\n/g;
 917                chomp $s;
 918                if (length $p == 0) {
 919                        $s =~ s#\n#\n/$p#g;
 920                        print $fh "/$s\n";
 921                } else {
 922                        $s =~ s#\n#\n/$p/#g;
 923                        print $fh "/$p/$s\n";
 924                }
 925        }
 926        foreach (sort keys %$dirent) {
 927                next if $dirent->{$_}->kind != $SVN::Node::dir;
 928                $self->traverse_ignore($fh, "$path/$_", $r);
 929        }
 930}
 931
 932sub last_rev { ($_[0]->last_rev_commit)[0] }
 933sub last_commit { ($_[0]->last_rev_commit)[1] }
 934
 935# returns the newest SVN revision number and newest commit SHA1
 936sub last_rev_commit {
 937        my ($self) = @_;
 938        if (defined $self->{last_rev} && defined $self->{last_commit}) {
 939                return ($self->{last_rev}, $self->{last_commit});
 940        }
 941        my $c = ::verify_ref($self->refname.'^0');
 942        if ($c) {
 943                my $rev = (::cmt_metadata($c))[1];
 944                if (defined $rev) {
 945                        ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
 946                        return ($rev, $c);
 947                }
 948        }
 949        my $offset = -41; # from tail
 950        my $rl;
 951        open my $fh, '<', $self->{db_path} or
 952                                 croak "$self->{db_path} not readable: $!\n";
 953        seek $fh, $offset, 2;
 954        $rl = readline $fh;
 955        defined $rl or return (undef, undef);
 956        chomp $rl;
 957        while ($c ne $rl && tell $fh != 0) {
 958                $offset -= 41;
 959                seek $fh, $offset, 2;
 960                $rl = readline $fh;
 961                defined $rl or return (undef, undef);
 962                chomp $rl;
 963        }
 964        my $rev = tell $fh;
 965        croak $! if ($rev < 0);
 966        $rev =  ($rev - 41) / 41;
 967        close $fh or croak $!;
 968        ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
 969        return ($rev, $c);
 970}
 971
 972sub get_fetch_range {
 973        my ($self, $min, $max) = @_;
 974        $max ||= $self->ra->get_latest_revnum;
 975        $min ||= $self->last_rev || 0;
 976        (++$min, $max);
 977}
 978
 979sub tmp_index_do {
 980        my ($self, $sub) = @_;
 981        my $old_index = $ENV{GIT_INDEX_FILE};
 982        $ENV{GIT_INDEX_FILE} = $self->{index};
 983        my @ret = &$sub;
 984        if ($old_index) {
 985                $ENV{GIT_INDEX_FILE} = $old_index;
 986        } else {
 987                delete $ENV{GIT_INDEX_FILE};
 988        }
 989        wantarray ? @ret : $ret[0];
 990}
 991
 992sub assert_index_clean {
 993        my ($self, $treeish) = @_;
 994
 995        $self->tmp_index_do(sub {
 996                command_noisy('read-tree', $treeish) unless -e $self->{index};
 997                my $x = command_oneline('write-tree');
 998                my ($y) = (command(qw/cat-file commit/, $treeish) =~
 999                           /^tree ($::sha1)/mo);
1000                if ($y ne $x) {
1001                        unlink $self->{index} or croak $!;
1002                        command_noisy('read-tree', $treeish);
1003                }
1004                $x = command_oneline('write-tree');
1005                if ($y ne $x) {
1006                        ::fatal "trees ($treeish) $y != $x\n",
1007                                "Something is seriously wrong...\n";
1008                }
1009        });
1010}
1011
1012sub get_commit_parents {
1013        my ($self, $log_entry, @parents) = @_;
1014        my (%seen, @ret, @tmp);
1015        # commit parents can be conditionally bound to a particular
1016        # svn revision via: "svn_revno=commit_sha1", filter them out here:
1017        foreach my $p (@parents) {
1018                next unless defined $p;
1019                if ($p =~ /^(\d+)=($::sha1_short)$/o) {
1020                        push @tmp, $2 if $1 == $log_entry->{revision};
1021                } else {
1022                        push @tmp, $p if $p =~ /^$::sha1_short$/o;
1023                }
1024        }
1025        if (my $cur = ::verify_ref($self->refname.'^0')) {
1026                push @tmp, $cur;
1027        }
1028        push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
1029        while (my $p = shift @tmp) {
1030                next if $seen{$p};
1031                $seen{$p} = 1;
1032                push @ret, $p;
1033                # MAXPARENT is defined to 16 in commit-tree.c:
1034                last if @ret >= 16;
1035        }
1036        if (@tmp) {
1037                die "r$log_entry->{revision}: No room for parents:\n\t",
1038                    join("\n\t", @tmp), "\n";
1039        }
1040        @ret;
1041}
1042
1043sub full_url {
1044        my ($self) = @_;
1045        $self->ra->{url} . (length $self->{path} ? '/' . $self->{path} : '');
1046}
1047
1048sub do_git_commit {
1049        my ($self, $log_entry, @parents) = @_;
1050        if (my $c = $self->rev_db_get($log_entry->{revision})) {
1051                croak "$log_entry->{revision} = $c already exists! ",
1052                      "Why are we refetching it?\n";
1053        }
1054        my $author = $log_entry->{author};
1055        my ($name, $email) = (defined $::users{$author} ? @{$::users{$author}}
1056                           : ($author, "$author\@".$self->ra->uuid));
1057        $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} = $name;
1058        $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} = $email;
1059        $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
1060
1061        my $tree = $log_entry->{tree};
1062        if (!defined $tree) {
1063                $tree = $self->tmp_index_do(sub {
1064                                            command_oneline('write-tree') });
1065        }
1066        die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
1067
1068        my @exec = ('git-commit-tree', $tree);
1069        foreach ($self->get_commit_parents($log_entry, @parents)) {
1070                push @exec, '-p', $_;
1071        }
1072        defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
1073                                                                   or croak $!;
1074        print $msg_fh $log_entry->{log} or croak $!;
1075        print $msg_fh "\ngit-svn-id: ", $self->full_url, '@',
1076                      $log_entry->{revision}, ' ',
1077                      $self->ra->uuid, "\n" or croak $!;
1078        $msg_fh->flush == 0 or croak $!;
1079        close $msg_fh or croak $!;
1080        chomp(my $commit = do { local $/; <$out_fh> });
1081        close $out_fh or croak $!;
1082        waitpid $pid, 0;
1083        croak $? if $?;
1084        if ($commit !~ /^$::sha1$/o) {
1085                die "Failed to commit, invalid sha1: $commit\n";
1086        }
1087
1088        command_noisy('update-ref',$self->refname, $commit);
1089        $self->rev_db_set($log_entry->{revision}, $commit);
1090
1091        $self->{last_rev} = $log_entry->{revision};
1092        $self->{last_commit} = $commit;
1093        print "r$log_entry->{revision} = $commit\n";
1094        return $commit;
1095}
1096
1097sub revisions_eq {
1098        my ($self, $r0, $r1) = @_;
1099        return 1 if $r0 == $r1;
1100        my $nr = 0;
1101        $self->ra->get_log([$self->{path}], $r0, $r1,
1102                           0, 0, 1, sub { $nr++ });
1103        return 0 if ($nr > 1);
1104        return 1;
1105}
1106
1107sub find_parent_branch {
1108        my ($self, $paths, $rev) = @_;
1109        return undef unless $::_follow_parent;
1110        unless (defined $paths) {
1111                $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1,
1112                                   sub { $paths = dup_changed_paths($_[0]) });
1113        }
1114        return undef unless defined $paths;
1115
1116        # look for a parent from another branch:
1117        my @b_path_components = split m#/#, $self->rel_path;
1118        my @a_path_components;
1119        my $i;
1120        while (@b_path_components) {
1121                $i = $paths->{'/'.join('/', @b_path_components)};
1122                last if $i;
1123                unshift(@a_path_components, pop(@b_path_components));
1124        }
1125        goto not_found unless defined $i;
1126        my $branch_from = $i->{copyfrom_path} or goto not_found;
1127        if (@a_path_components) {
1128                print STDERR "branch_from: $branch_from => ";
1129                $branch_from .= '/'.join('/', @a_path_components);
1130                print STDERR $branch_from, "\n";
1131        }
1132        my $r = $i->{copyfrom_rev};
1133        my $repos_root = $self->ra->{repos_root};
1134        my $url = $self->ra->{url};
1135        my $new_url = $repos_root . $branch_from;
1136        print STDERR  "Found possible branch point: ",
1137                      "$new_url => ", $self->full_url, ", $r\n";
1138        $branch_from =~ s#^/##;
1139        my $remotes = read_all_remotes();
1140        my $gs;
1141        foreach my $repo_id (keys %$remotes) {
1142                my $u = $remotes->{$repo_id}->{url} or next;
1143                next if $url ne $u;
1144                my $fetch = $remotes->{$repo_id}->{fetch};
1145                foreach my $f (keys %$fetch) {
1146                        next if $f ne $branch_from;
1147                        $gs = Git::SVN->new($fetch->{$f}, $repo_id, $f);
1148                        last;
1149                }
1150                last if $gs;
1151        }
1152        unless ($gs) {
1153                my $ref_id = $branch_from;
1154                $ref_id .= "\@$r" if find_ref($ref_id);
1155                # just grow a tail if we're not unique enough :x
1156                $ref_id .= '-' while find_ref($ref_id);
1157                $gs = Git::SVN->init($new_url, '', $ref_id, $ref_id);
1158        }
1159        my ($r0, $parent) = $gs->find_rev_before($r, 1);
1160        if ($::_follow_parent && (!defined $r0 || !defined $parent)) {
1161                $gs->fetch(0, $r);
1162                ($r0, $parent) = $gs->last_rev_commit;
1163        }
1164        if (defined $r0 && defined $parent && $gs->revisions_eq($r0, $r)) {
1165                print STDERR "Found branch parent: ($self->{ref_id}) $parent\n";
1166                $self->assert_index_clean($parent);
1167                my $ed;
1168                if ($self->ra->can_do_switch) {
1169                        print STDERR "Following parent with do_switch\n";
1170                        # do_switch works with svn/trunk >= r22312, but that
1171                        # is not included with SVN 1.4.2 (the latest version
1172                        # at the moment), so we can't rely on it
1173                        $self->{last_commit} = $parent;
1174                        $ed = SVN::Git::Fetcher->new($self);
1175                        $gs->ra->gs_do_switch($r0, $rev, $gs->{path}, 1,
1176                                              $self->full_url, $ed)
1177                          or die "SVN connection failed somewhere...\n";
1178                } else {
1179                        print STDERR "Following parent with do_update\n";
1180                        $ed = SVN::Git::Fetcher->new($self);
1181                        $self->ra->gs_do_update($rev, $rev, $self->{path},
1182                                                1, $ed)
1183                          or die "SVN connection failed somewhere...\n";
1184                }
1185                return $self->make_log_entry($rev, [$parent], $ed);
1186        }
1187not_found:
1188        print STDERR "Branch parent for path: '/",
1189                     $self->rel_path, "' @ r$rev not found:\n";
1190        return undef unless $paths;
1191        print STDERR "Changed paths:\n";
1192        foreach my $x (sort keys %$paths) {
1193                my $p = $paths->{$x};
1194                print STDERR "\t$p->{action}\t$x";
1195                if ($p->{copyfrom_path}) {
1196                        print STDERR "(from $p->{copyfrom_path}: ",
1197                                     "$p->{copyfrom_rev})";
1198                }
1199                print STDERR "\n";
1200        }
1201        print STDERR '-'x72, "\n";
1202        return undef;
1203}
1204
1205sub do_fetch {
1206        my ($self, $paths, $rev) = @_;
1207        my $ed;
1208        my ($last_rev, @parents);
1209        if ($self->{last_commit}) {
1210                $ed = SVN::Git::Fetcher->new($self);
1211                $last_rev = $self->{last_rev};
1212                $ed->{c} = $self->{last_commit};
1213                @parents = ($self->{last_commit});
1214        } else {
1215                $last_rev = $rev;
1216                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
1217                        return $log_entry;
1218                }
1219                $ed = SVN::Git::Fetcher->new($self);
1220        }
1221        unless ($self->ra->gs_do_update($last_rev, $rev,
1222                                        $self->{path}, 1, $ed)) {
1223                die "SVN connection failed somewhere...\n";
1224        }
1225        $self->make_log_entry($rev, \@parents, $ed);
1226}
1227
1228sub get_untracked {
1229        my ($self, $ed) = @_;
1230        my @out;
1231        my $h = $ed->{empty};
1232        foreach (sort keys %$h) {
1233                my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
1234                push @out, "  $act: " . uri_encode($_);
1235                warn "W: $act: $_\n";
1236        }
1237        foreach my $t (qw/dir_prop file_prop/) {
1238                $h = $ed->{$t} or next;
1239                foreach my $path (sort keys %$h) {
1240                        my $ppath = $path eq '' ? '.' : $path;
1241                        foreach my $prop (sort keys %{$h->{$path}}) {
1242                                next if $SKIP_PROP{$prop};
1243                                my $v = $h->{$path}->{$prop};
1244                                my $t_ppath_prop = "$t: " .
1245                                                    uri_encode($ppath) . ' ' .
1246                                                    uri_encode($prop);
1247                                if (defined $v) {
1248                                        push @out, "  +$t_ppath_prop " .
1249                                                   uri_encode($v);
1250                                } else {
1251                                        push @out, "  -$t_ppath_prop";
1252                                }
1253                        }
1254                }
1255        }
1256        foreach my $t (qw/absent_file absent_directory/) {
1257                $h = $ed->{$t} or next;
1258                foreach my $parent (sort keys %$h) {
1259                        foreach my $path (sort @{$h->{$parent}}) {
1260                                push @out, "  $t: " .
1261                                           uri_encode("$parent/$path");
1262                                warn "W: $t: $parent/$path ",
1263                                     "Insufficient permissions?\n";
1264                        }
1265                }
1266        }
1267        \@out;
1268}
1269
1270sub parse_svn_date {
1271        my $date = shift || return '+0000 1970-01-01 00:00:00';
1272        my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
1273                                            (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x) or
1274                                         croak "Unable to parse date: $date\n";
1275        "+0000 $Y-$m-$d $H:$M:$S";
1276}
1277
1278sub check_author {
1279        my ($author) = @_;
1280        if (!defined $author || length $author == 0) {
1281                $author = '(no author)';
1282        }
1283        if (defined $::_authors && ! defined $::users{$author}) {
1284                die "Author: $author not defined in $::_authors file\n";
1285        }
1286        $author;
1287}
1288
1289sub make_log_entry {
1290        my ($self, $rev, $parents, $ed) = @_;
1291        my $untracked = $self->get_untracked($ed);
1292
1293        return undef if ($ed->{nr} == 0 && scalar @$untracked == 0);
1294
1295        open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
1296        print $un "r$rev\n" or croak $!;
1297        print $un $_, "\n" foreach @$untracked;
1298        my %log_entry = ( parents => $parents || [], revision => $rev,
1299                          log => '');
1300        my $rp = $self->ra->rev_proplist($rev);
1301        foreach (sort keys %$rp) {
1302                my $v = $rp->{$_};
1303                if (/^svn:(author|date|log)$/) {
1304                        $log_entry{$1} = $v;
1305                } else {
1306                        print $un "  rev_prop: ", uri_encode($_), ' ',
1307                                  uri_encode($v), "\n";
1308                }
1309        }
1310        close $un or croak $!;
1311
1312        $log_entry{date} = parse_svn_date($log_entry{date});
1313        $log_entry{author} = check_author($log_entry{author});
1314        $log_entry{log} .= "\n";
1315        \%log_entry;
1316}
1317
1318sub fetch {
1319        my ($self, $min_rev, $max_rev, @parents) = @_;
1320        my ($last_rev, $last_commit) = $self->last_rev_commit;
1321        my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
1322        return if ($base > $head);
1323        if (defined $last_commit) {
1324                $self->assert_index_clean($last_commit);
1325        }
1326        my $inc = 1000;
1327        my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
1328        my $err_handler = $SVN::Error::handler;
1329        my $err;
1330        $SVN::Error::handler = sub { ($err) = @_; skip_unknown_revs($err); } ;
1331        while (1) {
1332                my @revs;
1333                $self->ra->get_log([$self->{path}], $min, $max, 0, 1, 1,
1334                    sub {
1335                        my ($paths, $rev) = @_;
1336                        push @revs, [ dup_changed_paths($paths), $rev ];
1337                        });
1338                if (! @revs && $err && $max >= $head) {
1339                        print STDERR "Branch probably deleted:\n  ",
1340                                     $err->expanded_message,
1341                                     "\nWill attempt to follow revisions ",
1342                                     "r$min .. r$max",
1343                                     "committed before the deletion\n";
1344                        @revs = map { [ undef, $_ ] } ($min .. $max);
1345                }
1346                foreach (@revs) {
1347                        if (my $log_entry = $self->do_fetch(@$_)) {
1348                                $self->do_git_commit($log_entry, @parents);
1349                        }
1350                }
1351                last if $max >= $head;
1352                $min = $max + 1;
1353                $max += $inc;
1354                $max = $head if ($max > $head);
1355        }
1356        $SVN::Error::handler = $err_handler;
1357}
1358
1359sub set_tree_cb {
1360        my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
1361        # TODO: enable and test optimized commits:
1362        if (0 && $rev == ($self->{last_rev} + 1)) {
1363                $log_entry->{revision} = $rev;
1364                $log_entry->{author} = $author;
1365                $self->do_git_commit($log_entry, "$rev=$tree");
1366        } else {
1367                $self->fetch(undef, undef, "$rev=$tree");
1368        }
1369}
1370
1371sub set_tree {
1372        my ($self, $tree) = (shift, shift);
1373        my $log_entry = ::get_commit_entry($tree);
1374        unless ($self->{last_rev}) {
1375                fatal("Must have an existing revision to commit\n");
1376        }
1377        my $pool = SVN::Pool->new;
1378        my $ed = SVN::Git::Editor->new({ r => $self->{last_rev},
1379                                         ra => $self->ra->dup,
1380                                         svn_path => $self->{path}
1381                                       },
1382                                       $self->ra->get_commit_editor(
1383                                         $log_entry->{log}, sub {
1384                                           $self->set_tree_cb($log_entry,
1385                                                              $tree, @_);
1386                                       }),
1387                                       $pool);
1388        my $mods = $ed->apply_diff($self->{last_commit}, $tree);
1389        if (@$mods == 0) {
1390                print "No changes\nr$self->{last_rev} = $tree\n";
1391        }
1392        $pool->clear;
1393}
1394
1395sub skip_unknown_revs {
1396        my ($err) = @_;
1397        my $errno = $err->apr_err();
1398        # Maybe the branch we're tracking didn't
1399        # exist when the repo started, so it's
1400        # not an error if it doesn't, just continue
1401        #
1402        # Wonderfully consistent library, eh?
1403        # 160013 - svn:// and file://
1404        # 175002 - http(s)://
1405        # 175007 - http(s):// (this repo required authorization, too...)
1406        #   More codes may be discovered later...
1407        if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
1408                return;
1409        }
1410        croak "Error from SVN, ($errno): ", $err->expanded_message,"\n";
1411}
1412
1413# svn_log_changed_path_t objects passed to get_log are likely to be
1414# overwritten even if only the refs are copied to an external variable,
1415# so we should dup the structures in their entirety.  Using an externally
1416# passed pool (instead of our temporary and quickly cleared pool in
1417# Git::SVN::Ra) does not help matters at all...
1418sub dup_changed_paths {
1419        my ($paths) = @_;
1420        return undef unless $paths;
1421        my %ret;
1422        foreach my $p (keys %$paths) {
1423                my $i = $paths->{$p};
1424                my %s = map { $_ => $i->$_ }
1425                              qw/copyfrom_path copyfrom_rev action/;
1426                $ret{$p} = \%s;
1427        }
1428        \%ret;
1429}
1430
1431# rev_db:
1432# Tie::File seems to be prone to offset errors if revisions get sparse,
1433# it's not that fast, either.  Tie::File is also not in Perl 5.6.  So
1434# one of my favorite modules is out :<  Next up would be one of the DBM
1435# modules, but I'm not sure which is most portable...  So I'll just
1436# go with something that's plain-text, but still capable of
1437# being randomly accessed.  So here's my ultra-simple fixed-width
1438# database.  All records are 40 characters + "\n", so it's easy to seek
1439# to a revision: (41 * rev) is the byte offset.
1440# A record of 40 0s denotes an empty revision.
1441# And yes, it's still pretty fast (faster than Tie::File).
1442
1443sub rev_db_set {
1444        my ($self, $rev, $commit) = @_;
1445        length $commit == 40 or croak "arg3 must be a full SHA1 hexsum\n";
1446        open my $fh, '+<', $self->{db_path} or croak $!;
1447        my $offset = $rev * 41;
1448        # assume that append is the common case:
1449        seek $fh, 0, 2 or croak $!;
1450        my $pos = tell $fh;
1451        if ($pos < $offset) {
1452                print $fh (('0' x 40),"\n") x (($offset - $pos) / 41)
1453                  or croak $!;
1454        }
1455        seek $fh, $offset, 0 or croak $!;
1456        print $fh $commit,"\n" or croak $!;
1457        close $fh or croak $!;
1458}
1459
1460sub rev_db_get {
1461        my ($self, $rev) = @_;
1462        my $ret;
1463        my $offset = $rev * 41;
1464        open my $fh, '<', $self->{db_path} or croak $!;
1465        if (seek $fh, $offset, 0) {
1466                $ret = readline $fh;
1467                if (defined $ret) {
1468                        chomp $ret;
1469                        $ret = undef if ($ret =~ /^0{40}$/);
1470                }
1471        }
1472        close $fh or croak $!;
1473        $ret;
1474}
1475
1476sub find_rev_before {
1477        my ($self, $rev, $eq_ok) = @_;
1478        --$rev unless $eq_ok;
1479        while ($rev > 0) {
1480                if (my $c = $self->rev_db_get($rev)) {
1481                        return ($rev, $c);
1482                }
1483                --$rev;
1484        }
1485        return (undef, undef);
1486}
1487
1488sub _new {
1489        my ($class, $repo_id, $ref_id, $path) = @_;
1490        unless (defined $repo_id && length $repo_id) {
1491                $repo_id = $Git::SVN::default_repo_id;
1492        }
1493        unless (defined $ref_id && length $ref_id) {
1494                $_[2] = $ref_id = $Git::SVN::default_ref_id;
1495        }
1496        $_[1] = $repo_id = sanitize_remote_name($repo_id);
1497        my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
1498        $_[3] = $path = '' unless (defined $path);
1499        mkpath([$dir]);
1500        unless (-f "$dir/.rev_db") {
1501                open my $fh, '>>', "$dir/.rev_db" or croak $!;
1502                close $fh or croak $!;
1503        }
1504        bless { ref_id => $ref_id, dir => $dir, index => "$dir/index",
1505                path => $path,
1506                db_path => "$dir/.rev_db", repo_id => $repo_id }, $class;
1507}
1508
1509sub uri_encode {
1510        my ($f) = @_;
1511        $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
1512        $f
1513}
1514
1515package Git::SVN::Prompt;
1516use strict;
1517use warnings;
1518require SVN::Core;
1519use vars qw/$_no_auth_cache $_username/;
1520
1521sub simple {
1522        my ($cred, $realm, $default_username, $may_save, $pool) = @_;
1523        $may_save = undef if $_no_auth_cache;
1524        $default_username = $_username if defined $_username;
1525        if (defined $default_username && length $default_username) {
1526                if (defined $realm && length $realm) {
1527                        print STDERR "Authentication realm: $realm\n";
1528                        STDERR->flush;
1529                }
1530                $cred->username($default_username);
1531        } else {
1532                username($cred, $realm, $may_save, $pool);
1533        }
1534        $cred->password(_read_password("Password for '" .
1535                                       $cred->username . "': ", $realm));
1536        $cred->may_save($may_save);
1537        $SVN::_Core::SVN_NO_ERROR;
1538}
1539
1540sub ssl_server_trust {
1541        my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
1542        $may_save = undef if $_no_auth_cache;
1543        print STDERR "Error validating server certificate for '$realm':\n";
1544        if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
1545                print STDERR " - The certificate is not issued by a trusted ",
1546                      "authority. Use the\n",
1547                      "   fingerprint to validate the certificate manually!\n";
1548        }
1549        if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
1550                print STDERR " - The certificate hostname does not match.\n";
1551        }
1552        if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
1553                print STDERR " - The certificate is not yet valid.\n";
1554        }
1555        if ($failures & $SVN::Auth::SSL::EXPIRED) {
1556                print STDERR " - The certificate has expired.\n";
1557        }
1558        if ($failures & $SVN::Auth::SSL::OTHER) {
1559                print STDERR " - The certificate has an unknown error.\n";
1560        }
1561        printf STDERR
1562                "Certificate information:\n".
1563                " - Hostname: %s\n".
1564                " - Valid: from %s until %s\n".
1565                " - Issuer: %s\n".
1566                " - Fingerprint: %s\n",
1567                map $cert_info->$_, qw(hostname valid_from valid_until
1568                                       issuer_dname fingerprint);
1569        my $choice;
1570prompt:
1571        print STDERR $may_save ?
1572              "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
1573              "(R)eject or accept (t)emporarily? ";
1574        STDERR->flush;
1575        $choice = lc(substr(<STDIN> || 'R', 0, 1));
1576        if ($choice =~ /^t$/i) {
1577                $cred->may_save(undef);
1578        } elsif ($choice =~ /^r$/i) {
1579                return -1;
1580        } elsif ($may_save && $choice =~ /^p$/i) {
1581                $cred->may_save($may_save);
1582        } else {
1583                goto prompt;
1584        }
1585        $cred->accepted_failures($failures);
1586        $SVN::_Core::SVN_NO_ERROR;
1587}
1588
1589sub ssl_client_cert {
1590        my ($cred, $realm, $may_save, $pool) = @_;
1591        $may_save = undef if $_no_auth_cache;
1592        print STDERR "Client certificate filename: ";
1593        STDERR->flush;
1594        chomp(my $filename = <STDIN>);
1595        $cred->cert_file($filename);
1596        $cred->may_save($may_save);
1597        $SVN::_Core::SVN_NO_ERROR;
1598}
1599
1600sub ssl_client_cert_pw {
1601        my ($cred, $realm, $may_save, $pool) = @_;
1602        $may_save = undef if $_no_auth_cache;
1603        $cred->password(_read_password("Password: ", $realm));
1604        $cred->may_save($may_save);
1605        $SVN::_Core::SVN_NO_ERROR;
1606}
1607
1608sub username {
1609        my ($cred, $realm, $may_save, $pool) = @_;
1610        $may_save = undef if $_no_auth_cache;
1611        if (defined $realm && length $realm) {
1612                print STDERR "Authentication realm: $realm\n";
1613        }
1614        my $username;
1615        if (defined $_username) {
1616                $username = $_username;
1617        } else {
1618                print STDERR "Username: ";
1619                STDERR->flush;
1620                chomp($username = <STDIN>);
1621        }
1622        $cred->username($username);
1623        $cred->may_save($may_save);
1624        $SVN::_Core::SVN_NO_ERROR;
1625}
1626
1627sub _read_password {
1628        my ($prompt, $realm) = @_;
1629        print STDERR $prompt;
1630        STDERR->flush;
1631        require Term::ReadKey;
1632        Term::ReadKey::ReadMode('noecho');
1633        my $password = '';
1634        while (defined(my $key = Term::ReadKey::ReadKey(0))) {
1635                last if $key =~ /[\012\015]/; # \n\r
1636                $password .= $key;
1637        }
1638        Term::ReadKey::ReadMode('restore');
1639        print STDERR "\n";
1640        STDERR->flush;
1641        $password;
1642}
1643
1644package main;
1645
1646sub uri_encode {
1647        my ($f) = @_;
1648        $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
1649        $f
1650}
1651
1652sub uri_decode {
1653        my ($f) = @_;
1654        $f =~ tr/+/ /;
1655        $f =~ s/%([A-F0-9]{2})/chr hex($1)/ge;
1656        $f
1657}
1658
1659{
1660        my $kill_stupid_warnings = $SVN::Node::none.$SVN::Node::file.
1661                                $SVN::Node::dir.$SVN::Node::unknown.
1662                                $SVN::Node::none.$SVN::Node::file.
1663                                $SVN::Node::dir.$SVN::Node::unknown.
1664                                $SVN::Auth::SSL::CNMISMATCH.
1665                                $SVN::Auth::SSL::NOTYETVALID.
1666                                $SVN::Auth::SSL::EXPIRED.
1667                                $SVN::Auth::SSL::UNKNOWNCA.
1668                                $SVN::Auth::SSL::OTHER;
1669}
1670
1671package SVN::Git::Fetcher;
1672use vars qw/@ISA/;
1673use strict;
1674use warnings;
1675use Carp qw/croak/;
1676use IO::File qw//;
1677
1678# file baton members: path, mode_a, mode_b, pool, fh, blob, base
1679sub new {
1680        my ($class, $git_svn) = @_;
1681        my $self = SVN::Delta::Editor->new;
1682        bless $self, $class;
1683        $self->{c} = $git_svn->{last_commit} if exists $git_svn->{last_commit};
1684        $self->{empty} = {};
1685        $self->{dir_prop} = {};
1686        $self->{file_prop} = {};
1687        $self->{absent_dir} = {};
1688        $self->{absent_file} = {};
1689        $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
1690        require Digest::MD5;
1691        $self;
1692}
1693
1694sub set_path_strip {
1695        my ($self, $path) = @_;
1696        $self->{path_strip} = qr/^\Q$path\E\/?/;
1697}
1698
1699sub open_root {
1700        { path => '' };
1701}
1702
1703sub open_directory {
1704        my ($self, $path, $pb, $rev) = @_;
1705        { path => $path };
1706}
1707
1708sub git_path {
1709        my ($self, $path) = @_;
1710        $path =~ s!$self->{path_strip}!! if $self->{path_strip};
1711        $path;
1712}
1713
1714sub delete_entry {
1715        my ($self, $path, $rev, $pb) = @_;
1716
1717        my $gpath = $self->git_path($path);
1718        # remove entire directories.
1719        if (command('ls-tree', $self->{c}, '--', $gpath) =~ /^040000 tree/) {
1720                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
1721                                                     -r --name-only -z/,
1722                                                     $self->{c}, '--', $gpath);
1723                local $/ = "\0";
1724                while (<$ls>) {
1725                        chomp;
1726                        $self->{gii}->remove($_);
1727                        print "\tD\t$_\n" unless $self->{q};
1728                }
1729                print "\tD\t$gpath/\n" unless $self->{q};
1730                command_close_pipe($ls, $ctx);
1731                $self->{empty}->{$path} = 0
1732        } else {
1733                $self->{gii}->remove($gpath);
1734                print "\tD\t$gpath\n" unless $self->{q};
1735        }
1736        undef;
1737}
1738
1739sub open_file {
1740        my ($self, $path, $pb, $rev) = @_;
1741        my $gpath = $self->git_path($path);
1742        my ($mode, $blob) = (command('ls-tree', $self->{c}, '--', $gpath)
1743                             =~ /^(\d{6}) blob ([a-f\d]{40})\t/);
1744        unless (defined $mode && defined $blob) {
1745                die "$path was not found in commit $self->{c} (r$rev)\n";
1746        }
1747        { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
1748          pool => SVN::Pool->new, action => 'M' };
1749}
1750
1751sub add_file {
1752        my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
1753        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
1754        delete $self->{empty}->{$dir};
1755        { path => $path, mode_a => 100644, mode_b => 100644,
1756          pool => SVN::Pool->new, action => 'A' };
1757}
1758
1759sub add_directory {
1760        my ($self, $path, $cp_path, $cp_rev) = @_;
1761        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
1762        delete $self->{empty}->{$dir};
1763        $self->{empty}->{$path} = 1;
1764        { path => $path };
1765}
1766
1767sub change_dir_prop {
1768        my ($self, $db, $prop, $value) = @_;
1769        $self->{dir_prop}->{$db->{path}} ||= {};
1770        $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
1771        undef;
1772}
1773
1774sub absent_directory {
1775        my ($self, $path, $pb) = @_;
1776        $self->{absent_dir}->{$pb->{path}} ||= [];
1777        push @{$self->{absent_dir}->{$pb->{path}}}, $path;
1778        undef;
1779}
1780
1781sub absent_file {
1782        my ($self, $path, $pb) = @_;
1783        $self->{absent_file}->{$pb->{path}} ||= [];
1784        push @{$self->{absent_file}->{$pb->{path}}}, $path;
1785        undef;
1786}
1787
1788sub change_file_prop {
1789        my ($self, $fb, $prop, $value) = @_;
1790        if ($prop eq 'svn:executable') {
1791                if ($fb->{mode_b} != 120000) {
1792                        $fb->{mode_b} = defined $value ? 100755 : 100644;
1793                }
1794        } elsif ($prop eq 'svn:special') {
1795                $fb->{mode_b} = defined $value ? 120000 : 100644;
1796        } else {
1797                $self->{file_prop}->{$fb->{path}} ||= {};
1798                $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
1799        }
1800        undef;
1801}
1802
1803sub apply_textdelta {
1804        my ($self, $fb, $exp) = @_;
1805        my $fh = IO::File->new_tmpfile;
1806        $fh->autoflush(1);
1807        # $fh gets auto-closed() by SVN::TxDelta::apply(),
1808        # (but $base does not,) so dup() it for reading in close_file
1809        open my $dup, '<&', $fh or croak $!;
1810        my $base = IO::File->new_tmpfile;
1811        $base->autoflush(1);
1812        if ($fb->{blob}) {
1813                defined (my $pid = fork) or croak $!;
1814                if (!$pid) {
1815                        open STDOUT, '>&', $base or croak $!;
1816                        print STDOUT 'link ' if ($fb->{mode_a} == 120000);
1817                        exec qw/git-cat-file blob/, $fb->{blob} or croak $!;
1818                }
1819                waitpid $pid, 0;
1820                croak $? if $?;
1821
1822                if (defined $exp) {
1823                        seek $base, 0, 0 or croak $!;
1824                        my $md5 = Digest::MD5->new;
1825                        $md5->addfile($base);
1826                        my $got = $md5->hexdigest;
1827                        die "Checksum mismatch: $fb->{path} $fb->{blob}\n",
1828                            "expected: $exp\n",
1829                            "     got: $got\n" if ($got ne $exp);
1830                }
1831        }
1832        seek $base, 0, 0 or croak $!;
1833        $fb->{fh} = $dup;
1834        $fb->{base} = $base;
1835        [ SVN::TxDelta::apply($base, $fh, undef, $fb->{path}, $fb->{pool}) ];
1836}
1837
1838sub close_file {
1839        my ($self, $fb, $exp) = @_;
1840        my $hash;
1841        my $path = $self->git_path($fb->{path});
1842        if (my $fh = $fb->{fh}) {
1843                seek($fh, 0, 0) or croak $!;
1844                my $md5 = Digest::MD5->new;
1845                $md5->addfile($fh);
1846                my $got = $md5->hexdigest;
1847                die "Checksum mismatch: $path\n",
1848                    "expected: $exp\n    got: $got\n" if ($got ne $exp);
1849                seek($fh, 0, 0) or croak $!;
1850                if ($fb->{mode_b} == 120000) {
1851                        read($fh, my $buf, 5) == 5 or croak $!;
1852                        $buf eq 'link ' or die "$path has mode 120000",
1853                                               "but is not a link\n";
1854                }
1855                defined(my $pid = open my $out,'-|') or die "Can't fork: $!\n";
1856                if (!$pid) {
1857                        open STDIN, '<&', $fh or croak $!;
1858                        exec qw/git-hash-object -w --stdin/ or croak $!;
1859                }
1860                chomp($hash = do { local $/; <$out> });
1861                close $out or croak $!;
1862                close $fh or croak $!;
1863                $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
1864                close $fb->{base} or croak $!;
1865        } else {
1866                $hash = $fb->{blob} or die "no blob information\n";
1867        }
1868        $fb->{pool}->clear;
1869        $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
1870        print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $self->{q};
1871        undef;
1872}
1873
1874sub abort_edit {
1875        my $self = shift;
1876        $self->{nr} = $self->{gii}->{nr};
1877        delete $self->{gii};
1878        $self->SUPER::abort_edit(@_);
1879}
1880
1881sub close_edit {
1882        my $self = shift;
1883        $self->{git_commit_ok} = 1;
1884        $self->{nr} = $self->{gii}->{nr};
1885        delete $self->{gii};
1886        $self->SUPER::close_edit(@_);
1887}
1888
1889package SVN::Git::Editor;
1890use vars qw/@ISA/;
1891use strict;
1892use warnings;
1893use Carp qw/croak/;
1894use IO::File;
1895
1896sub new {
1897        my $class = shift;
1898        my $git_svn = shift;
1899        my $self = SVN::Delta::Editor->new(@_);
1900        bless $self, $class;
1901        foreach (qw/svn_path r ra/) {
1902                die "$_ required!\n" unless (defined $git_svn->{$_});
1903                $self->{$_} = $git_svn->{$_};
1904        }
1905        $self->{pool} = SVN::Pool->new;
1906        $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
1907        $self->{rm} = { };
1908        $self->{path_prefix} = length $self->{svn_path} ?
1909                               "$self->{svn_path}/" : '';
1910        require Digest::MD5;
1911        return $self;
1912}
1913
1914sub split_path {
1915        return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
1916}
1917
1918sub repo_path {
1919        my ($self, $path) = @_;
1920        $self->{path_prefix}.(defined $path ? $path : '');
1921}
1922
1923sub url_path {
1924        my ($self, $path) = @_;
1925        $self->{ra}->{url} . '/' . $self->repo_path($path);
1926}
1927
1928sub rmdirs {
1929        my ($self, $tree_b) = @_;
1930        my $rm = $self->{rm};
1931        delete $rm->{''}; # we never delete the url we're tracking
1932        return unless %$rm;
1933
1934        foreach (keys %$rm) {
1935                my @d = split m#/#, $_;
1936                my $c = shift @d;
1937                $rm->{$c} = 1;
1938                while (@d) {
1939                        $c .= '/' . shift @d;
1940                        $rm->{$c} = 1;
1941                }
1942        }
1943        delete $rm->{$self->{svn_path}};
1944        delete $rm->{''}; # we never delete the url we're tracking
1945        return unless %$rm;
1946
1947        my ($fh, $ctx) = command_output_pipe(
1948                                   qw/ls-tree --name-only -r -z/, $tree_b);
1949        local $/ = "\0";
1950        while (<$fh>) {
1951                chomp;
1952                my @dn = split m#/#, $_;
1953                while (pop @dn) {
1954                        delete $rm->{join '/', @dn};
1955                }
1956                unless (%$rm) {
1957                        close $fh;
1958                        return;
1959                }
1960        }
1961        command_close_pipe($fh, $ctx);
1962
1963        my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
1964        foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
1965                $self->close_directory($bat->{$d}, $p);
1966                my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
1967                print "\tD+\t$d/\n" unless $::_q;
1968                $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
1969                delete $bat->{$d};
1970        }
1971}
1972
1973sub open_or_add_dir {
1974        my ($self, $full_path, $baton) = @_;
1975        my $t = $self->{ra}->check_path($full_path, $self->{r});
1976        if ($t == $SVN::Node::none) {
1977                return $self->add_directory($full_path, $baton,
1978                                                undef, -1, $self->{pool});
1979        } elsif ($t == $SVN::Node::dir) {
1980                return $self->open_directory($full_path, $baton,
1981                                                $self->{r}, $self->{pool});
1982        }
1983        print STDERR "$full_path already exists in repository at ",
1984                "r$self->{r} and it is not a directory (",
1985                ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
1986        exit 1;
1987}
1988
1989sub ensure_path {
1990        my ($self, $path) = @_;
1991        my $bat = $self->{bat};
1992        $path = $self->repo_path($path);
1993        return $bat->{''} unless (length $path);
1994        my @p = split m#/+#, $path;
1995        my $c = shift @p;
1996        $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
1997        while (@p) {
1998                my $c0 = $c;
1999                $c .= '/' . shift @p;
2000                $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
2001        }
2002        return $bat->{$c};
2003}
2004
2005sub A {
2006        my ($self, $m) = @_;
2007        my ($dir, $file) = split_path($m->{file_b});
2008        my $pbat = $self->ensure_path($dir);
2009        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2010                                        undef, -1);
2011        print "\tA\t$m->{file_b}\n" unless $::_q;
2012        $self->chg_file($fbat, $m);
2013        $self->close_file($fbat,undef,$self->{pool});
2014}
2015
2016sub C {
2017        my ($self, $m) = @_;
2018        my ($dir, $file) = split_path($m->{file_b});
2019        my $pbat = $self->ensure_path($dir);
2020        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2021                                $self->url_path($m->{file_a}), $self->{r});
2022        print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
2023        $self->chg_file($fbat, $m);
2024        $self->close_file($fbat,undef,$self->{pool});
2025}
2026
2027sub delete_entry {
2028        my ($self, $path, $pbat) = @_;
2029        my $rpath = $self->repo_path($path);
2030        my ($dir, $file) = split_path($rpath);
2031        $self->{rm}->{$dir} = 1;
2032        $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
2033}
2034
2035sub R {
2036        my ($self, $m) = @_;
2037        my ($dir, $file) = split_path($m->{file_b});
2038        my $pbat = $self->ensure_path($dir);
2039        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2040                                $self->url_path($m->{file_a}), $self->{r});
2041        print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
2042        $self->chg_file($fbat, $m);
2043        $self->close_file($fbat,undef,$self->{pool});
2044
2045        ($dir, $file) = split_path($m->{file_a});
2046        $pbat = $self->ensure_path($dir);
2047        $self->delete_entry($m->{file_a}, $pbat);
2048}
2049
2050sub M {
2051        my ($self, $m) = @_;
2052        my ($dir, $file) = split_path($m->{file_b});
2053        my $pbat = $self->ensure_path($dir);
2054        my $fbat = $self->open_file($self->repo_path($m->{file_b}),
2055                                $pbat,$self->{r},$self->{pool});
2056        print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
2057        $self->chg_file($fbat, $m);
2058        $self->close_file($fbat,undef,$self->{pool});
2059}
2060
2061sub T { shift->M(@_) }
2062
2063sub change_file_prop {
2064        my ($self, $fbat, $pname, $pval) = @_;
2065        $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
2066}
2067
2068sub chg_file {
2069        my ($self, $fbat, $m) = @_;
2070        if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
2071                $self->change_file_prop($fbat,'svn:executable','*');
2072        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
2073                $self->change_file_prop($fbat,'svn:executable',undef);
2074        }
2075        my $fh = IO::File->new_tmpfile or croak $!;
2076        if ($m->{mode_b} =~ /^120/) {
2077                print $fh 'link ' or croak $!;
2078                $self->change_file_prop($fbat,'svn:special','*');
2079        } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
2080                $self->change_file_prop($fbat,'svn:special',undef);
2081        }
2082        defined(my $pid = fork) or croak $!;
2083        if (!$pid) {
2084                open STDOUT, '>&', $fh or croak $!;
2085                exec qw/git-cat-file blob/, $m->{sha1_b} or croak $!;
2086        }
2087        waitpid $pid, 0;
2088        croak $? if $?;
2089        $fh->flush == 0 or croak $!;
2090        seek $fh, 0, 0 or croak $!;
2091
2092        my $md5 = Digest::MD5->new;
2093        $md5->addfile($fh) or croak $!;
2094        seek $fh, 0, 0 or croak $!;
2095
2096        my $exp = $md5->hexdigest;
2097        my $pool = SVN::Pool->new;
2098        my $atd = $self->apply_textdelta($fbat, undef, $pool);
2099        my $got = SVN::TxDelta::send_stream($fh, @$atd, $pool);
2100        die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
2101        $pool->clear;
2102
2103        close $fh or croak $!;
2104}
2105
2106sub D {
2107        my ($self, $m) = @_;
2108        my ($dir, $file) = split_path($m->{file_b});
2109        my $pbat = $self->ensure_path($dir);
2110        print "\tD\t$m->{file_b}\n" unless $::_q;
2111        $self->delete_entry($m->{file_b}, $pbat);
2112}
2113
2114sub close_edit {
2115        my ($self) = @_;
2116        my ($p,$bat) = ($self->{pool}, $self->{bat});
2117        foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
2118                $self->close_directory($bat->{$_}, $p);
2119        }
2120        $self->SUPER::close_edit($p);
2121        $p->clear;
2122}
2123
2124sub abort_edit {
2125        my ($self) = @_;
2126        $self->SUPER::abort_edit($self->{pool});
2127        $self->{pool}->clear;
2128}
2129
2130# this drives the editor
2131sub apply_diff {
2132        my ($self, $tree_a, $tree_b) = @_;
2133        my @diff_tree = qw(diff-tree -z -r);
2134        if ($::_cp_similarity) {
2135                push @diff_tree, "-C$::_cp_similarity";
2136        } else {
2137                push @diff_tree, '-C';
2138        }
2139        push @diff_tree, '--find-copies-harder' if $::_find_copies_harder;
2140        push @diff_tree, "-l$::_l" if defined $::_l;
2141        push @diff_tree, $tree_a, $tree_b;
2142        my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
2143        my $nl = $/;
2144        local $/ = "\0";
2145        my $state = 'meta';
2146        my @mods;
2147        while (<$diff_fh>) {
2148                chomp $_; # this gets rid of the trailing "\0"
2149                if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
2150                                        $::sha1\s($::sha1)\s
2151                                        ([MTCRAD])\d*$/xo) {
2152                        push @mods, {   mode_a => $1, mode_b => $2,
2153                                        sha1_b => $3, chg => $4 };
2154                        if ($4 =~ /^(?:C|R)$/) {
2155                                $state = 'file_a';
2156                        } else {
2157                                $state = 'file_b';
2158                        }
2159                } elsif ($state eq 'file_a') {
2160                        my $x = $mods[$#mods] or croak "Empty array\n";
2161                        if ($x->{chg} !~ /^(?:C|R)$/) {
2162                                croak "Error parsing $_, $x->{chg}\n";
2163                        }
2164                        $x->{file_a} = $_;
2165                        $state = 'file_b';
2166                } elsif ($state eq 'file_b') {
2167                        my $x = $mods[$#mods] or croak "Empty array\n";
2168                        if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
2169                                croak "Error parsing $_, $x->{chg}\n";
2170                        }
2171                        if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
2172                                croak "Error parsing $_, $x->{chg}\n";
2173                        }
2174                        $x->{file_b} = $_;
2175                        $state = 'meta';
2176                } else {
2177                        croak "Error parsing $_\n";
2178                }
2179        }
2180        command_close_pipe($diff_fh, $ctx);
2181        $/ = $nl;
2182
2183        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
2184        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @mods) {
2185                my $f = $m->{chg};
2186                if (defined $o{$f}) {
2187                        $self->$f($m);
2188                } else {
2189                        fatal("Invalid change type: $f\n");
2190                }
2191        }
2192        $self->rmdirs($tree_b) if $::_rmdir;
2193        if (@mods == 0) {
2194                $self->abort_edit;
2195        } else {
2196                $self->close_edit;
2197        }
2198        \@mods;
2199}
2200
2201package Git::SVN::Ra;
2202use vars qw/@ISA $config_dir/;
2203use strict;
2204use warnings;
2205my ($can_do_switch);
2206my %RA;
2207
2208BEGIN {
2209        # enforce temporary pool usage for some simple functions
2210        my $e;
2211        foreach (qw/get_latest_revnum rev_proplist get_file
2212                    check_path get_dir get_uuid get_repos_root/) {
2213                $e .= "sub $_ {
2214                        my \$self = shift;
2215                        my \$pool = SVN::Pool->new;
2216                        my \@ret = \$self->SUPER::$_(\@_,\$pool);
2217                        \$pool->clear;
2218                        wantarray ? \@ret : \$ret[0]; }\n";
2219        }
2220        eval $e;
2221}
2222
2223sub new {
2224        my ($class, $url) = @_;
2225        $url =~ s!/+$!!;
2226        return $RA{$url} if $RA{$url};
2227
2228        SVN::_Core::svn_config_ensure($config_dir, undef);
2229        my ($baton, $callbacks) = SVN::Core::auth_open_helper([
2230            SVN::Client::get_simple_provider(),
2231            SVN::Client::get_ssl_server_trust_file_provider(),
2232            SVN::Client::get_simple_prompt_provider(
2233              \&Git::SVN::Prompt::simple, 2),
2234            SVN::Client::get_ssl_client_cert_prompt_provider(
2235              \&Git::SVN::Prompt::ssl_client_cert, 2),
2236            SVN::Client::get_ssl_client_cert_pw_prompt_provider(
2237              \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
2238            SVN::Client::get_username_provider(),
2239            SVN::Client::get_ssl_server_trust_prompt_provider(
2240              \&Git::SVN::Prompt::ssl_server_trust),
2241            SVN::Client::get_username_prompt_provider(
2242              \&Git::SVN::Prompt::username, 2),
2243          ]);
2244        my $config = SVN::Core::config_get_config($config_dir);
2245        my $self = SVN::Ra->new(url => $url, auth => $baton,
2246                              config => $config,
2247                              pool => SVN::Pool->new,
2248                              auth_provider_callbacks => $callbacks);
2249        $self->{svn_path} = $url;
2250        $self->{repos_root} = $self->get_repos_root;
2251        $self->{svn_path} =~ s#^\Q$self->{repos_root}\E/*##;
2252        $RA{$url} = bless $self, $class;
2253}
2254
2255sub DESTROY {
2256        # do not call the real DESTROY since we store ourselves in %RA
2257}
2258
2259sub dup {
2260        my ($self) = @_;
2261        my $dup = SVN::Ra->new(pool => SVN::Pool->new,
2262                                map { $_ => $self->{$_} } qw/config url
2263                     auth auth_provider_callbacks repos_root svn_path/);
2264        bless $dup, ref $self;
2265}
2266
2267sub get_log {
2268        my ($self, @args) = @_;
2269        my $pool = SVN::Pool->new;
2270        splice(@args, 3, 1) if ($SVN::Core::VERSION le '1.2.0');
2271        my $ret = $self->SUPER::get_log(@args, $pool);
2272        $pool->clear;
2273        $ret;
2274}
2275
2276sub get_commit_editor {
2277        my ($self, $log, $cb, $pool) = @_;
2278        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
2279        $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
2280}
2281
2282sub uuid {
2283        my ($self) = @_;
2284        $self->{uuid} ||= $self->get_uuid;
2285}
2286
2287sub gs_do_update {
2288        my ($self, $rev_a, $rev_b, $path, $recurse, $editor) = @_;
2289        my $pool = SVN::Pool->new;
2290        $editor->set_path_strip($path);
2291        my $reporter = $self->do_update($rev_b, $path, $recurse,
2292                                        $editor, $pool);
2293        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
2294        my $new = ($rev_a == $rev_b);
2295        $reporter->set_path('', $rev_a, $new, @lock, $pool);
2296        $reporter->finish_report($pool);
2297        $pool->clear;
2298        $editor->{git_commit_ok};
2299}
2300
2301sub gs_do_switch {
2302        my ($self, $rev_a, $rev_b, $path, $recurse, $url_b, $editor) = @_;
2303        my $pool = SVN::Pool->new;
2304        $editor->set_path_strip($path);
2305        my $reporter = $self->do_switch($rev_b, $path, $recurse,
2306                                        $url_b, $editor, $pool);
2307        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
2308        $reporter->set_path('', $rev_a, 0, @lock, $pool);
2309        $reporter->finish_report($pool);
2310        $pool->clear;
2311        $editor->{git_commit_ok};
2312}
2313
2314sub minimize_url {
2315        my ($self) = @_;
2316        return $self->{url} if ($self->{url} eq $self->{repos_root});
2317        my $url = $self->{repos_root};
2318        my @components = split(m!/!, $self->{svn_path});
2319        my $c = '';
2320        do {
2321                $url .= "/$c" if length $c;
2322                eval { (ref $self)->new($url)->get_latest_revnum };
2323        } while ($@ && ($c = shift @components));
2324        $url;
2325}
2326
2327sub can_do_switch {
2328        my $self = shift;
2329        unless (defined $can_do_switch) {
2330                my $pool = SVN::Pool->new;
2331                my $rep = eval {
2332                        $self->do_switch(1, '', 0, $self->{url},
2333                                         SVN::Delta::Editor->new, $pool);
2334                };
2335                if ($@) {
2336                        $can_do_switch = 0;
2337                } else {
2338                        $rep->abort_report($pool);
2339                        $can_do_switch = 1;
2340                }
2341                $pool->clear;
2342        }
2343        $can_do_switch;
2344}
2345
2346package Git::SVN::Log;
2347use strict;
2348use warnings;
2349use POSIX qw/strftime/;
2350use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
2351            %rusers $show_commit $incremental/;
2352my $l_fmt;
2353
2354sub cmt_showable {
2355        my ($c) = @_;
2356        return 1 if defined $c->{r};
2357        if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
2358                                $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
2359                my @log = command(qw/cat-file commit/, $c->{c});
2360                shift @log while ($log[0] ne "\n");
2361                shift @log;
2362                @{$c->{l}} = grep !/^git-svn-id: /, @log;
2363
2364                (undef, $c->{r}, undef) = ::extract_metadata(
2365                                (grep(/^git-svn-id: /, @log))[-1]);
2366        }
2367        return defined $c->{r};
2368}
2369
2370sub log_use_color {
2371        return 1 if $color;
2372        my ($dc, $dcvar);
2373        $dcvar = 'color.diff';
2374        $dc = `git-config --get $dcvar`;
2375        if ($dc eq '') {
2376                # nothing at all; fallback to "diff.color"
2377                $dcvar = 'diff.color';
2378                $dc = `git-config --get $dcvar`;
2379        }
2380        chomp($dc);
2381        if ($dc eq 'auto') {
2382                my $pc;
2383                $pc = `git-config --get color.pager`;
2384                if ($pc eq '') {
2385                        # does not have it -- fallback to pager.color
2386                        $pc = `git-config --bool --get pager.color`;
2387                }
2388                else {
2389                        $pc = `git-config --bool --get color.pager`;
2390                        if ($?) {
2391                                $pc = 'false';
2392                        }
2393                }
2394                chomp($pc);
2395                if (-t *STDOUT || (defined $pager && $pc eq 'true')) {
2396                        return ($ENV{TERM} && $ENV{TERM} ne 'dumb');
2397                }
2398                return 0;
2399        }
2400        return 0 if $dc eq 'never';
2401        return 1 if $dc eq 'always';
2402        chomp($dc = `git-config --bool --get $dcvar`);
2403        return ($dc eq 'true');
2404}
2405
2406sub git_svn_log_cmd {
2407        my ($r_min, $r_max) = @_;
2408        my $gs = Git::SVN->_new;
2409        my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
2410                   $gs->refname);
2411        push @cmd, '-r' unless $non_recursive;
2412        push @cmd, qw/--raw --name-status/ if $verbose;
2413        push @cmd, '--color' if log_use_color();
2414        return @cmd unless defined $r_max;
2415        if ($r_max == $r_min) {
2416                push @cmd, '--max-count=1';
2417                if (my $c = $gs->rev_db_get($r_max)) {
2418                        push @cmd, $c;
2419                }
2420        } else {
2421                my ($c_min, $c_max);
2422                $c_max = $gs->rev_db_get($r_max);
2423                $c_min = $gs->rev_db_get($r_min);
2424                if (defined $c_min && defined $c_max) {
2425                        if ($r_max > $r_max) {
2426                                push @cmd, "$c_min..$c_max";
2427                        } else {
2428                                push @cmd, "$c_max..$c_min";
2429                        }
2430                } elsif ($r_max > $r_min) {
2431                        push @cmd, $c_max;
2432                } else {
2433                        push @cmd, $c_min;
2434                }
2435        }
2436        return @cmd;
2437}
2438
2439# adapted from pager.c
2440sub config_pager {
2441        $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
2442        if (!defined $pager) {
2443                $pager = 'less';
2444        } elsif (length $pager == 0 || $pager eq 'cat') {
2445                $pager = undef;
2446        }
2447}
2448
2449sub run_pager {
2450        return unless -t *STDOUT;
2451        pipe my $rfd, my $wfd or return;
2452        defined(my $pid = fork) or ::fatal "Can't fork: $!\n";
2453        if (!$pid) {
2454                open STDOUT, '>&', $wfd or
2455                                     ::fatal "Can't redirect to stdout: $!\n";
2456                return;
2457        }
2458        open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!\n";
2459        $ENV{LESS} ||= 'FRSX';
2460        exec $pager or ::fatal "Can't run pager: $! ($pager)\n";
2461}
2462
2463sub get_author_info {
2464        my ($dest, $author, $t, $tz) = @_;
2465        $author =~ s/(?:^\s*|\s*$)//g;
2466        $dest->{a_raw} = $author;
2467        my $au;
2468        if ($::_authors) {
2469                $au = $rusers{$author} || undef;
2470        }
2471        if (!$au) {
2472                ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
2473        }
2474        $dest->{t} = $t;
2475        $dest->{tz} = $tz;
2476        $dest->{a} = $au;
2477        # Date::Parse isn't in the standard Perl distro :(
2478        if ($tz =~ s/^\+//) {
2479                $t += ::tz_to_s_offset($tz);
2480        } elsif ($tz =~ s/^\-//) {
2481                $t -= ::tz_to_s_offset($tz);
2482        }
2483        $dest->{t_utc} = $t;
2484}
2485
2486sub process_commit {
2487        my ($c, $r_min, $r_max, $defer) = @_;
2488        if (defined $r_min && defined $r_max) {
2489                if ($r_min == $c->{r} && $r_min == $r_max) {
2490                        show_commit($c);
2491                        return 0;
2492                }
2493                return 1 if $r_min == $r_max;
2494                if ($r_min < $r_max) {
2495                        # we need to reverse the print order
2496                        return 0 if (defined $limit && --$limit < 0);
2497                        push @$defer, $c;
2498                        return 1;
2499                }
2500                if ($r_min != $r_max) {
2501                        return 1 if ($r_min < $c->{r});
2502                        return 1 if ($r_max > $c->{r});
2503                }
2504        }
2505        return 0 if (defined $limit && --$limit < 0);
2506        show_commit($c);
2507        return 1;
2508}
2509
2510sub show_commit {
2511        my $c = shift;
2512        if ($oneline) {
2513                my $x = "\n";
2514                if (my $l = $c->{l}) {
2515                        while ($l->[0] =~ /^\s*$/) { shift @$l }
2516                        $x = $l->[0];
2517                }
2518                $l_fmt ||= 'A' . length($c->{r});
2519                print 'r',pack($l_fmt, $c->{r}),' | ';
2520                print "$c->{c} | " if $show_commit;
2521                print $x;
2522        } else {
2523                show_commit_normal($c);
2524        }
2525}
2526
2527sub show_commit_changed_paths {
2528        my ($c) = @_;
2529        return unless $c->{changed};
2530        print "Changed paths:\n", @{$c->{changed}};
2531}
2532
2533sub show_commit_normal {
2534        my ($c) = @_;
2535        print '-' x72, "\nr$c->{r} | ";
2536        print "$c->{c} | " if $show_commit;
2537        print "$c->{a} | ", strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)",
2538                                 localtime($c->{t_utc})), ' | ';
2539        my $nr_line = 0;
2540
2541        if (my $l = $c->{l}) {
2542                while ($l->[$#$l] eq "\n" && $#$l > 0
2543                                          && $l->[($#$l - 1)] eq "\n") {
2544                        pop @$l;
2545                }
2546                $nr_line = scalar @$l;
2547                if (!$nr_line) {
2548                        print "1 line\n\n\n";
2549                } else {
2550                        if ($nr_line == 1) {
2551                                $nr_line = '1 line';
2552                        } else {
2553                                $nr_line .= ' lines';
2554                        }
2555                        print $nr_line, "\n";
2556                        show_commit_changed_paths($c);
2557                        print "\n";
2558                        print $_ foreach @$l;
2559                }
2560        } else {
2561                print "1 line\n";
2562                show_commit_changed_paths($c);
2563                print "\n";
2564
2565        }
2566        foreach my $x (qw/raw diff/) {
2567                if ($c->{$x}) {
2568                        print "\n";
2569                        print $_ foreach @{$c->{$x}}
2570                }
2571        }
2572}
2573
2574sub cmd_show_log {
2575        my (@args) = @_;
2576        my ($r_min, $r_max);
2577        my $r_last = -1; # prevent dupes
2578        if (defined $TZ) {
2579                $ENV{TZ} = $TZ;
2580        } else {
2581                delete $ENV{TZ};
2582        }
2583        if (defined $::_revision) {
2584                if ($::_revision =~ /^(\d+):(\d+)$/) {
2585                        ($r_min, $r_max) = ($1, $2);
2586                } elsif ($::_revision =~ /^\d+$/) {
2587                        $r_min = $r_max = $::_revision;
2588                } else {
2589                        ::fatal "-r$::_revision is not supported, use ",
2590                                "standard \'git log\' arguments instead\n";
2591                }
2592        }
2593
2594        config_pager();
2595        @args = (git_svn_log_cmd($r_min, $r_max), @args);
2596        my $log = command_output_pipe(@args);
2597        run_pager();
2598        my (@k, $c, $d);
2599        my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
2600        while (<$log>) {
2601                if (/^${esc_color}commit ($::sha1_short)/o) {
2602                        my $cmt = $1;
2603                        if ($c && cmt_showable($c) && $c->{r} != $r_last) {
2604                                $r_last = $c->{r};
2605                                process_commit($c, $r_min, $r_max, \@k) or
2606                                                                goto out;
2607                        }
2608                        $d = undef;
2609                        $c = { c => $cmt };
2610                } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
2611                        get_author_info($c, $1, $2, $3);
2612                } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
2613                        # ignore
2614                } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
2615                        push @{$c->{raw}}, $_;
2616                } elsif (/^${esc_color}[ACRMDT]\t/) {
2617                        # we could add $SVN->{svn_path} here, but that requires
2618                        # remote access at the moment (repo_path_split)...
2619                        s#^(${esc_color})([ACRMDT])\t#$1   $2 #o;
2620                        push @{$c->{changed}}, $_;
2621                } elsif (/^${esc_color}diff /o) {
2622                        $d = 1;
2623                        push @{$c->{diff}}, $_;
2624                } elsif ($d) {
2625                        push @{$c->{diff}}, $_;
2626                } elsif (/^${esc_color}    (git-svn-id:.+)$/o) {
2627                        ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
2628                } elsif (s/^${esc_color}    //o) {
2629                        push @{$c->{l}}, $_;
2630                }
2631        }
2632        if ($c && defined $c->{r} && $c->{r} != $r_last) {
2633                $r_last = $c->{r};
2634                process_commit($c, $r_min, $r_max, \@k);
2635        }
2636        if (@k) {
2637                my $swap = $r_max;
2638                $r_max = $r_min;
2639                $r_min = $swap;
2640                process_commit($_, $r_min, $r_max) foreach reverse @k;
2641        }
2642out:
2643        close $log;
2644        print '-' x72,"\n" unless $incremental || $oneline;
2645}
2646
2647package Git::SVN::Migration;
2648# these version numbers do NOT correspond to actual version numbers
2649# of git nor git-svn.  They are just relative.
2650#
2651# v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
2652#
2653# v1 layout: .git/$id/info/url, refs/remotes/$id
2654#
2655# v2 layout: .git/svn/$id/info/url, refs/remotes/$id
2656#
2657# v3 layout: .git/svn/$id, refs/remotes/$id
2658#            - info/url may remain for backwards compatibility
2659#            - this is what we migrate up to this layout automatically,
2660#            - this will be used by git svn init on single branches
2661#
2662# v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
2663#            - this is only created for newly multi-init-ed
2664#              repositories.  Similar in spirit to the
2665#              --use-separate-remotes option in git-clone (now default)
2666#            - we do not automatically migrate to this (following
2667#              the example set by core git)
2668use strict;
2669use warnings;
2670use Carp qw/croak/;
2671use File::Path qw/mkpath/;
2672use File::Basename qw/dirname basename/;
2673use vars qw/$_minimize/;
2674
2675sub migrate_from_v0 {
2676        my $git_dir = $ENV{GIT_DIR};
2677        return undef unless -d $git_dir;
2678        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
2679        my $migrated = 0;
2680        while (<$fh>) {
2681                chomp;
2682                my ($id, $orig_ref) = ($_, $_);
2683                next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
2684                next unless -f "$git_dir/$id/info/url";
2685                my $new_ref = "refs/remotes/$id";
2686                if (::verify_ref("$new_ref^0")) {
2687                        print STDERR "W: $orig_ref is probably an old ",
2688                                     "branch used by an ancient version of ",
2689                                     "git-svn.\n",
2690                                     "However, $new_ref also exists.\n",
2691                                     "We will not be able ",
2692                                     "to use this branch until this ",
2693                                     "ambiguity is resolved.\n";
2694                        next;
2695                }
2696                print STDERR "Migrating from v0 layout...\n" if !$migrated;
2697                print STDERR "Renaming ref: $orig_ref => $new_ref\n";
2698                command_noisy('update-ref', $new_ref, $orig_ref);
2699                command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
2700                $migrated++;
2701        }
2702        command_close_pipe($fh, $ctx);
2703        print STDERR "Done migrating from v0 layout...\n" if $migrated;
2704        $migrated;
2705}
2706
2707sub migrate_from_v1 {
2708        my $git_dir = $ENV{GIT_DIR};
2709        my $migrated = 0;
2710        return $migrated unless -d $git_dir;
2711        my $svn_dir = "$git_dir/svn";
2712
2713        # just in case somebody used 'svn' as their $id at some point...
2714        return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
2715
2716        print STDERR "Migrating from a git-svn v1 layout...\n";
2717        mkpath([$svn_dir]);
2718        print STDERR "Data from a previous version of git-svn exists, but\n\t",
2719                     "$svn_dir\n\t(required for this version ",
2720                     "($::VERSION) of git-svn) does not. exist\n";
2721        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
2722        while (<$fh>) {
2723                my $x = $_;
2724                next unless $x =~ s#^refs/remotes/##;
2725                chomp $x;
2726                next unless -f "$git_dir/$x/info/url";
2727                my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
2728                next unless $u;
2729                my $dn = dirname("$git_dir/svn/$x");
2730                mkpath([$dn]) unless -d $dn;
2731                if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
2732                        mkpath(["$git_dir/svn/svn"]);
2733                        print STDERR " - $git_dir/$x/info => ",
2734                                        "$git_dir/svn/$x/info\n";
2735                        rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
2736                               croak "$!: $x";
2737                        # don't worry too much about these, they probably
2738                        # don't exist with repos this old (save for index,
2739                        # and we can easily regenerate that)
2740                        foreach my $f (qw/unhandled.log index .rev_db/) {
2741                                rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
2742                        }
2743                } else {
2744                        print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
2745                        rename "$git_dir/$x", "$git_dir/svn/$x" or
2746                               croak "$!: $x";
2747                }
2748                $migrated++;
2749        }
2750        command_close_pipe($fh, $ctx);
2751        print STDERR "Done migrating from a git-svn v1 layout\n";
2752        $migrated;
2753}
2754
2755sub read_old_urls {
2756        my ($l_map, $pfx, $path) = @_;
2757        my @dir;
2758        foreach (<$path/*>) {
2759                if (-r "$_/info/url") {
2760                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
2761                        my $ref_id = $pfx . basename $_;
2762                        my $url = ::file_to_s("$_/info/url");
2763                        $l_map->{$ref_id} = $url;
2764                } elsif (-d $_) {
2765                        push @dir, $_;
2766                }
2767        }
2768        foreach (@dir) {
2769                my $x = $_;
2770                $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
2771                read_old_urls($l_map, $x, $_);
2772        }
2773}
2774
2775sub migrate_from_v2 {
2776        my @cfg = command(qw/config -l/);
2777        return if grep /^svn-remote\..+\.url=/, @cfg;
2778        my %l_map;
2779        read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
2780        my $migrated = 0;
2781
2782        foreach my $ref_id (sort keys %l_map) {
2783                Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
2784                $migrated++;
2785        }
2786        $migrated;
2787}
2788
2789sub minimize_connections {
2790        my $r = Git::SVN::read_all_remotes();
2791        my $new_urls = {};
2792        my $root_repos = {};
2793        foreach my $repo_id (keys %$r) {
2794                my $url = $r->{$repo_id}->{url} or next;
2795                my $fetch = $r->{$repo_id}->{fetch} or next;
2796                my $ra = Git::SVN::Ra->new($url);
2797
2798                # skip existing cases where we already connect to the root
2799                if (($ra->{url} eq $ra->{repos_root}) ||
2800                    (Git::SVN::sanitize_remote_name($ra->{repos_root}) eq
2801                     $repo_id)) {
2802                        $root_repos->{$ra->{url}} = $repo_id;
2803                        next;
2804                }
2805
2806                my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
2807                my $root_path = $ra->{url};
2808                $root_path =~ s#^\Q$ra->{repos_root}\E/*##;
2809                foreach my $path (keys %$fetch) {
2810                        my $ref_id = $fetch->{$path};
2811                        my $gs = Git::SVN->new($ref_id, $repo_id, $path);
2812
2813                        # make sure we can read when connecting to
2814                        # a higher level of a repository
2815                        my ($last_rev, undef) = $gs->last_rev_commit;
2816                        if (!defined $last_rev) {
2817                                $last_rev = eval {
2818                                        $root_ra->get_latest_revnum;
2819                                };
2820                                next if $@;
2821                        }
2822                        my $new = $root_path;
2823                        $new .= length $path ? "/$path" : '';
2824                        eval {
2825                                $root_ra->get_log([$new], $last_rev, $last_rev,
2826                                                  0, 0, 1, sub { });
2827                        };
2828                        next if $@;
2829                        $new_urls->{$ra->{repos_root}}->{$new} =
2830                                { ref_id => $ref_id,
2831                                  old_repo_id => $repo_id,
2832                                  old_path => $path };
2833                }
2834        }
2835
2836        my @emptied;
2837        foreach my $url (keys %$new_urls) {
2838                # see if we can re-use an existing [svn-remote "repo_id"]
2839                # instead of creating a(n ugly) new section:
2840                my $repo_id = $root_repos->{$url} ||
2841                              Git::SVN::sanitize_remote_name($url);
2842
2843                my $fetch = $new_urls->{$url};
2844                foreach my $path (keys %$fetch) {
2845                        my $x = $fetch->{$path};
2846                        Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
2847                        my $pfx = "svn-remote.$x->{old_repo_id}";
2848
2849                        my $old_fetch = quotemeta("$x->{old_path}:".
2850                                                  "refs/remotes/$x->{ref_id}");
2851                        command_noisy(qw/config --unset/,
2852                                      "$pfx.fetch", '^'. $old_fetch . '$');
2853                        delete $r->{$x->{old_repo_id}}->
2854                               {fetch}->{$x->{old_path}};
2855                        if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
2856                                command_noisy(qw/config --unset/,
2857                                              "$pfx.url");
2858                                push @emptied, $x->{old_repo_id}
2859                        }
2860                }
2861        }
2862        if (@emptied) {
2863                my $file = $ENV{GIT_CONFIG} || $ENV{GIT_CONFIG_LOCAL} ||
2864                           "$ENV{GIT_DIR}/config";
2865                print STDERR <<EOF;
2866The following [svn-remote] sections in your config file ($file) are empty
2867and can be safely removed:
2868EOF
2869                print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
2870        }
2871}
2872
2873sub migration_check {
2874        migrate_from_v0();
2875        migrate_from_v1();
2876        migrate_from_v2();
2877        minimize_connections() if $_minimize;
2878}
2879
2880package Git::IndexInfo;
2881use strict;
2882use warnings;
2883use Git qw/command_input_pipe command_close_pipe/;
2884
2885sub new {
2886        my ($class) = @_;
2887        my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
2888        bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
2889}
2890
2891sub remove {
2892        my ($self, $path) = @_;
2893        if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
2894                return ++$self->{nr};
2895        }
2896        undef;
2897}
2898
2899sub update {
2900        my ($self, $mode, $hash, $path) = @_;
2901        if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
2902                return ++$self->{nr};
2903        }
2904        undef;
2905}
2906
2907sub DESTROY {
2908        my ($self) = @_;
2909        command_close_pipe($self->{gui}, $self->{ctx});
2910}
2911
2912__END__
2913
2914Data structures:
2915
2916$log_entry hashref as returned by libsvn_log_entry()
2917{
2918        log => 'whitespace-formatted log entry
2919',                                              # trailing newline is preserved
2920        revision => '8',                        # integer
2921        date => '2004-02-24T17:01:44.108345Z',  # commit date
2922        author => 'committer name'
2923};
2924
2925@mods = array of diff-index line hashes, each element represents one line
2926        of diff-index output
2927
2928diff-index line ($m hash)
2929{
2930        mode_a => first column of diff-index output, no leading ':',
2931        mode_b => second column of diff-index output,
2932        sha1_b => sha1sum of the final blob,
2933        chg => change type [MCRADT],
2934        file_a => original file name of a file (iff chg is 'C' or 'R')
2935        file_b => new/current file name of a file (any chg)
2936}
2937;
2938
2939# retval of read_url_paths{,_all}();
2940$l_map = {
2941        # repository root url
2942        'https://svn.musicpd.org' => {
2943                # repository path               # GIT_SVN_ID
2944                'mpd/trunk'             =>      'trunk',
2945                'mpd/tags/0.11.5'       =>      'tags/0.11.5',
2946        },
2947}
2948
2949Notes:
2950        I don't trust the each() function on unless I created %hash myself
2951        because the internal iterator may not have started at base.