02786f1a6acc0c8b6cc8c276df9573e35a4db088
   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 $REVDB/;
  10$AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
  11$VERSION = '@@GIT_VERSION@@';
  12
  13use Cwd qw/abs_path/;
  14$GIT_DIR = abs_path($ENV{GIT_DIR} || '.git');
  15$ENV{GIT_DIR} = $GIT_DIR;
  16
  17my $LC_ALL = $ENV{LC_ALL};
  18my $TZ = $ENV{TZ};
  19# make sure the svn binary gives consistent output between locales and TZs:
  20$ENV{TZ} = 'UTC';
  21$ENV{LC_ALL} = 'C';
  22$| = 1; # unbuffer STDOUT
  23
  24# properties that we do not log:
  25my %SKIP = ( 'svn:wc:ra_dav:version-url' => 1,
  26             'svn:special' => 1,
  27             'svn:executable' => 1,
  28             'svn:entry:committed-rev' => 1,
  29             'svn:entry:last-author' => 1,
  30             'svn:entry:uuid' => 1,
  31             'svn:entry:committed-date' => 1,
  32);
  33
  34sub fatal (@) { print STDERR @_; exit 1 }
  35require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
  36require SVN::Ra;
  37require SVN::Delta;
  38if ($SVN::Core::VERSION lt '1.1.0') {
  39        fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)\n";
  40}
  41push @Git::SVN::Ra::ISA, 'SVN::Ra';
  42push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
  43push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
  44use Carp qw/croak/;
  45use IO::File qw//;
  46use File::Basename qw/dirname basename/;
  47use File::Path qw/mkpath/;
  48use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev pass_through/;
  49use POSIX qw/strftime/;
  50use IPC::Open3;
  51use Memoize;
  52use Git;
  53memoize('revisions_eq');
  54memoize('cmt_metadata');
  55memoize('get_commit_time');
  56
  57BEGIN {
  58        my $s;
  59        foreach (qw/command command_oneline command_noisy command_output_pipe
  60                    command_input_pipe command_close_pipe/) {
  61                $s .= "*SVN::Git::Editor::$_ = *SVN::Git::Fetcher::$_ = ".
  62                      "*$_ = *Git::$_; ";
  63        }
  64        eval $s;
  65}
  66
  67my ($SVN);
  68
  69my $_optimize_commits = 1 unless $ENV{GIT_SVN_NO_OPTIMIZE_COMMITS};
  70my $sha1 = qr/[a-f\d]{40}/;
  71my $sha1_short = qr/[a-f\d]{4,40}/;
  72my $_esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
  73my ($_revision,$_stdin,$_help,$_rmdir,$_edit,
  74        $_find_copies_harder, $_l, $_cp_similarity, $_cp_remote,
  75        $_repack, $_repack_nr, $_repack_flags, $_q,
  76        $_message, $_file, $_no_metadata,
  77        $_template, $_shared, $_no_default_regex, $_no_graft_copy,
  78        $_limit, $_verbose, $_incremental, $_oneline, $_l_fmt, $_show_commit,
  79        $_version, $_upgrade, $_authors, $_branch_all_refs, @_opt_m,
  80        $_merge, $_strategy, $_dry_run, $_ignore_nodate, $_non_recursive,
  81        $_pager, $_color, $_prefix);
  82my (@_branch_from, %tree_map, %users, %rusers);
  83my @repo_path_split_cache;
  84use vars qw/$_follow_parent/;
  85
  86my %fc_opts = ( 'branch|b=s' => \@_branch_from,
  87                'follow-parent|follow' => \$_follow_parent,
  88                'branch-all-refs|B' => \$_branch_all_refs,
  89                'authors-file|A=s' => \$_authors,
  90                'repack:i' => \$_repack,
  91                'no-metadata' => \$_no_metadata,
  92                'quiet|q' => \$_q,
  93                'username=s' => \$Git::SVN::Prompt::_username,
  94                'config-dir=s' => \$Git::SVN::Ra::config_dir,
  95                'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache,
  96                'ignore-nodate' => \$_ignore_nodate,
  97                'repack-flags|repack-args|repack-opts=s' => \$_repack_flags);
  98
  99my ($_trunk, $_tags, $_branches);
 100my %multi_opts = ( 'trunk|T=s' => \$_trunk,
 101                'tags|t=s' => \$_tags,
 102                'branches|b=s' => \$_branches );
 103my %init_opts = ( 'template=s' => \$_template, 'shared' => \$_shared );
 104my %cmt_opts = ( 'edit|e' => \$_edit,
 105                'rmdir' => \$_rmdir,
 106                'find-copies-harder' => \$_find_copies_harder,
 107                'l=i' => \$_l,
 108                'copy-similarity|C=i'=> \$_cp_similarity
 109);
 110
 111my %cmd = (
 112        fetch => [ \&cmd_fetch, "Download new revisions from SVN",
 113                        { 'revision|r=s' => \$_revision, %fc_opts } ],
 114        init => [ \&cmd_init, "Initialize a repo for tracking" .
 115                          " (requires URL argument)",
 116                          \%init_opts ],
 117        dcommit => [ \&dcommit, 'Commit several diffs to merge with upstream',
 118                        { 'merge|m|M' => \$_merge,
 119                          'strategy|s=s' => \$_strategy,
 120                          'dry-run|n' => \$_dry_run,
 121                        %cmt_opts, %fc_opts } ],
 122        'set-tree' => [ \&commit, "Set an SVN repository to a git tree-ish",
 123                        {       'stdin|' => \$_stdin, %cmt_opts, %fc_opts, } ],
 124        'show-ignore' => [ \&show_ignore, "Show svn:ignore listings",
 125                        { 'revision|r=i' => \$_revision } ],
 126        rebuild => [ \&rebuild, "Rebuild git-svn metadata (after git clone)",
 127                        { 'copy-remote|remote=s' => \$_cp_remote,
 128                          'upgrade' => \$_upgrade } ],
 129        'graft-branches' => [ \&graft_branches,
 130                        'Detect merges/branches from already imported history',
 131                        { 'merge-rx|m' => \@_opt_m,
 132                          'branch|b=s' => \@_branch_from,
 133                          'branch-all-refs|B' => \$_branch_all_refs,
 134                          'no-default-regex' => \$_no_default_regex,
 135                          'no-graft-copy' => \$_no_graft_copy } ],
 136        'multi-init' => [ \&cmd_multi_init,
 137                        'Initialize multiple trees (like git-svnimport)',
 138                        { %multi_opts, %init_opts,
 139                         'revision|r=i' => \$_revision,
 140                         'username=s' => \$Git::SVN::Prompt::_username,
 141                         'config-dir=s' => \$Git::SVN::Ra::config_dir,
 142                         'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache,
 143                         'prefix=s' => \$_prefix,
 144                        } ],
 145        'multi-fetch' => [ \&multi_fetch,
 146                        'Fetch multiple trees (like git-svnimport)',
 147                        \%fc_opts ],
 148        'log' => [ \&show_log, 'Show commit logs',
 149                        { 'limit=i' => \$_limit,
 150                          'revision|r=s' => \$_revision,
 151                          'verbose|v' => \$_verbose,
 152                          'incremental' => \$_incremental,
 153                          'oneline' => \$_oneline,
 154                          'show-commit' => \$_show_commit,
 155                          'non-recursive' => \$_non_recursive,
 156                          'authors-file|A=s' => \$_authors,
 157                          'color' => \$_color,
 158                          'pager=s' => \$_pager,
 159                        } ],
 160        'commit-diff' => [ \&commit_diff, 'Commit a diff between two trees',
 161                        { 'message|m=s' => \$_message,
 162                          'file|F=s' => \$_file,
 163                          'revision|r=s' => \$_revision,
 164                        %cmt_opts } ],
 165);
 166
 167my $cmd;
 168for (my $i = 0; $i < @ARGV; $i++) {
 169        if (defined $cmd{$ARGV[$i]}) {
 170                $cmd = $ARGV[$i];
 171                splice @ARGV, $i, 1;
 172                last;
 173        }
 174};
 175
 176my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
 177
 178read_repo_config(\%opts);
 179my $rv = GetOptions(%opts, 'help|H|h' => \$_help,
 180                                'version|V' => \$_version,
 181                                'id|i=s' => \$GIT_SVN);
 182exit 1 if (!$rv && $cmd ne 'log');
 183
 184set_default_vals();
 185usage(0) if $_help;
 186version() if $_version;
 187usage(1) unless defined $cmd;
 188init_vars();
 189load_authors() if $_authors;
 190load_all_refs() if $_branch_all_refs;
 191migration_check() unless $cmd =~ /^(?:init|rebuild|multi-init|commit-diff)$/;
 192$cmd{$cmd}->[0]->(@ARGV);
 193exit 0;
 194
 195####################### primary functions ######################
 196sub usage {
 197        my $exit = shift || 0;
 198        my $fd = $exit ? \*STDERR : \*STDOUT;
 199        print $fd <<"";
 200git-svn - bidirectional operations between a single Subversion tree and git
 201Usage: $0 <command> [options] [arguments]\n
 202
 203        print $fd "Available commands:\n" unless $cmd;
 204
 205        foreach (sort keys %cmd) {
 206                next if $cmd && $cmd ne $_;
 207                print $fd '  ',pack('A17',$_),$cmd{$_}->[1],"\n";
 208                foreach (keys %{$cmd{$_}->[2]}) {
 209                        # prints out arguments as they should be passed:
 210                        my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
 211                        print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
 212                                                        "--$_" : "-$_" }
 213                                                split /\|/,$_)," $x\n";
 214                }
 215        }
 216        print $fd <<"";
 217\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
 218arbitrary identifier if you're tracking multiple SVN branches/repositories in
 219one git repository and want to keep them separate.  See git-svn(1) for more
 220information.
 221
 222        exit $exit;
 223}
 224
 225sub version {
 226        print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
 227        exit 0;
 228}
 229
 230sub rebuild {
 231        if (!verify_ref("refs/remotes/$GIT_SVN^0")) {
 232                copy_remote_ref();
 233        }
 234        $SVN_URL = shift or undef;
 235        my $newest_rev = 0;
 236        if ($_upgrade) {
 237                command_noisy('update-ref',"refs/remotes/$GIT_SVN","
 238                              $GIT_SVN-HEAD");
 239        } else {
 240                check_upgrade_needed();
 241        }
 242
 243        my ($rev_list, $ctx) = command_output_pipe("rev-list",
 244                                                   "refs/remotes/$GIT_SVN");
 245        my $latest;
 246        my $svn_uuid;
 247        while (<$rev_list>) {
 248                chomp;
 249                my $c = $_;
 250                croak "Non-SHA1: $c\n" unless $c =~ /^$sha1$/o;
 251                my @commit = grep(/^git-svn-id: /,
 252                                  command(qw/cat-file commit/, $c));
 253                next if (!@commit); # skip merges
 254                my ($url, $rev, $uuid) = extract_metadata($commit[$#commit]);
 255                if (!defined $rev || !$uuid) {
 256                        croak "Unable to extract revision or UUID from ",
 257                                "$c, $commit[$#commit]\n";
 258                }
 259
 260                # if we merged or otherwise started elsewhere, this is
 261                # how we break out of it
 262                next if (defined $svn_uuid && ($uuid ne $svn_uuid));
 263                next if (defined $SVN_URL && defined $url && ($url ne $SVN_URL));
 264
 265                unless (defined $latest) {
 266                        if (!$SVN_URL && !$url) {
 267                                croak "SVN repository location required: $url\n";
 268                        }
 269                        $SVN_URL ||= $url;
 270                        $svn_uuid ||= $uuid;
 271                        setup_git_svn();
 272                        $latest = $rev;
 273                }
 274                revdb_set($REVDB, $rev, $c);
 275                print "r$rev = $c\n";
 276                $newest_rev = $rev if ($rev > $newest_rev);
 277        }
 278        command_close_pipe($rev_list, $ctx);
 279}
 280
 281sub do_git_init_db {
 282        unless (-d $ENV{GIT_DIR}) {
 283                my @init_db = ('init');
 284                push @init_db, "--template=$_template" if defined $_template;
 285                push @init_db, "--shared" if defined $_shared;
 286                command_noisy(@init_db);
 287        }
 288}
 289
 290sub cmd_init {
 291        my $url = shift or die "SVN repository location required " .
 292                                "as a command-line argument\n";
 293        if (my $repo_path = shift) {
 294                unless (-d $repo_path) {
 295                        mkpath([$repo_path]);
 296                }
 297                chdir $repo_path or croak $!;
 298                $ENV{GIT_DIR} = $repo_path . "/.git";
 299        }
 300        do_git_init_db();
 301
 302        Git::SVN->init(undef, $url);
 303}
 304
 305sub cmd_fetch {
 306        fetch_child_id($GIT_SVN, @_);
 307}
 308
 309sub fetch {
 310        check_upgrade_needed();
 311        $SVN_URL ||= file_to_s("$GIT_SVN_DIR/info/url");
 312        my $ret = fetch_lib(@_);
 313        if ($ret->{commit} && !verify_ref('refs/heads/master^0')) {
 314                command_noisy(qw(update-ref refs/heads/master),$ret->{commit});
 315        }
 316        return $ret;
 317}
 318
 319sub fetch_lib {
 320        my (@parents) = @_;
 321        $SVN_URL ||= file_to_s("$GIT_SVN_DIR/info/url");
 322        $SVN ||= Git::SVN::Ra->new($SVN_URL);
 323        my ($last_rev, $last_commit) = svn_grab_base_rev();
 324        my ($base, $head) = libsvn_parse_revision($last_rev);
 325        if ($base > $head) {
 326                return { revision => $last_rev, commit => $last_commit }
 327        }
 328        my $index = set_index($GIT_SVN_INDEX);
 329
 330        # limit ourselves and also fork() since get_log won't release memory
 331        # after processing a revision and SVN stuff seems to leak
 332        my $inc = 1000;
 333        my ($min, $max) = ($base, $head < $base+$inc ? $head : $base+$inc);
 334        if (defined $last_commit) {
 335                unless (-e $GIT_SVN_INDEX) {
 336                        command_noisy('read-tree', $last_commit);
 337                }
 338                my $x = command_oneline('write-tree');
 339                my ($y) = (command(qw/cat-file commit/, $last_commit)
 340                                                        =~ /^tree ($sha1)/m);
 341                if ($y ne $x) {
 342                        unlink $GIT_SVN_INDEX or croak $!;
 343                        command_noisy('read-tree', $last_commit);
 344                }
 345                $x = command_oneline('write-tree');
 346                if ($y ne $x) {
 347                        print STDERR "trees ($last_commit) $y != $x\n",
 348                                 "Something is seriously wrong...\n";
 349                }
 350        }
 351        while (1) {
 352                # fork, because using SVN::Pool with get_log() still doesn't
 353                # seem to help enough to keep memory usage down.
 354                defined(my $pid = fork) or croak $!;
 355                if (!$pid) {
 356                        $SVN::Error::handler = \&libsvn_skip_unknown_revs;
 357
 358                        # Yes I'm perfectly aware that the fourth argument
 359                        # below is the limit revisions number.  Unfortunately
 360                        # performance sucks with it enabled, so it's much
 361                        # faster to fetch revision ranges instead of relying
 362                        # on the limiter.
 363                        $SVN->dup->get_log([''], $min, $max, 0, 1, 1,
 364                                sub {
 365                                        my $log_msg;
 366                                        if ($last_commit) {
 367                                                $log_msg = libsvn_fetch(
 368                                                        $last_commit, @_);
 369                                                $last_commit = git_commit(
 370                                                        $log_msg,
 371                                                        $last_commit,
 372                                                        @parents);
 373                                        } else {
 374                                                $log_msg = libsvn_new_tree(@_);
 375                                                $last_commit = git_commit(
 376                                                        $log_msg, @parents);
 377                                        }
 378                                });
 379                        exit 0;
 380                }
 381                waitpid $pid, 0;
 382                croak $? if $?;
 383                ($last_rev, $last_commit) = svn_grab_base_rev();
 384                last if ($max >= $head);
 385                $min = $max + 1;
 386                $max += $inc;
 387                $max = $head if ($max > $head);
 388                $SVN = Git::SVN::Ra->new($SVN_URL);
 389        }
 390        restore_index($index);
 391        return { revision => $last_rev, commit => $last_commit };
 392}
 393
 394sub commit {
 395        my (@commits) = @_;
 396        check_upgrade_needed();
 397        if ($_stdin || !@commits) {
 398                print "Reading from stdin...\n";
 399                @commits = ();
 400                while (<STDIN>) {
 401                        if (/\b($sha1_short)\b/o) {
 402                                unshift @commits, $1;
 403                        }
 404                }
 405        }
 406        my @revs;
 407        foreach my $c (@commits) {
 408                my @tmp = command('rev-parse',$c);
 409                if (scalar @tmp == 1) {
 410                        push @revs, $tmp[0];
 411                } elsif (scalar @tmp > 1) {
 412                        push @revs, reverse(command('rev-list',@tmp));
 413                } else {
 414                        die "Failed to rev-parse $c\n";
 415                }
 416        }
 417        commit_lib(@revs);
 418        print "Done committing ",scalar @revs," revisions to SVN\n";
 419}
 420
 421sub commit_lib {
 422        my (@revs) = @_;
 423        my ($r_last, $cmt_last) = svn_grab_base_rev();
 424        defined $r_last or die "Must have an existing revision to commit\n";
 425        my $fetched = fetch();
 426        if ($r_last != $fetched->{revision}) {
 427                print STDERR "There are new revisions that were fetched ",
 428                                "and need to be merged (or acknowledged) ",
 429                                "before committing.\n",
 430                                "last rev: $r_last\n",
 431                                " current: $fetched->{revision}\n";
 432                exit 1;
 433        }
 434        my $commit_msg = "$GIT_SVN_DIR/.svn-commit.tmp.$$";
 435
 436        my $repo;
 437        set_svn_commit_env();
 438        foreach my $c (@revs) {
 439                my $log_msg = get_commit_message($c, $commit_msg);
 440
 441                # fork for each commit because there's a memory leak I
 442                # can't track down... (it's probably in the SVN code)
 443                defined(my $pid = open my $fh, '-|') or croak $!;
 444                if (!$pid) {
 445                        my $pool = SVN::Pool->new;
 446                        my $ed = SVN::Git::Editor->new(
 447                                        {       r => $r_last,
 448                                                ra => $SVN->dup,
 449                                                c => $c,
 450                                                svn_path => $SVN->{svn_path},
 451                                        },
 452                                        $SVN->get_commit_editor(
 453                                                $log_msg->{msg},
 454                                                sub {
 455                                                        libsvn_commit_cb(
 456                                                                @_, $c,
 457                                                                $log_msg->{msg},
 458                                                                $r_last,
 459                                                                $cmt_last)
 460                                                }, $pool)
 461                                        );
 462                        my $mods = libsvn_checkout_tree($cmt_last, $c, $ed);
 463                        if (@$mods == 0) {
 464                                print "No changes\nr$r_last = $cmt_last\n";
 465                                $ed->abort_edit;
 466                        } else {
 467                                $ed->close_edit;
 468                        }
 469                        $pool->clear;
 470                        exit 0;
 471                }
 472                my ($r_new, $cmt_new, $no);
 473                while (<$fh>) {
 474                        print $_;
 475                        chomp;
 476                        if (/^r(\d+) = ($sha1)$/o) {
 477                                ($r_new, $cmt_new) = ($1, $2);
 478                        } elsif ($_ eq 'No changes') {
 479                                $no = 1;
 480                        }
 481                }
 482                close $fh or exit 1;
 483                if (! defined $r_new && ! defined $cmt_new) {
 484                        unless ($no) {
 485                                die "Failed to parse revision information\n";
 486                        }
 487                } else {
 488                        ($r_last, $cmt_last) = ($r_new, $cmt_new);
 489                }
 490        }
 491        $ENV{LC_ALL} = 'C';
 492        unlink $commit_msg;
 493}
 494
 495sub dcommit {
 496        my $head = shift || 'HEAD';
 497        my $gs = "refs/remotes/$GIT_SVN";
 498        my @refs = command(qw/rev-list --no-merges/, "$gs..$head");
 499        my $last_rev;
 500        foreach my $d (reverse @refs) {
 501                if (!verify_ref("$d~1")) {
 502                        die "Commit $d\n",
 503                            "has no parent commit, and therefore ",
 504                            "nothing to diff against.\n",
 505                            "You should be working from a repository ",
 506                            "originally created by git-svn\n";
 507                }
 508                unless (defined $last_rev) {
 509                        (undef, $last_rev, undef) = cmt_metadata("$d~1");
 510                        unless (defined $last_rev) {
 511                                die "Unable to extract revision information ",
 512                                    "from commit $d~1\n";
 513                        }
 514                }
 515                if ($_dry_run) {
 516                        print "diff-tree $d~1 $d\n";
 517                } else {
 518                        if (my $r = commit_diff("$d~1", $d, undef, $last_rev)) {
 519                                $last_rev = $r;
 520                        } # else: no changes, same $last_rev
 521                }
 522        }
 523        return if $_dry_run;
 524        fetch();
 525        my @diff = command('diff-tree', 'HEAD', $gs, '--');
 526        my @finish;
 527        if (@diff) {
 528                @finish = qw/rebase/;
 529                push @finish, qw/--merge/ if $_merge;
 530                push @finish, "--strategy=$_strategy" if $_strategy;
 531                print STDERR "W: HEAD and $gs differ, using @finish:\n", @diff;
 532        } else {
 533                print "No changes between current HEAD and $gs\n",
 534                      "Resetting to the latest $gs\n";
 535                @finish = qw/reset --mixed/;
 536        }
 537        command_noisy(@finish, $gs);
 538}
 539
 540sub show_ignore {
 541        $SVN_URL ||= file_to_s("$GIT_SVN_DIR/info/url");
 542        my $repo;
 543        $SVN ||= Git::SVN::Ra->new($SVN_URL);
 544        my $r = defined $_revision ? $_revision : $SVN->get_latest_revnum;
 545        libsvn_traverse_ignore(\*STDOUT, '', $r);
 546}
 547
 548sub graft_branches {
 549        my $gr_file = "$GIT_DIR/info/grafts";
 550        my ($grafts, $comments) = read_grafts($gr_file);
 551        my $gr_sha1;
 552
 553        if (%$grafts) {
 554                # temporarily disable our grafts file to make this idempotent
 555                chomp($gr_sha1 = command(qw/hash-object -w/,$gr_file));
 556                rename $gr_file, "$gr_file~$gr_sha1" or croak $!;
 557        }
 558
 559        my $l_map = read_url_paths();
 560        my @re = map { qr/$_/is } @_opt_m if @_opt_m;
 561        unless ($_no_default_regex) {
 562                push @re, (qr/\b(?:merge|merging|merged)\s+with\s+([\w\.\-]+)/i,
 563                        qr/\b(?:merge|merging|merged)\s+([\w\.\-]+)/i,
 564                        qr/\b(?:from|of)\s+([\w\.\-]+)/i );
 565        }
 566        foreach my $u (keys %$l_map) {
 567                if (@re) {
 568                        foreach my $p (keys %{$l_map->{$u}}) {
 569                                graft_merge_msg($grafts,$l_map,$u,$p,@re);
 570                        }
 571                }
 572                unless ($_no_graft_copy) {
 573                        graft_file_copy_lib($grafts,$l_map,$u);
 574                }
 575        }
 576        graft_tree_joins($grafts);
 577
 578        write_grafts($grafts, $comments, $gr_file);
 579        unlink "$gr_file~$gr_sha1" if $gr_sha1;
 580}
 581
 582sub cmd_multi_init {
 583        my $url = shift;
 584        unless (defined $_trunk || defined $_branches || defined $_tags) {
 585                usage(1);
 586        }
 587        do_git_init_db();
 588        $_prefix = '' unless defined $_prefix;
 589        if (defined $_trunk) {
 590                my $gs_trunk = eval { Git::SVN->new($_prefix . 'trunk') };
 591                unless ($gs_trunk) {
 592                        my $trunk_url = complete_svn_url($url, $_trunk);
 593                        $gs_trunk = Git::SVN->init($_prefix . 'trunk',
 594                                                   $trunk_url);
 595                        command_noisy('config', 'svn.trunk', $trunk_url);
 596                }
 597        }
 598        my $ra = $url ? Git::SVN::Ra->new($url) : undef;
 599        complete_url_ls_init($ra, $_branches, '--branches/-b', $_prefix);
 600        complete_url_ls_init($ra, $_tags, '--tags/-t', $_prefix . 'tags/');
 601}
 602
 603sub multi_fetch {
 604        # try to do trunk first, since branches/tags
 605        # may be descended from it.
 606        if (-e "$GIT_DIR/svn/trunk/info/url") {
 607                fetch_child_id('trunk', @_);
 608        }
 609        rec_fetch('', "$GIT_DIR/svn", @_);
 610}
 611
 612sub show_log {
 613        my (@args) = @_;
 614        my ($r_min, $r_max);
 615        my $r_last = -1; # prevent dupes
 616        rload_authors() if $_authors;
 617        if (defined $TZ) {
 618                $ENV{TZ} = $TZ;
 619        } else {
 620                delete $ENV{TZ};
 621        }
 622        if (defined $_revision) {
 623                if ($_revision =~ /^(\d+):(\d+)$/) {
 624                        ($r_min, $r_max) = ($1, $2);
 625                } elsif ($_revision =~ /^\d+$/) {
 626                        $r_min = $r_max = $_revision;
 627                } else {
 628                        print STDERR "-r$_revision is not supported, use ",
 629                                "standard \'git log\' arguments instead\n";
 630                        exit 1;
 631                }
 632        }
 633
 634        config_pager();
 635        @args = (git_svn_log_cmd($r_min, $r_max), @args);
 636        my $log = command_output_pipe(@args);
 637        run_pager();
 638        my (@k, $c, $d);
 639
 640        while (<$log>) {
 641                if (/^${_esc_color}commit ($sha1_short)/o) {
 642                        my $cmt = $1;
 643                        if ($c && cmt_showable($c) && $c->{r} != $r_last) {
 644                                $r_last = $c->{r};
 645                                process_commit($c, $r_min, $r_max, \@k) or
 646                                                                goto out;
 647                        }
 648                        $d = undef;
 649                        $c = { c => $cmt };
 650                } elsif (/^${_esc_color}author (.+) (\d+) ([\-\+]?\d+)$/) {
 651                        get_author_info($c, $1, $2, $3);
 652                } elsif (/^${_esc_color}(?:tree|parent|committer) /) {
 653                        # ignore
 654                } elsif (/^${_esc_color}:\d{6} \d{6} $sha1_short/o) {
 655                        push @{$c->{raw}}, $_;
 656                } elsif (/^${_esc_color}[ACRMDT]\t/) {
 657                        # we could add $SVN->{svn_path} here, but that requires
 658                        # remote access at the moment (repo_path_split)...
 659                        s#^(${_esc_color})([ACRMDT])\t#$1   $2 #;
 660                        push @{$c->{changed}}, $_;
 661                } elsif (/^${_esc_color}diff /) {
 662                        $d = 1;
 663                        push @{$c->{diff}}, $_;
 664                } elsif ($d) {
 665                        push @{$c->{diff}}, $_;
 666                } elsif (/^${_esc_color}    (git-svn-id:.+)$/) {
 667                        ($c->{url}, $c->{r}, undef) = extract_metadata($1);
 668                } elsif (s/^${_esc_color}    //) {
 669                        push @{$c->{l}}, $_;
 670                }
 671        }
 672        if ($c && defined $c->{r} && $c->{r} != $r_last) {
 673                $r_last = $c->{r};
 674                process_commit($c, $r_min, $r_max, \@k);
 675        }
 676        if (@k) {
 677                my $swap = $r_max;
 678                $r_max = $r_min;
 679                $r_min = $swap;
 680                process_commit($_, $r_min, $r_max) foreach reverse @k;
 681        }
 682out:
 683        close $log;
 684        print '-' x72,"\n" unless $_incremental || $_oneline;
 685}
 686
 687sub commit_diff_usage {
 688        print STDERR "Usage: $0 commit-diff <tree-ish> <tree-ish> [<URL>]\n";
 689        exit 1
 690}
 691
 692sub commit_diff {
 693        my $ta = shift or commit_diff_usage();
 694        my $tb = shift or commit_diff_usage();
 695        if (!eval { $SVN_URL = shift || file_to_s("$GIT_SVN_DIR/info/url") }) {
 696                print STDERR "Needed URL or usable git-svn id command-line\n";
 697                commit_diff_usage();
 698        }
 699        my $r = shift;
 700        unless (defined $r) {
 701                if (defined $_revision) {
 702                        $r = $_revision
 703                } else {
 704                        die "-r|--revision is a required argument\n";
 705                }
 706        }
 707        if (defined $_message && defined $_file) {
 708                print STDERR "Both --message/-m and --file/-F specified ",
 709                                "for the commit message.\n",
 710                                "I have no idea what you mean\n";
 711                exit 1;
 712        }
 713        if (defined $_file) {
 714                $_message = file_to_s($_file);
 715        } else {
 716                $_message ||= get_commit_message($tb,
 717                                        "$GIT_DIR/.svn-commit.tmp.$$")->{msg};
 718        }
 719        $SVN ||= Git::SVN::Ra->new($SVN_URL);
 720        if ($r eq 'HEAD') {
 721                $r = $SVN->get_latest_revnum;
 722        } elsif ($r !~ /^\d+$/) {
 723                die "revision argument: $r not understood by git-svn\n";
 724        }
 725        my $rev_committed;
 726        my $pool = SVN::Pool->new;
 727        my $ed = SVN::Git::Editor->new({        r => $r,
 728                                                ra => $SVN->dup,
 729                                                c => $tb,
 730                                                svn_path => $SVN->{svn_path}
 731                                        },
 732                                $SVN->get_commit_editor($_message,
 733                                        sub {
 734                                                $rev_committed = $_[0];
 735                                                print "Committed $_[0]\n";
 736                                        },
 737                                        $pool)
 738                                );
 739        eval {
 740                my $mods = libsvn_checkout_tree($ta, $tb, $ed);
 741                if (@$mods == 0) {
 742                        print "No changes\n$ta == $tb\n";
 743                        $ed->abort_edit;
 744                } else {
 745                        $ed->close_edit;
 746                }
 747        };
 748        $pool->clear;
 749        fatal "$@\n" if $@;
 750        $_message = $_file = undef;
 751        return $rev_committed;
 752}
 753
 754########################### utility functions #########################
 755
 756sub cmt_showable {
 757        my ($c) = @_;
 758        return 1 if defined $c->{r};
 759        if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
 760                                $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
 761                my @msg = command(qw/cat-file commit/, $c->{c});
 762                shift @msg while ($msg[0] ne "\n");
 763                shift @msg;
 764                @{$c->{l}} = grep !/^git-svn-id: /, @msg;
 765
 766                (undef, $c->{r}, undef) = extract_metadata(
 767                                (grep(/^git-svn-id: /, @msg))[-1]);
 768        }
 769        return defined $c->{r};
 770}
 771
 772sub log_use_color {
 773        return 1 if $_color;
 774        my ($dc, $dcvar);
 775        $dcvar = 'color.diff';
 776        $dc = `git-config --get $dcvar`;
 777        if ($dc eq '') {
 778                # nothing at all; fallback to "diff.color"
 779                $dcvar = 'diff.color';
 780                $dc = `git-config --get $dcvar`;
 781        }
 782        chomp($dc);
 783        if ($dc eq 'auto') {
 784                my $pc;
 785                $pc = `git-config --get color.pager`;
 786                if ($pc eq '') {
 787                        # does not have it -- fallback to pager.color
 788                        $pc = `git-config --bool --get pager.color`;
 789                }
 790                else {
 791                        $pc = `git-config --bool --get color.pager`;
 792                        if ($?) {
 793                                $pc = 'false';
 794                        }
 795                }
 796                chomp($pc);
 797                if (-t *STDOUT || (defined $_pager && $pc eq 'true')) {
 798                        return ($ENV{TERM} && $ENV{TERM} ne 'dumb');
 799                }
 800                return 0;
 801        }
 802        return 0 if $dc eq 'never';
 803        return 1 if $dc eq 'always';
 804        chomp($dc = `git-config --bool --get $dcvar`);
 805        return ($dc eq 'true');
 806}
 807
 808sub git_svn_log_cmd {
 809        my ($r_min, $r_max) = @_;
 810        my @cmd = (qw/log --abbrev-commit --pretty=raw
 811                        --default/, "refs/remotes/$GIT_SVN");
 812        push @cmd, '-r' unless $_non_recursive;
 813        push @cmd, qw/--raw --name-status/ if $_verbose;
 814        push @cmd, '--color' if log_use_color();
 815        return @cmd unless defined $r_max;
 816        if ($r_max == $r_min) {
 817                push @cmd, '--max-count=1';
 818                if (my $c = revdb_get($REVDB, $r_max)) {
 819                        push @cmd, $c;
 820                }
 821        } else {
 822                my ($c_min, $c_max);
 823                $c_max = revdb_get($REVDB, $r_max);
 824                $c_min = revdb_get($REVDB, $r_min);
 825                if (defined $c_min && defined $c_max) {
 826                        if ($r_max > $r_max) {
 827                                push @cmd, "$c_min..$c_max";
 828                        } else {
 829                                push @cmd, "$c_max..$c_min";
 830                        }
 831                } elsif ($r_max > $r_min) {
 832                        push @cmd, $c_max;
 833                } else {
 834                        push @cmd, $c_min;
 835                }
 836        }
 837        return @cmd;
 838}
 839
 840sub fetch_child_id {
 841        my $id = shift;
 842        print "Fetching $id\n";
 843        my $ref = "$GIT_DIR/refs/remotes/$id";
 844        defined(my $pid = open my $fh, '-|') or croak $!;
 845        if (!$pid) {
 846                $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
 847                init_vars();
 848                fetch(@_);
 849                exit 0;
 850        }
 851        while (<$fh>) {
 852                print $_;
 853                check_repack() if (/^r\d+ = $sha1/o);
 854        }
 855        close $fh or croak $?;
 856}
 857
 858sub rec_fetch {
 859        my ($pfx, $p, @args) = @_;
 860        my @dir;
 861        foreach (sort <$p/*>) {
 862                if (-r "$_/info/url") {
 863                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
 864                        my $id = $pfx . basename $_;
 865                        next if $id eq 'trunk';
 866                        fetch_child_id($id, @args);
 867                } elsif (-d $_) {
 868                        push @dir, $_;
 869                }
 870        }
 871        foreach (@dir) {
 872                my $x = $_;
 873                $x =~ s!^\Q$GIT_DIR\E/svn/!!;
 874                rec_fetch($x, $_);
 875        }
 876}
 877
 878sub complete_svn_url {
 879        my ($url, $path) = @_;
 880        $path =~ s#/+$##;
 881        $url =~ s#/+$## if $url;
 882        if ($path !~ m#^[a-z\+]+://#) {
 883                $path = '/' . $path if ($path !~ m#^/#);
 884                if (!defined $url || $url !~ m#^[a-z\+]+://#) {
 885                        fatal("E: '$path' is not a complete URL ",
 886                              "and a separate URL is not specified\n");
 887                }
 888                $path = $url . $path;
 889        }
 890        return $path;
 891}
 892
 893sub complete_url_ls_init {
 894        my ($ra, $path, $switch, $pfx) = @_;
 895        unless ($path) {
 896                print STDERR "W: $switch not specified\n";
 897                return;
 898        }
 899        $path =~ s#/+$##;
 900        if ($path =~ m#^[a-z\+]+://#) {
 901                $ra = Git::SVN::Ra->new($path);
 902                $path = '';
 903        } else {
 904                $path =~ s#^/+##;
 905                unless ($ra) {
 906                        fatal("E: '$path' is not a complete URL ",
 907                              "and a separate URL is not specified\n");
 908                }
 909        }
 910        my $r = defined $_revision ? $_revision : $ra->get_latest_revnum;
 911        my ($dirent, undef, undef) = $ra->get_dir($path, $r);
 912        my $url = $ra->{url} . (length $path ? "/$path" : '');
 913        foreach my $d (sort keys %$dirent) {
 914                next if ($dirent->{$d}->kind != $SVN::Node::dir);
 915                my $u =  "$url/$d";
 916                my $id = "$pfx$d";
 917                my $gs = eval { Git::SVN->new($id) };
 918                # don't try to init already existing refs
 919                unless ($gs) {
 920                        print "init $u => $id\n";
 921                        Git::SVN->init($id, $u);
 922                }
 923        }
 924        my ($n) = ($switch =~ /^--(\w+)/);
 925        command_noisy('config', "svn.$n", $url);
 926}
 927
 928sub common_prefix {
 929        my $paths = shift;
 930        my %common;
 931        foreach (@$paths) {
 932                my @tmp = split m#/#, $_;
 933                my $p = '';
 934                while (my $x = shift @tmp) {
 935                        $p .= "/$x";
 936                        $common{$p} ||= 0;
 937                        $common{$p}++;
 938                }
 939        }
 940        foreach (sort {length $b <=> length $a} keys %common) {
 941                if ($common{$_} == @$paths) {
 942                        return $_;
 943                }
 944        }
 945        return '';
 946}
 947
 948# grafts set here are 'stronger' in that they're based on actual tree
 949# matches, and won't be deleted from merge-base checking in write_grafts()
 950sub graft_tree_joins {
 951        my $grafts = shift;
 952        map_tree_joins() if (@_branch_from && !%tree_map);
 953        return unless %tree_map;
 954
 955        git_svn_each(sub {
 956                my $i = shift;
 957                my @args = (qw/rev-list --pretty=raw/, "refs/remotes/$i");
 958                my ($fh, $ctx) = command_output_pipe(@args);
 959                while (<$fh>) {
 960                        next unless /^commit ($sha1)$/o;
 961                        my $c = $1;
 962                        my ($t) = (<$fh> =~ /^tree ($sha1)$/o);
 963                        next unless $tree_map{$t};
 964
 965                        my $l;
 966                        do {
 967                                $l = readline $fh;
 968                        } until ($l =~ /^committer (?:.+) (\d+) ([\-\+]?\d+)$/);
 969
 970                        my ($s, $tz) = ($1, $2);
 971                        if ($tz =~ s/^\+//) {
 972                                $s += tz_to_s_offset($tz);
 973                        } elsif ($tz =~ s/^\-//) {
 974                                $s -= tz_to_s_offset($tz);
 975                        }
 976
 977                        my ($url_a, $r_a, $uuid_a) = cmt_metadata($c);
 978
 979                        foreach my $p (@{$tree_map{$t}}) {
 980                                next if $p eq $c;
 981                                my $mb = eval { command('merge-base', $c, $p) };
 982                                next unless ($@ || $?);
 983                                if (defined $r_a) {
 984                                        # see if SVN says it's a relative
 985                                        my ($url_b, $r_b, $uuid_b) =
 986                                                        cmt_metadata($p);
 987                                        next if (defined $url_b &&
 988                                                        defined $url_a &&
 989                                                        ($url_a eq $url_b) &&
 990                                                        ($uuid_a eq $uuid_b));
 991                                        if ($uuid_a eq $uuid_b) {
 992                                                if ($r_b < $r_a) {
 993                                                        $grafts->{$c}->{$p} = 2;
 994                                                        next;
 995                                                } elsif ($r_b > $r_a) {
 996                                                        $grafts->{$p}->{$c} = 2;
 997                                                        next;
 998                                                }
 999                                        }
1000                                }
1001                                my $ct = get_commit_time($p);
1002                                if ($ct < $s) {
1003                                        $grafts->{$c}->{$p} = 2;
1004                                } elsif ($ct > $s) {
1005                                        $grafts->{$p}->{$c} = 2;
1006                                }
1007                                # what should we do when $ct == $s ?
1008                        }
1009                }
1010                command_close_pipe($fh, $ctx);
1011        });
1012}
1013
1014sub graft_file_copy_lib {
1015        my ($grafts, $l_map, $u) = @_;
1016        my $tree_paths = $l_map->{$u};
1017        my $pfx = common_prefix([keys %$tree_paths]);
1018        my ($repo, $path) = repo_path_split($u.$pfx);
1019        $SVN = Git::SVN::Ra->new($repo);
1020
1021        my ($base, $head) = libsvn_parse_revision();
1022        my $inc = 1000;
1023        my ($min, $max) = ($base, $head < $base+$inc ? $head : $base+$inc);
1024        my $eh = $SVN::Error::handler;
1025        $SVN::Error::handler = \&libsvn_skip_unknown_revs;
1026        while (1) {
1027                $SVN->dup->get_log([$path], $min, $max, 0, 2, 1,
1028                        sub {
1029                                libsvn_graft_file_copies($grafts, $tree_paths,
1030                                                        $path, @_);
1031                        });
1032                last if ($max >= $head);
1033                $min = $max + 1;
1034                $max += $inc;
1035                $max = $head if ($max > $head);
1036        }
1037        $SVN::Error::handler = $eh;
1038}
1039
1040sub process_merge_msg_matches {
1041        my ($grafts, $l_map, $u, $p, $c, @matches) = @_;
1042        my (@strong, @weak);
1043        foreach (@matches) {
1044                # merging with ourselves is not interesting
1045                next if $_ eq $p;
1046                if ($l_map->{$u}->{$_}) {
1047                        push @strong, $_;
1048                } else {
1049                        push @weak, $_;
1050                }
1051        }
1052        foreach my $w (@weak) {
1053                last if @strong;
1054                # no exact match, use branch name as regexp.
1055                my $re = qr/\Q$w\E/i;
1056                foreach (keys %{$l_map->{$u}}) {
1057                        if (/$re/) {
1058                                push @strong, $l_map->{$u}->{$_};
1059                                last;
1060                        }
1061                }
1062                last if @strong;
1063                $w = basename($w);
1064                $re = qr/\Q$w\E/i;
1065                foreach (keys %{$l_map->{$u}}) {
1066                        if (/$re/) {
1067                                push @strong, $l_map->{$u}->{$_};
1068                                last;
1069                        }
1070                }
1071        }
1072        my ($rev) = ($c->{m} =~ /^git-svn-id:\s(?:\S+?)\@(\d+)
1073                                        \s(?:[a-f\d\-]+)$/xsm);
1074        unless (defined $rev) {
1075                ($rev) = ($c->{m} =~/^git-svn-id:\s(\d+)
1076                                        \@(?:[a-f\d\-]+)/xsm);
1077                return unless defined $rev;
1078        }
1079        foreach my $m (@strong) {
1080                my ($r0, $s0) = find_rev_before($rev, $m, 1);
1081                $grafts->{$c->{c}}->{$s0} = 1 if defined $s0;
1082        }
1083}
1084
1085sub graft_merge_msg {
1086        my ($grafts, $l_map, $u, $p, @re) = @_;
1087
1088        my $x = $l_map->{$u}->{$p};
1089        my $rl = rev_list_raw("refs/remotes/$x");
1090        while (my $c = next_rev_list_entry($rl)) {
1091                foreach my $re (@re) {
1092                        my (@br) = ($c->{m} =~ /$re/g);
1093                        next unless @br;
1094                        process_merge_msg_matches($grafts,$l_map,$u,$p,$c,@br);
1095                }
1096        }
1097}
1098
1099sub verify_ref {
1100        my ($ref) = @_;
1101        eval { command_oneline([ 'rev-parse', '--verify', $ref ],
1102                               { STDERR => 0 }); };
1103}
1104
1105sub repo_path_split {
1106        my $full_url = shift;
1107        $full_url =~ s#/+$##;
1108
1109        foreach (@repo_path_split_cache) {
1110                if ($full_url =~ s#$_##) {
1111                        my $u = $1;
1112                        $full_url =~ s#^/+##;
1113                        return ($u, $full_url);
1114                }
1115        }
1116        my $tmp = Git::SVN::Ra->new($full_url);
1117        return ($tmp->{repos_root}, $tmp->{svn_path});
1118}
1119
1120sub setup_git_svn {
1121        defined $SVN_URL or croak "SVN repository location required\n";
1122        unless (-d $GIT_DIR) {
1123                croak "GIT_DIR=$GIT_DIR does not exist!\n";
1124        }
1125        mkpath([$GIT_SVN_DIR]);
1126        mkpath(["$GIT_SVN_DIR/info"]);
1127        open my $fh, '>>',$REVDB or croak $!;
1128        close $fh;
1129        s_to_file($SVN_URL,"$GIT_SVN_DIR/info/url");
1130
1131}
1132
1133sub get_tree_from_treeish {
1134        my ($treeish) = @_;
1135        croak "Not a sha1: $treeish\n" unless $treeish =~ /^$sha1$/o;
1136        my $type = command_oneline(qw/cat-file -t/, $treeish);
1137        my $expected;
1138        while ($type eq 'tag') {
1139                ($treeish, $type) = command(qw/cat-file tag/, $treeish);
1140        }
1141        if ($type eq 'commit') {
1142                $expected = (grep /^tree /, command(qw/cat-file commit/,
1143                                                    $treeish))[0];
1144                ($expected) = ($expected =~ /^tree ($sha1)$/);
1145                die "Unable to get tree from $treeish\n" unless $expected;
1146        } elsif ($type eq 'tree') {
1147                $expected = $treeish;
1148        } else {
1149                die "$treeish is a $type, expected tree, tag or commit\n";
1150        }
1151        return $expected;
1152}
1153
1154sub get_diff {
1155        my ($from, $treeish) = @_;
1156        print "diff-tree $from $treeish\n";
1157        my @diff_tree = qw(diff-tree -z -r);
1158        if ($_cp_similarity) {
1159                push @diff_tree, "-C$_cp_similarity";
1160        } else {
1161                push @diff_tree, '-C';
1162        }
1163        push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
1164        push @diff_tree, "-l$_l" if defined $_l;
1165        push @diff_tree, $from, $treeish;
1166        my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
1167        local $/ = "\0";
1168        my $state = 'meta';
1169        my @mods;
1170        while (<$diff_fh>) {
1171                chomp $_; # this gets rid of the trailing "\0"
1172                if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
1173                                        $sha1\s($sha1)\s([MTCRAD])\d*$/xo) {
1174                        push @mods, {   mode_a => $1, mode_b => $2,
1175                                        sha1_b => $3, chg => $4 };
1176                        if ($4 =~ /^(?:C|R)$/) {
1177                                $state = 'file_a';
1178                        } else {
1179                                $state = 'file_b';
1180                        }
1181                } elsif ($state eq 'file_a') {
1182                        my $x = $mods[$#mods] or croak "Empty array\n";
1183                        if ($x->{chg} !~ /^(?:C|R)$/) {
1184                                croak "Error parsing $_, $x->{chg}\n";
1185                        }
1186                        $x->{file_a} = $_;
1187                        $state = 'file_b';
1188                } elsif ($state eq 'file_b') {
1189                        my $x = $mods[$#mods] or croak "Empty array\n";
1190                        if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
1191                                croak "Error parsing $_, $x->{chg}\n";
1192                        }
1193                        if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
1194                                croak "Error parsing $_, $x->{chg}\n";
1195                        }
1196                        $x->{file_b} = $_;
1197                        $state = 'meta';
1198                } else {
1199                        croak "Error parsing $_\n";
1200                }
1201        }
1202        command_close_pipe($diff_fh, $ctx);
1203        return \@mods;
1204}
1205
1206sub libsvn_checkout_tree {
1207        my ($from, $treeish, $ed) = @_;
1208        my $mods = get_diff($from, $treeish);
1209        return $mods unless (scalar @$mods);
1210        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
1211        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
1212                my $f = $m->{chg};
1213                if (defined $o{$f}) {
1214                        $ed->$f($m, $_q);
1215                } else {
1216                        croak "Invalid change type: $f\n";
1217                }
1218        }
1219        $ed->rmdirs($_q) if $_rmdir;
1220        return $mods;
1221}
1222
1223sub get_commit_message {
1224        my ($commit, $commit_msg) = (@_);
1225        my %log_msg = ( msg => '' );
1226        open my $msg, '>', $commit_msg or croak $!;
1227
1228        my $type = command_oneline(qw/cat-file -t/, $commit);
1229        if ($type eq 'commit' || $type eq 'tag') {
1230                my ($msg_fh, $ctx) = command_output_pipe('cat-file',
1231                                                         $type, $commit);
1232                my $in_msg = 0;
1233                while (<$msg_fh>) {
1234                        if (!$in_msg) {
1235                                $in_msg = 1 if (/^\s*$/);
1236                        } elsif (/^git-svn-id: /) {
1237                                # skip this, we regenerate the correct one
1238                                # on re-fetch anyways
1239                        } else {
1240                                print $msg $_ or croak $!;
1241                        }
1242                }
1243                command_close_pipe($msg_fh, $ctx);
1244        }
1245        close $msg or croak $!;
1246
1247        if ($_edit || ($type eq 'tree')) {
1248                my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
1249                system($editor, $commit_msg);
1250        }
1251
1252        # file_to_s removes all trailing newlines, so just use chomp() here:
1253        open $msg, '<', $commit_msg or croak $!;
1254        { local $/; chomp($log_msg{msg} = <$msg>); }
1255        close $msg or croak $!;
1256
1257        return \%log_msg;
1258}
1259
1260sub set_svn_commit_env {
1261        if (defined $LC_ALL) {
1262                $ENV{LC_ALL} = $LC_ALL;
1263        } else {
1264                delete $ENV{LC_ALL};
1265        }
1266}
1267
1268sub rev_list_raw {
1269        my ($fh, $c) = command_output_pipe(qw/rev-list --pretty=raw/, @_);
1270        return { fh => $fh, ctx => $c, t => { } };
1271}
1272
1273sub next_rev_list_entry {
1274        my $rl = shift;
1275        my $fh = $rl->{fh};
1276        my $x = $rl->{t};
1277        while (<$fh>) {
1278                if (/^commit ($sha1)$/o) {
1279                        if ($x->{c}) {
1280                                $rl->{t} = { c => $1 };
1281                                return $x;
1282                        } else {
1283                                $x->{c} = $1;
1284                        }
1285                } elsif (/^parent ($sha1)$/o) {
1286                        $x->{p}->{$1} = 1;
1287                } elsif (s/^    //) {
1288                        $x->{m} ||= '';
1289                        $x->{m} .= $_;
1290                }
1291        }
1292        command_close_pipe($fh, $rl->{ctx});
1293        return ($x != $rl->{t}) ? $x : undef;
1294}
1295
1296sub s_to_file {
1297        my ($str, $file, $mode) = @_;
1298        open my $fd,'>',$file or croak $!;
1299        print $fd $str,"\n" or croak $!;
1300        close $fd or croak $!;
1301        chmod ($mode &~ umask, $file) if (defined $mode);
1302}
1303
1304sub file_to_s {
1305        my $file = shift;
1306        open my $fd,'<',$file or croak "$!: file: $file\n";
1307        local $/;
1308        my $ret = <$fd>;
1309        close $fd or croak $!;
1310        $ret =~ s/\s*$//s;
1311        return $ret;
1312}
1313
1314sub assert_revision_unknown {
1315        my $r = shift;
1316        if (my $c = revdb_get($REVDB, $r)) {
1317                croak "$r = $c already exists! Why are we refetching it?";
1318        }
1319}
1320
1321sub git_commit {
1322        my ($log_msg, @parents) = @_;
1323        assert_revision_unknown($log_msg->{revision});
1324        map_tree_joins() if (@_branch_from && !%tree_map);
1325
1326        my (@tmp_parents, @exec_parents, %seen_parent);
1327        if (my $lparents = $log_msg->{parents}) {
1328                @tmp_parents = @$lparents
1329        }
1330        # commit parents can be conditionally bound to a particular
1331        # svn revision via: "svn_revno=commit_sha1", filter them out here:
1332        foreach my $p (@parents) {
1333                next unless defined $p;
1334                if ($p =~ /^(\d+)=($sha1_short)$/o) {
1335                        if ($1 == $log_msg->{revision}) {
1336                                push @tmp_parents, $2;
1337                        }
1338                } else {
1339                        push @tmp_parents, $p if $p =~ /$sha1_short/o;
1340                }
1341        }
1342        my $tree = $log_msg->{tree};
1343        if (!defined $tree) {
1344                my $index = set_index($GIT_SVN_INDEX);
1345                $tree = command_oneline('write-tree');
1346                croak $? if $?;
1347                restore_index($index);
1348        }
1349        # just in case we clobber the existing ref, we still want that ref
1350        # as our parent:
1351        if (my $cur = verify_ref("refs/remotes/$GIT_SVN^0")) {
1352                chomp $cur;
1353                push @tmp_parents, $cur;
1354        }
1355
1356        if (exists $tree_map{$tree}) {
1357                foreach my $p (@{$tree_map{$tree}}) {
1358                        my $skip;
1359                        foreach (@tmp_parents) {
1360                                # see if a common parent is found
1361                                my $mb = eval { command('merge-base', $_, $p) };
1362                                next if ($@ || $?);
1363                                $skip = 1;
1364                                last;
1365                        }
1366                        next if $skip;
1367                        my ($url_p, $r_p, $uuid_p) = cmt_metadata($p);
1368                        next if (($SVN->uuid eq $uuid_p) &&
1369                                                ($log_msg->{revision} > $r_p));
1370                        next if (defined $url_p && defined $SVN_URL &&
1371                                                ($SVN->uuid eq $uuid_p) &&
1372                                                ($url_p eq $SVN_URL));
1373                        push @tmp_parents, $p;
1374                }
1375        }
1376        foreach (@tmp_parents) {
1377                next if $seen_parent{$_};
1378                $seen_parent{$_} = 1;
1379                push @exec_parents, $_;
1380                # MAXPARENT is defined to 16 in commit-tree.c:
1381                last if @exec_parents > 16;
1382        }
1383
1384        set_commit_env($log_msg);
1385        my @exec = ('git-commit-tree', $tree);
1386        push @exec, '-p', $_  foreach @exec_parents;
1387        defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
1388                                                                or croak $!;
1389        print $msg_fh $log_msg->{msg} or croak $!;
1390        unless ($_no_metadata) {
1391                print $msg_fh "\ngit-svn-id: $SVN_URL\@$log_msg->{revision} ",
1392                                        $SVN->uuid,"\n" or croak $!;
1393        }
1394        $msg_fh->flush == 0 or croak $!;
1395        close $msg_fh or croak $!;
1396        chomp(my $commit = do { local $/; <$out_fh> });
1397        close $out_fh or croak $!;
1398        waitpid $pid, 0;
1399        croak $? if $?;
1400        if ($commit !~ /^$sha1$/o) {
1401                die "Failed to commit, invalid sha1: $commit\n";
1402        }
1403        command_noisy('update-ref',"refs/remotes/$GIT_SVN",$commit);
1404        revdb_set($REVDB, $log_msg->{revision}, $commit);
1405
1406        # this output is read via pipe, do not change:
1407        print "r$log_msg->{revision} = $commit\n";
1408        return $commit;
1409}
1410
1411sub check_repack {
1412        if ($_repack && (--$_repack_nr == 0)) {
1413                $_repack_nr = $_repack;
1414                # repack doesn't use any arguments with spaces in them, does it?
1415                command_noisy('repack', split(/\s+/, $_repack_flags));
1416        }
1417}
1418
1419sub set_commit_env {
1420        my ($log_msg) = @_;
1421        my $author = $log_msg->{author};
1422        if (!defined $author || length $author == 0) {
1423                $author = '(no author)';
1424        }
1425        my ($name,$email) = defined $users{$author} ?  @{$users{$author}}
1426                                : ($author,$author . '@' . $SVN->uuid);
1427        $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} = $name;
1428        $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} = $email;
1429        $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_msg->{date};
1430}
1431
1432sub check_upgrade_needed {
1433        if (!-r $REVDB) {
1434                -d $GIT_SVN_DIR or mkpath([$GIT_SVN_DIR]);
1435                open my $fh, '>>',$REVDB or croak $!;
1436                close $fh;
1437        }
1438        return unless eval {
1439                command([qw/rev-parse --verify/,"$GIT_SVN-HEAD^0"],
1440                        {STDERR => 0});
1441        };
1442        my $head = eval { command('rev-parse',"refs/remotes/$GIT_SVN") };
1443        if ($@ || !$head) {
1444                print STDERR "Please run: $0 rebuild --upgrade\n";
1445                exit 1;
1446        }
1447}
1448
1449# fills %tree_map with a reverse mapping of trees to commits.  Useful
1450# for finding parents to commit on.
1451sub map_tree_joins {
1452        my %seen;
1453        foreach my $br (@_branch_from) {
1454                my $pipe = command_output_pipe(qw/rev-list
1455                                            --topo-order --pretty=raw/, $br);
1456                while (<$pipe>) {
1457                        if (/^commit ($sha1)$/o) {
1458                                my $commit = $1;
1459
1460                                # if we've seen a commit,
1461                                # we've seen its parents
1462                                last if $seen{$commit};
1463                                my ($tree) = (<$pipe> =~ /^tree ($sha1)$/o);
1464                                unless (defined $tree) {
1465                                        die "Failed to parse commit $commit\n";
1466                                }
1467                                push @{$tree_map{$tree}}, $commit;
1468                                $seen{$commit} = 1;
1469                        }
1470                }
1471                close $pipe;
1472        }
1473}
1474
1475sub load_all_refs {
1476        if (@_branch_from) {
1477                print STDERR '--branch|-b parameters are ignored when ',
1478                        "--branch-all-refs|-B is passed\n";
1479        }
1480
1481        # don't worry about rev-list on non-commit objects/tags,
1482        # it shouldn't blow up if a ref is a blob or tree...
1483        @_branch_from = command(qw/rev-parse --symbolic --all/);
1484}
1485
1486# '<svn username> = real-name <email address>' mapping based on git-svnimport:
1487sub load_authors {
1488        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1489        while (<$authors>) {
1490                chomp;
1491                next unless /^(\S+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
1492                my ($user, $name, $email) = ($1, $2, $3);
1493                $users{$user} = [$name, $email];
1494        }
1495        close $authors or croak $!;
1496}
1497
1498sub rload_authors {
1499        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1500        while (<$authors>) {
1501                chomp;
1502                next unless /^(\S+?)\s*=\s*(.+?)\s*<(.+)>\s*$/;
1503                my ($user, $name, $email) = ($1, $2, $3);
1504                $rusers{"$name <$email>"} = $user;
1505        }
1506        close $authors or croak $!;
1507}
1508
1509sub git_svn_each {
1510        my $sub = shift;
1511        foreach (command(qw/rev-parse --symbolic --all/)) {
1512                next unless s#^refs/remotes/##;
1513                chomp $_;
1514                next unless -f "$GIT_DIR/svn/$_/info/url";
1515                &$sub($_);
1516        }
1517}
1518
1519sub migrate_revdb {
1520        git_svn_each(sub {
1521                my $id = shift;
1522                defined(my $pid = fork) or croak $!;
1523                if (!$pid) {
1524                        $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
1525                        init_vars();
1526                        exit 0 if -r $REVDB;
1527                        print "Upgrading svn => git mapping...\n";
1528                        -d $GIT_SVN_DIR or mkpath([$GIT_SVN_DIR]);
1529                        open my $fh, '>>',$REVDB or croak $!;
1530                        close $fh;
1531                        rebuild();
1532                        print "Done upgrading. You may now delete the ",
1533                                "deprecated $GIT_SVN_DIR/revs directory\n";
1534                        exit 0;
1535                }
1536                waitpid $pid, 0;
1537                croak $? if $?;
1538        });
1539}
1540
1541sub migration_check {
1542        migrate_revdb() unless (-e $REVDB);
1543        return if (-d "$GIT_DIR/svn" || !-d $GIT_DIR);
1544        print "Upgrading repository...\n";
1545        unless (-d "$GIT_DIR/svn") {
1546                mkdir "$GIT_DIR/svn" or croak $!;
1547        }
1548        print "Data from a previous version of git-svn exists, but\n\t",
1549                                "$GIT_SVN_DIR\n\t(required for this version ",
1550                                "($VERSION) of git-svn) does not.\n";
1551
1552        foreach my $x (command(qw/rev-parse --symbolic --all/)) {
1553                next unless $x =~ s#^refs/remotes/##;
1554                chomp $x;
1555                next unless -f "$GIT_DIR/$x/info/url";
1556                my $u = eval { file_to_s("$GIT_DIR/$x/info/url") };
1557                next unless $u;
1558                my $dn = dirname("$GIT_DIR/svn/$x");
1559                mkpath([$dn]) unless -d $dn;
1560                rename "$GIT_DIR/$x", "$GIT_DIR/svn/$x" or croak "$!: $x";
1561        }
1562        migrate_revdb() if (-d $GIT_SVN_DIR && !-w $REVDB);
1563        print "Done upgrading.\n";
1564}
1565
1566sub find_rev_before {
1567        my ($r, $id, $eq_ok) = @_;
1568        my $f = "$GIT_DIR/svn/$id/.rev_db";
1569        return (undef,undef) unless -r $f;
1570        --$r unless $eq_ok;
1571        while ($r > 0) {
1572                if (my $c = revdb_get($f, $r)) {
1573                        return ($r, $c);
1574                }
1575                --$r;
1576        }
1577        return (undef, undef);
1578}
1579
1580sub init_vars {
1581        $GIT_SVN ||= $ENV{GIT_SVN_ID} || 'git-svn';
1582        $Git::SVN::default = $GIT_SVN;
1583        $GIT_SVN_DIR = "$GIT_DIR/svn/$GIT_SVN";
1584        $REVDB = "$GIT_SVN_DIR/.rev_db";
1585        $GIT_SVN_INDEX = "$GIT_SVN_DIR/index";
1586        $SVN_URL = undef;
1587        %tree_map = ();
1588}
1589
1590# convert GetOpt::Long specs for use by git-config
1591sub read_repo_config {
1592        return unless -d $GIT_DIR;
1593        my $opts = shift;
1594        foreach my $o (keys %$opts) {
1595                my $v = $opts->{$o};
1596                my ($key) = ($o =~ /^([a-z\-]+)/);
1597                $key =~ s/-//g;
1598                my $arg = 'git-config';
1599                $arg .= ' --int' if ($o =~ /[:=]i$/);
1600                $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
1601                if (ref $v eq 'ARRAY') {
1602                        chomp(my @tmp = `$arg --get-all svn.$key`);
1603                        @$v = @tmp if @tmp;
1604                } else {
1605                        chomp(my $tmp = `$arg --get svn.$key`);
1606                        if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
1607                                $$v = $tmp;
1608                        }
1609                }
1610        }
1611}
1612
1613sub set_default_vals {
1614        if (defined $_repack) {
1615                $_repack = 1000 if ($_repack <= 0);
1616                $_repack_nr = $_repack;
1617                $_repack_flags ||= '-d';
1618        }
1619}
1620
1621sub read_grafts {
1622        my $gr_file = shift;
1623        my ($grafts, $comments) = ({}, {});
1624        if (open my $fh, '<', $gr_file) {
1625                my @tmp;
1626                while (<$fh>) {
1627                        if (/^($sha1)\s+/) {
1628                                my $c = $1;
1629                                if (@tmp) {
1630                                        @{$comments->{$c}} = @tmp;
1631                                        @tmp = ();
1632                                }
1633                                foreach my $p (split /\s+/, $_) {
1634                                        $grafts->{$c}->{$p} = 1;
1635                                }
1636                        } else {
1637                                push @tmp, $_;
1638                        }
1639                }
1640                close $fh or croak $!;
1641                @{$comments->{'END'}} = @tmp if @tmp;
1642        }
1643        return ($grafts, $comments);
1644}
1645
1646sub write_grafts {
1647        my ($grafts, $comments, $gr_file) = @_;
1648
1649        open my $fh, '>', $gr_file or croak $!;
1650        foreach my $c (sort keys %$grafts) {
1651                if ($comments->{$c}) {
1652                        print $fh $_ foreach @{$comments->{$c}};
1653                }
1654                my $p = $grafts->{$c};
1655                my %x; # real parents
1656                delete $p->{$c}; # commits are not self-reproducing...
1657                my $ch = command_output_pipe(qw/cat-file commit/, $c);
1658                while (<$ch>) {
1659                        if (/^parent ($sha1)/) {
1660                                $x{$1} = $p->{$1} = 1;
1661                        } else {
1662                                last unless /^\S/;
1663                        }
1664                }
1665                close $ch; # breaking the pipe
1666
1667                # if real parents are the only ones in the grafts, drop it
1668                next if join(' ',sort keys %$p) eq join(' ',sort keys %x);
1669
1670                my (@ip, @jp, $mb);
1671                my %del = %x;
1672                @ip = @jp = keys %$p;
1673                foreach my $i (@ip) {
1674                        next if $del{$i} || $p->{$i} == 2;
1675                        foreach my $j (@jp) {
1676                                next if $i eq $j || $del{$j} || $p->{$j} == 2;
1677                                $mb = eval { command('merge-base', $i, $j) };
1678                                next unless $mb;
1679                                chomp $mb;
1680                                next if $x{$mb};
1681                                if ($mb eq $j) {
1682                                        delete $p->{$i};
1683                                        $del{$i} = 1;
1684                                } elsif ($mb eq $i) {
1685                                        delete $p->{$j};
1686                                        $del{$j} = 1;
1687                                }
1688                        }
1689                }
1690
1691                # if real parents are the only ones in the grafts, drop it
1692                next if join(' ',sort keys %$p) eq join(' ',sort keys %x);
1693
1694                print $fh $c, ' ', join(' ', sort keys %$p),"\n";
1695        }
1696        if ($comments->{'END'}) {
1697                print $fh $_ foreach @{$comments->{'END'}};
1698        }
1699        close $fh or croak $!;
1700}
1701
1702sub read_url_paths_all {
1703        my ($l_map, $pfx, $p) = @_;
1704        my @dir;
1705        foreach (<$p/*>) {
1706                if (-r "$_/info/url") {
1707                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
1708                        my $id = $pfx . basename $_;
1709                        my $url = file_to_s("$_/info/url");
1710                        my ($u, $p) = repo_path_split($url);
1711                        $l_map->{$u}->{$p} = $id;
1712                } elsif (-d $_) {
1713                        push @dir, $_;
1714                }
1715        }
1716        foreach (@dir) {
1717                my $x = $_;
1718                $x =~ s!^\Q$GIT_DIR\E/svn/!!o;
1719                read_url_paths_all($l_map, $x, $_);
1720        }
1721}
1722
1723# this one only gets ids that have been imported, not new ones
1724sub read_url_paths {
1725        my $l_map = {};
1726        git_svn_each(sub { my $x = shift;
1727                        my $url = file_to_s("$GIT_DIR/svn/$x/info/url");
1728                        my ($u, $p) = repo_path_split($url);
1729                        $l_map->{$u}->{$p} = $x;
1730                        });
1731        return $l_map;
1732}
1733
1734sub extract_metadata {
1735        my $id = shift or return (undef, undef, undef);
1736        my ($url, $rev, $uuid) = ($id =~ /^git-svn-id:\s(\S+?)\@(\d+)
1737                                                        \s([a-f\d\-]+)$/x);
1738        if (!defined $rev || !$uuid || !$url) {
1739                # some of the original repositories I made had
1740                # identifiers like this:
1741                ($rev, $uuid) = ($id =~/^git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
1742        }
1743        return ($url, $rev, $uuid);
1744}
1745
1746sub cmt_metadata {
1747        return extract_metadata((grep(/^git-svn-id: /,
1748                command(qw/cat-file commit/, shift)))[-1]);
1749}
1750
1751sub get_commit_time {
1752        my $cmt = shift;
1753        my $fh = command_output_pipe(qw/rev-list --pretty=raw -n1/, $cmt);
1754        while (<$fh>) {
1755                /^committer\s(?:.+) (\d+) ([\-\+]?\d+)$/ or next;
1756                my ($s, $tz) = ($1, $2);
1757                if ($tz =~ s/^\+//) {
1758                        $s += tz_to_s_offset($tz);
1759                } elsif ($tz =~ s/^\-//) {
1760                        $s -= tz_to_s_offset($tz);
1761                }
1762                close $fh;
1763                return $s;
1764        }
1765        die "Can't get commit time for commit: $cmt\n";
1766}
1767
1768sub tz_to_s_offset {
1769        my ($tz) = @_;
1770        $tz =~ s/(\d\d)$//;
1771        return ($1 * 60) + ($tz * 3600);
1772}
1773
1774# adapted from pager.c
1775sub config_pager {
1776        $_pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
1777        if (!defined $_pager) {
1778                $_pager = 'less';
1779        } elsif (length $_pager == 0 || $_pager eq 'cat') {
1780                $_pager = undef;
1781        }
1782}
1783
1784sub run_pager {
1785        return unless -t *STDOUT;
1786        pipe my $rfd, my $wfd or return;
1787        defined(my $pid = fork) or croak $!;
1788        if (!$pid) {
1789                open STDOUT, '>&', $wfd or croak $!;
1790                return;
1791        }
1792        open STDIN, '<&', $rfd or croak $!;
1793        $ENV{LESS} ||= 'FRSX';
1794        exec $_pager or croak "Can't run pager: $! ($_pager)\n";
1795}
1796
1797sub get_author_info {
1798        my ($dest, $author, $t, $tz) = @_;
1799        $author =~ s/(?:^\s*|\s*$)//g;
1800        $dest->{a_raw} = $author;
1801        my $_a;
1802        if ($_authors) {
1803                $_a = $rusers{$author} || undef;
1804        }
1805        if (!$_a) {
1806                ($_a) = ($author =~ /<([^>]+)\@[^>]+>$/);
1807        }
1808        $dest->{t} = $t;
1809        $dest->{tz} = $tz;
1810        $dest->{a} = $_a;
1811        # Date::Parse isn't in the standard Perl distro :(
1812        if ($tz =~ s/^\+//) {
1813                $t += tz_to_s_offset($tz);
1814        } elsif ($tz =~ s/^\-//) {
1815                $t -= tz_to_s_offset($tz);
1816        }
1817        $dest->{t_utc} = $t;
1818}
1819
1820sub process_commit {
1821        my ($c, $r_min, $r_max, $defer) = @_;
1822        if (defined $r_min && defined $r_max) {
1823                if ($r_min == $c->{r} && $r_min == $r_max) {
1824                        show_commit($c);
1825                        return 0;
1826                }
1827                return 1 if $r_min == $r_max;
1828                if ($r_min < $r_max) {
1829                        # we need to reverse the print order
1830                        return 0 if (defined $_limit && --$_limit < 0);
1831                        push @$defer, $c;
1832                        return 1;
1833                }
1834                if ($r_min != $r_max) {
1835                        return 1 if ($r_min < $c->{r});
1836                        return 1 if ($r_max > $c->{r});
1837                }
1838        }
1839        return 0 if (defined $_limit && --$_limit < 0);
1840        show_commit($c);
1841        return 1;
1842}
1843
1844sub show_commit {
1845        my $c = shift;
1846        if ($_oneline) {
1847                my $x = "\n";
1848                if (my $l = $c->{l}) {
1849                        while ($l->[0] =~ /^\s*$/) { shift @$l }
1850                        $x = $l->[0];
1851                }
1852                $_l_fmt ||= 'A' . length($c->{r});
1853                print 'r',pack($_l_fmt, $c->{r}),' | ';
1854                print "$c->{c} | " if $_show_commit;
1855                print $x;
1856        } else {
1857                show_commit_normal($c);
1858        }
1859}
1860
1861sub show_commit_changed_paths {
1862        my ($c) = @_;
1863        return unless $c->{changed};
1864        print "Changed paths:\n", @{$c->{changed}};
1865}
1866
1867sub show_commit_normal {
1868        my ($c) = @_;
1869        print '-' x72, "\nr$c->{r} | ";
1870        print "$c->{c} | " if $_show_commit;
1871        print "$c->{a} | ", strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)",
1872                                 localtime($c->{t_utc})), ' | ';
1873        my $nr_line = 0;
1874
1875        if (my $l = $c->{l}) {
1876                while ($l->[$#$l] eq "\n" && $#$l > 0
1877                                          && $l->[($#$l - 1)] eq "\n") {
1878                        pop @$l;
1879                }
1880                $nr_line = scalar @$l;
1881                if (!$nr_line) {
1882                        print "1 line\n\n\n";
1883                } else {
1884                        if ($nr_line == 1) {
1885                                $nr_line = '1 line';
1886                        } else {
1887                                $nr_line .= ' lines';
1888                        }
1889                        print $nr_line, "\n";
1890                        show_commit_changed_paths($c);
1891                        print "\n";
1892                        print $_ foreach @$l;
1893                }
1894        } else {
1895                print "1 line\n";
1896                show_commit_changed_paths($c);
1897                print "\n";
1898
1899        }
1900        foreach my $x (qw/raw diff/) {
1901                if ($c->{$x}) {
1902                        print "\n";
1903                        print $_ foreach @{$c->{$x}}
1904                }
1905        }
1906}
1907
1908package Git::SVN;
1909use strict;
1910use warnings;
1911use vars qw/$default/;
1912use Carp qw/croak/;
1913use File::Path qw/mkpath/;
1914use IPC::Open3;
1915
1916# properties that we do not log:
1917my %SKIP_PROP;
1918BEGIN {
1919        %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
1920                                        svn:special svn:executable
1921                                        svn:entry:committed-rev
1922                                        svn:entry:last-author
1923                                        svn:entry:uuid
1924                                        svn:entry:committed-date/;
1925}
1926
1927sub init {
1928        my ($class, $id, $url) = @_;
1929        my $self = _new($class, $id);
1930        mkpath(["$self->{dir}/info"]);
1931        if (defined $url) {
1932                $url =~ s!/+$!!; # strip trailing slash
1933                ::s_to_file($url, "$self->{dir}/info/url");
1934        }
1935        $self->{url} = $url;
1936        open my $fh, '>>', $self->{db_path} or croak $!;
1937        close $fh or croak $!;
1938        $self;
1939}
1940
1941sub new {
1942        my ($class, $id) = @_;
1943        my $self = _new($class, $id);
1944        $self->{url} = ::file_to_s("$self->{dir}/info/url");
1945        $self;
1946}
1947
1948sub refname { "refs/remotes/$_[0]->{id}" }
1949
1950sub ra {
1951        my ($self) = shift;
1952        $self->{ra} ||= Git::SVN::Ra->new($self->{url});
1953}
1954
1955sub copy_remote_ref {
1956        my ($self) = @_;
1957        my $origin = $::_cp_remote ? $::_cp_remote : 'origin';
1958        my $ref = $self->refname;
1959        if (command('ls-remote', $origin, $ref)) {
1960                command_noisy('fetch', $origin, "$ref:$ref");
1961        } elsif ($::_cp_remote && !$::_upgrade) {
1962                die "Unable to find remote reference: $ref on $origin\n";
1963        }
1964}
1965
1966sub traverse_ignore {
1967        my ($self, $fh, $path, $r) = @_;
1968        $path =~ s#^/+##g;
1969        my ($dirent, undef, $props) = $self->ra->get_dir($path, $r);
1970        my $p = $path;
1971        $p =~ s#^\Q$self->{ra}->{svn_path}\E/##;
1972        print $fh length $p ? "\n# $p\n" : "\n# /\n";
1973        if (my $s = $props->{'svn:ignore'}) {
1974                $s =~ s/[\r\n]+/\n/g;
1975                chomp $s;
1976                if (length $p == 0) {
1977                        $s =~ s#\n#\n/$p#g;
1978                        print $fh "/$s\n";
1979                } else {
1980                        $s =~ s#\n#\n/$p/#g;
1981                        print $fh "/$p/$s\n";
1982                }
1983        }
1984        foreach (sort keys %$dirent) {
1985                next if $dirent->{$_}->kind != $SVN::Node::dir;
1986                $self->traverse_ignore($fh, "$path/$_", $r);
1987        }
1988}
1989
1990# returns the newest SVN revision number and newest commit SHA1
1991sub last_rev_commit {
1992        my ($self) = @_;
1993        if (defined $self->{last_rev} && defined $self->{last_commit}) {
1994                return ($self->{last_rev}, $self->{last_commit});
1995        }
1996        my $c = ::verify_ref($self->refname.'^0');
1997        if (defined $c && length $c) {
1998                my $rev = (::cmt_metadata($c))[1];
1999                if (defined $rev) {
2000                        ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
2001                        return ($rev, $c);
2002                }
2003        }
2004        my $offset = -41; # from tail
2005        my $rl;
2006        open my $fh, '<', $self->{db_path} or
2007                                 croak "$self->{db_path} not readable: $!\n";
2008        seek $fh, $offset, 2;
2009        $rl = readline $fh;
2010        defined $rl or return (undef, undef);
2011        chomp $rl;
2012        while ($c ne $rl && tell $fh != 0) {
2013                $offset -= 41;
2014                seek $fh, $offset, 2;
2015                $rl = readline $fh;
2016                defined $rl or return (undef, undef);
2017                chomp $rl;
2018        }
2019        my $rev = tell $fh;
2020        croak $! if ($rev < 0);
2021        $rev =  ($rev - 41) / 41;
2022        close $fh or croak $!;
2023        ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
2024        return ($rev, $c);
2025}
2026
2027sub parse_revision {
2028        my ($self, $base) = @_;
2029        my $head = $self->ra->get_latest_revnum;
2030        if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
2031                return ($base + 1, $head) if (defined $base);
2032                return (0, $head);
2033        }
2034        return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
2035        return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
2036        if ($::_revision =~ /^BASE:(\d+)$/) {
2037                return ($base + 1, $1) if (defined $base);
2038                return (0, $head);
2039        }
2040        return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
2041        die "revision argument: $::_revision not understood by git-svn\n",
2042                "Try using the command-line svn client instead\n";
2043}
2044
2045sub tmp_index_do {
2046        my ($self, $sub) = @_;
2047        my $old_index = $ENV{GIT_INDEX_FILE};
2048        $ENV{GIT_INDEX_FILE} = $self->{index};
2049        my @ret = &$sub;
2050        if ($old_index) {
2051                $ENV{GIT_INDEX_FILE} = $old_index;
2052        } else {
2053                delete $ENV{GIT_INDEX_FILE};
2054        }
2055        wantarray ? @ret : $ret[0];
2056}
2057
2058sub assert_index_clean {
2059        my ($self, $treeish) = @_;
2060
2061        $self->tmp_index_do(sub {
2062                command_noisy('read-tree', $treeish) unless -e $self->{index};
2063                my $x = command_oneline('write-tree');
2064                my ($y) = (command(qw/cat-file commit/, $treeish) =~
2065                           /^tree ($::sha1)/mo);
2066                if ($y ne $x) {
2067                        unlink $self->{index} or croak $!;
2068                        command_noisy('read-tree', $treeish);
2069                }
2070                $x = command_oneline('write-tree');
2071                if ($y ne $x) {
2072                        ::fatal "trees ($treeish) $y != $x\n",
2073                                "Something is seriously wrong...\n";
2074                }
2075        });
2076}
2077
2078sub get_commit_parents {
2079        my ($self, $log_msg, @parents) = @_;
2080        my (%seen, @ret, @tmp);
2081        # commit parents can be conditionally bound to a particular
2082        # svn revision via: "svn_revno=commit_sha1", filter them out here:
2083        foreach my $p (@parents) {
2084                next unless defined $p;
2085                if ($p =~ /^(\d+)=($::sha1_short)$/o) {
2086                        push @tmp, $2 if $1 == $log_msg->{revision};
2087                } else {
2088                        push @tmp, $p if $p =~ /^$::sha1_short$/o;
2089                }
2090        }
2091        if (my $cur = ::verify_ref($self->refname.'^0')) {
2092                push @tmp, $cur;
2093        }
2094        push @tmp, $_ foreach (@{$log_msg->{parents}}, @tmp);
2095        while (my $p = shift @tmp) {
2096                next if $seen{$p};
2097                $seen{$p} = 1;
2098                push @ret, $p;
2099                # MAXPARENT is defined to 16 in commit-tree.c:
2100                last if @ret >= 16;
2101        }
2102        if (@tmp) {
2103                die "r$log_msg->{revision}: No room for parents:\n\t",
2104                    join("\n\t", @tmp), "\n";
2105        }
2106        @ret;
2107}
2108
2109sub check_upgrade_needed {
2110        my ($self) = @_;
2111        if (!-r $self->{db_path}) {
2112                -d $self->{dir} or mkpath([$self->{dir}]);
2113                open my $fh, '>>', $self->{db_path} or croak $!;
2114                close $fh;
2115        }
2116        return unless ::verify_ref($self->{id}.'-HEAD^0');
2117        my $head = ::verify_ref($self->refname.'^0');
2118        if ($@ || !$head) {
2119                ::fatal("Please run: $0 rebuild --upgrade\n");
2120        }
2121}
2122
2123sub do_git_commit {
2124        my ($self, $log_msg, @parents) = @_;
2125        if (my $c = $self->rev_db_get($log_msg->{revision})) {
2126                croak "$log_msg->{revision} = $c already exists! ",
2127                      "Why are we refetching it?\n";
2128        }
2129        my ($name, $email) = ::author_name_email($log_msg->{author}, $self->ra);
2130        $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} = $name;
2131        $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} = $email;
2132        $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_msg->{date};
2133
2134        my $tree = $log_msg->{tree};
2135        if (!defined $tree) {
2136                $tree = $self->tmp_index_do(sub {
2137                                            command_oneline('write-tree') });
2138        }
2139        die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
2140
2141        my @exec = ('git-commit-tree', $tree);
2142        foreach ($self->get_commit_parents($log_msg, @parents)) {
2143                push @exec, '-p', $_;
2144        }
2145        defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
2146                                                                   or croak $!;
2147        print $msg_fh $log_msg->{log} or croak $!;
2148        print $msg_fh "\ngit-svn-id: $self->{ra}->{url}\@$log_msg->{revision}",
2149                      " ", $self->ra->uuid,"\n" or croak $!;
2150        $msg_fh->flush == 0 or croak $!;
2151        close $msg_fh or croak $!;
2152        chomp(my $commit = do { local $/; <$out_fh> });
2153        close $out_fh or croak $!;
2154        waitpid $pid, 0;
2155        croak $? if $?;
2156        if ($commit !~ /^$::sha1$/o) {
2157                die "Failed to commit, invalid sha1: $commit\n";
2158        }
2159
2160        command_noisy('update-ref',$self->refname, $commit);
2161        $self->rev_db_set($log_msg->{revision}, $commit);
2162
2163        $self->{last_rev} = $log_msg->{revision};
2164        $self->{last_commit} = $commit;
2165        print "r$log_msg->{revision} = $commit\n";
2166        return $commit;
2167}
2168
2169sub do_fetch {
2170        my ($self, $paths, $rev) = @_; #, $author, $date, $msg) = @_;
2171        my $ed = SVN::Git::Fetcher->new($self);
2172        my ($last_rev, @parents);
2173        if ($self->{last_commit}) {
2174                $last_rev = $self->{last_rev};
2175                $ed->{c} = $self->{last_commit};
2176                @parents = ($self->{last_commit});
2177        } else {
2178                $last_rev = $rev;
2179        }
2180        unless ($self->ra->do_update($last_rev, $rev, '', 1, $ed)) {
2181                die "SVN connection failed somewhere...\n";
2182        }
2183        $self->make_log_entry($rev, \@parents, $ed);
2184}
2185
2186sub write_untracked {
2187        my ($self, $rev, $fh, $untracked) = @_;
2188        my $h;
2189        print $fh "r$rev\n" or croak $!;
2190        $h = $untracked->{empty};
2191        foreach (sort keys %$h) {
2192                my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
2193                print $fh "  $act: ", uri_encode($_), "\n" or croak $!;
2194                warn "W: $act: $_\n";
2195        }
2196        foreach my $t (qw/dir_prop file_prop/) {
2197                $h = $untracked->{$t} or next;
2198                foreach my $path (sort keys %$h) {
2199                        my $ppath = $path eq '' ? '.' : $path;
2200                        foreach my $prop (sort keys %{$h->{$path}}) {
2201                                next if $SKIP{$prop};
2202                                my $v = $h->{$path}->{$prop};
2203                                if (defined $v) {
2204                                        print $fh "  +$t: ",
2205                                                  uri_encode($ppath), ' ',
2206                                                  uri_encode($prop), ' ',
2207                                                  uri_encode($v), "\n"
2208                                                  or croak $!;
2209                                } else {
2210                                        print $fh "  -$t: ",
2211                                                  uri_encode($ppath), ' ',
2212                                                  uri_encode($prop), "\n"
2213                                                  or croak $!;
2214                                }
2215                        }
2216                }
2217        }
2218        foreach my $t (qw/absent_file absent_directory/) {
2219                $h = $untracked->{$t} or next;
2220                foreach my $parent (sort keys %$h) {
2221                        foreach my $path (sort @{$h->{$parent}}) {
2222                                print $fh "  $t: ",
2223                                      uri_encode("$parent/$path"), "\n"
2224                                      or croak $!;
2225                                warn "W: $t: $parent/$path ",
2226                                     "Insufficient permissions?\n";
2227                        }
2228                }
2229        }
2230}
2231
2232sub make_log_entry {
2233        my ($self, $rev, $parents, $untracked) = @_;
2234        my $rp = $self->ra->rev_proplist($rev);
2235        my %log_entry = ( parents => $parents || [], revision => $rev,
2236                          revprops => $rp, log => '');
2237        open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
2238        $self->write_untracked($rev, $un, $untracked);
2239        foreach (sort keys %$rp) {
2240                my $v = $rp->{$_};
2241                if (/^svn:(author|date|log)$/) {
2242                        $log_entry{$1} = $v;
2243                } else {
2244                        print $un "  rev_prop: ", uri_encode($_), ' ',
2245                                  uri_encode($v), "\n";
2246                }
2247        }
2248        close $un or croak $!;
2249        $log_entry{date} = parse_svn_date($log_entry{date});
2250        $log_entry{author} = check_author($log_entry{author});
2251        $log_entry{log} .= "\n";
2252        \%log_entry;
2253}
2254
2255sub fetch {
2256        my ($self, @parents) = @_;
2257        my ($last_rev, $last_commit) = $self->last_rev_commit;
2258        my ($base, $head) = $self->parse_revision($last_rev);
2259        return if ($base > $head);
2260        if (defined $last_commit) {
2261                $self->assert_index_clean($last_commit);
2262        }
2263        my $inc = 1000;
2264        my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
2265        my $err_handler = $SVN::Error::handler;
2266        $SVN::Error::handler = \&skip_unknown_revs;
2267        while (1) {
2268                my @revs;
2269                $self->ra->get_log([''], $min, $max, 0, 1, 1, sub {
2270                        my ($paths, $rev, $author, $date, $msg) = @_;
2271                        push @revs, $rev });
2272                foreach (@revs) {
2273                        my $log_entry = $self->do_fetch(undef, $_);
2274                        $self->do_git_commit($log_entry, @parents);
2275                }
2276                last if $max >= $head;
2277                $min = $max + 1;
2278                $max += $inc;
2279                $max = $head if ($max > $head);
2280        }
2281        $SVN::Error::handler = $err_handler;
2282}
2283
2284sub set_tree_cb {
2285        my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
2286        # TODO: enable and test optimized commits:
2287        if (0 && $rev == ($self->{last_rev} + 1)) {
2288                $log_entry->{revision} = $rev;
2289                $log_entry->{author} = $author;
2290                $self->do_git_commit($log_entry, "$rev=$tree");
2291        } else {
2292                $self->fetch("$rev=$tree");
2293        }
2294}
2295
2296sub set_tree {
2297        my ($self, $tree) = (shift, shift);
2298        my $log_entry = get_commit_entry($tree);
2299        unless ($self->{last_rev}) {
2300                fatal("Must have an existing revision to commit\n");
2301        }
2302        my $pool = SVN::Pool->new;
2303        my $ed = SVN::Git::Editor->new({ r => $self->{last_rev},
2304                                         ra => $self->ra->dup,
2305                                         c => $tree,
2306                                         svn_path => $self->ra->{svn_path}
2307                                       },
2308                                       $self->ra->get_commit_editor(
2309                                         $log_entry->{log}, sub {
2310                                           $self->set_tree_cb($log_entry,
2311                                                              $tree, @_);
2312                                       }),
2313                                       $pool);
2314        my $mods = $ed->apply_diff($self->{last_commit}, $tree);
2315        if (@$mods == 0) {
2316                print "No changes\nr$self->{last_rev} = $tree\n";
2317        }
2318        $pool->clear;
2319}
2320
2321sub skip_unknown_revs {
2322        my ($err) = @_;
2323        my $errno = $err->apr_err();
2324        # Maybe the branch we're tracking didn't
2325        # exist when the repo started, so it's
2326        # not an error if it doesn't, just continue
2327        #
2328        # Wonderfully consistent library, eh?
2329        # 160013 - svn:// and file://
2330        # 175002 - http(s)://
2331        # 175007 - http(s):// (this repo required authorization, too...)
2332        #   More codes may be discovered later...
2333        if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
2334                return;
2335        }
2336        croak "Error from SVN, ($errno): ", $err->expanded_message,"\n";
2337}
2338
2339# rev_db:
2340# Tie::File seems to be prone to offset errors if revisions get sparse,
2341# it's not that fast, either.  Tie::File is also not in Perl 5.6.  So
2342# one of my favorite modules is out :<  Next up would be one of the DBM
2343# modules, but I'm not sure which is most portable...  So I'll just
2344# go with something that's plain-text, but still capable of
2345# being randomly accessed.  So here's my ultra-simple fixed-width
2346# database.  All records are 40 characters + "\n", so it's easy to seek
2347# to a revision: (41 * rev) is the byte offset.
2348# A record of 40 0s denotes an empty revision.
2349# And yes, it's still pretty fast (faster than Tie::File).
2350
2351sub rev_db_set {
2352        my ($self, $rev, $commit) = @_;
2353        length $commit == 40 or croak "arg3 must be a full SHA1 hexsum\n";
2354        open my $fh, '+<', $self->{db_path} or croak $!;
2355        my $offset = $rev * 41;
2356        # assume that append is the common case:
2357        seek $fh, 0, 2 or croak $!;
2358        my $pos = tell $fh;
2359        if ($pos < $offset) {
2360                print $fh (('0' x 40),"\n") x (($offset - $pos) / 41)
2361                  or croak $!;
2362        }
2363        seek $fh, $offset, 0 or croak $!;
2364        print $fh $commit,"\n" or croak $!;
2365        close $fh or croak $!;
2366}
2367
2368sub rev_db_get {
2369        my ($self, $rev) = @_;
2370        my $ret;
2371        my $offset = $rev * 41;
2372        open my $fh, '<', $self->{db_path} or croak $!;
2373        if (seek $fh, $offset, 0) {
2374                $ret = readline $fh;
2375                if (defined $ret) {
2376                        chomp $ret;
2377                        $ret = undef if ($ret =~ /^0{40}$/);
2378                }
2379        }
2380        close $fh or croak $!;
2381        $ret;
2382}
2383
2384sub _new {
2385        my ($class, $id) = @_;
2386        $id ||= $Git::SVN::default;
2387        my $dir = "$ENV{GIT_DIR}/svn/$id";
2388        bless { id => $id, dir => $dir, index => "$dir/index",
2389                db_path => "$dir/.rev_db" }, $class;
2390}
2391
2392
2393package Git::SVN::Prompt;
2394use strict;
2395use warnings;
2396require SVN::Core;
2397use vars qw/$_no_auth_cache $_username/;
2398
2399sub simple {
2400        my ($cred, $realm, $default_username, $may_save, $pool) = @_;
2401        $may_save = undef if $_no_auth_cache;
2402        $default_username = $_username if defined $_username;
2403        if (defined $default_username && length $default_username) {
2404                if (defined $realm && length $realm) {
2405                        print STDERR "Authentication realm: $realm\n";
2406                        STDERR->flush;
2407                }
2408                $cred->username($default_username);
2409        } else {
2410                username($cred, $realm, $may_save, $pool);
2411        }
2412        $cred->password(_read_password("Password for '" .
2413                                       $cred->username . "': ", $realm));
2414        $cred->may_save($may_save);
2415        $SVN::_Core::SVN_NO_ERROR;
2416}
2417
2418sub ssl_server_trust {
2419        my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
2420        $may_save = undef if $_no_auth_cache;
2421        print STDERR "Error validating server certificate for '$realm':\n";
2422        if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
2423                print STDERR " - The certificate is not issued by a trusted ",
2424                      "authority. Use the\n",
2425                      "   fingerprint to validate the certificate manually!\n";
2426        }
2427        if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
2428                print STDERR " - The certificate hostname does not match.\n";
2429        }
2430        if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
2431                print STDERR " - The certificate is not yet valid.\n";
2432        }
2433        if ($failures & $SVN::Auth::SSL::EXPIRED) {
2434                print STDERR " - The certificate has expired.\n";
2435        }
2436        if ($failures & $SVN::Auth::SSL::OTHER) {
2437                print STDERR " - The certificate has an unknown error.\n";
2438        }
2439        printf STDERR
2440                "Certificate information:\n".
2441                " - Hostname: %s\n".
2442                " - Valid: from %s until %s\n".
2443                " - Issuer: %s\n".
2444                " - Fingerprint: %s\n",
2445                map $cert_info->$_, qw(hostname valid_from valid_until
2446                                       issuer_dname fingerprint);
2447        my $choice;
2448prompt:
2449        print STDERR $may_save ?
2450              "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
2451              "(R)eject or accept (t)emporarily? ";
2452        STDERR->flush;
2453        $choice = lc(substr(<STDIN> || 'R', 0, 1));
2454        if ($choice =~ /^t$/i) {
2455                $cred->may_save(undef);
2456        } elsif ($choice =~ /^r$/i) {
2457                return -1;
2458        } elsif ($may_save && $choice =~ /^p$/i) {
2459                $cred->may_save($may_save);
2460        } else {
2461                goto prompt;
2462        }
2463        $cred->accepted_failures($failures);
2464        $SVN::_Core::SVN_NO_ERROR;
2465}
2466
2467sub ssl_client_cert {
2468        my ($cred, $realm, $may_save, $pool) = @_;
2469        $may_save = undef if $_no_auth_cache;
2470        print STDERR "Client certificate filename: ";
2471        STDERR->flush;
2472        chomp(my $filename = <STDIN>);
2473        $cred->cert_file($filename);
2474        $cred->may_save($may_save);
2475        $SVN::_Core::SVN_NO_ERROR;
2476}
2477
2478sub ssl_client_cert_pw {
2479        my ($cred, $realm, $may_save, $pool) = @_;
2480        $may_save = undef if $_no_auth_cache;
2481        $cred->password(_read_password("Password: ", $realm));
2482        $cred->may_save($may_save);
2483        $SVN::_Core::SVN_NO_ERROR;
2484}
2485
2486sub username {
2487        my ($cred, $realm, $may_save, $pool) = @_;
2488        $may_save = undef if $_no_auth_cache;
2489        if (defined $realm && length $realm) {
2490                print STDERR "Authentication realm: $realm\n";
2491        }
2492        my $username;
2493        if (defined $_username) {
2494                $username = $_username;
2495        } else {
2496                print STDERR "Username: ";
2497                STDERR->flush;
2498                chomp($username = <STDIN>);
2499        }
2500        $cred->username($username);
2501        $cred->may_save($may_save);
2502        $SVN::_Core::SVN_NO_ERROR;
2503}
2504
2505sub _read_password {
2506        my ($prompt, $realm) = @_;
2507        print STDERR $prompt;
2508        STDERR->flush;
2509        require Term::ReadKey;
2510        Term::ReadKey::ReadMode('noecho');
2511        my $password = '';
2512        while (defined(my $key = Term::ReadKey::ReadKey(0))) {
2513                last if $key =~ /[\012\015]/; # \n\r
2514                $password .= $key;
2515        }
2516        Term::ReadKey::ReadMode('restore');
2517        print STDERR "\n";
2518        STDERR->flush;
2519        $password;
2520}
2521
2522package main;
2523
2524sub uri_encode {
2525        my ($f) = @_;
2526        $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
2527        $f
2528}
2529
2530sub uri_decode {
2531        my ($f) = @_;
2532        $f =~ tr/+/ /;
2533        $f =~ s/%([A-F0-9]{2})/chr hex($1)/ge;
2534        $f
2535}
2536
2537sub libsvn_log_entry {
2538        my ($rev, $author, $date, $msg, $parents, $untracked) = @_;
2539        my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2540                                         (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x)
2541                                or die "Unable to parse date: $date\n";
2542        if (defined $author && length $author > 0 &&
2543            defined $_authors && ! defined $users{$author}) {
2544                die "Author: $author not defined in $_authors file\n";
2545        }
2546        $msg = '' if ($rev == 0 && !defined $msg);
2547
2548        open my $un, '>>', "$GIT_SVN_DIR/unhandled.log" or croak $!;
2549        my $h;
2550        print $un "r$rev\n" or croak $!;
2551        $h = $untracked->{empty};
2552        foreach (sort keys %$h) {
2553                my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
2554                print $un "  $act: ", uri_encode($_), "\n" or croak $!;
2555                warn "W: $act: $_\n";
2556        }
2557        foreach my $t (qw/dir_prop file_prop/) {
2558                $h = $untracked->{$t} or next;
2559                foreach my $path (sort keys %$h) {
2560                        my $ppath = $path eq '' ? '.' : $path;
2561                        foreach my $prop (sort keys %{$h->{$path}}) {
2562                                next if $SKIP{$prop};
2563                                my $v = $h->{$path}->{$prop};
2564                                if (defined $v) {
2565                                        print $un "  +$t: ",
2566                                                  uri_encode($ppath), ' ',
2567                                                  uri_encode($prop), ' ',
2568                                                  uri_encode($v), "\n"
2569                                                  or croak $!;
2570                                } else {
2571                                        print $un "  -$t: ",
2572                                                  uri_encode($ppath), ' ',
2573                                                  uri_encode($prop), "\n"
2574                                                  or croak $!;
2575                                }
2576                        }
2577                }
2578        }
2579        foreach my $t (qw/absent_file absent_directory/) {
2580                $h = $untracked->{$t} or next;
2581                foreach my $parent (sort keys %$h) {
2582                        foreach my $path (sort @{$h->{$parent}}) {
2583                                print $un "  $t: ",
2584                                      uri_encode("$parent/$path"), "\n"
2585                                      or croak $!;
2586                                warn "W: $t: $parent/$path ",
2587                                     "Insufficient permissions?\n";
2588                        }
2589                }
2590        }
2591
2592        # revprops (make this optional? it's an extra network trip...)
2593        my $rp = $SVN->rev_proplist($rev);
2594        foreach (sort keys %$rp) {
2595                next if /^svn:(?:author|date|log)$/;
2596                print $un "  rev_prop: ", uri_encode($_), ' ',
2597                          uri_encode($rp->{$_}), "\n";
2598        }
2599        close $un or croak $!;
2600
2601        { revision => $rev, date => "+0000 $Y-$m-$d $H:$M:$S",
2602          author => $author, msg => $msg."\n", parents => $parents || [],
2603          revprops => $rp }
2604}
2605
2606sub libsvn_fetch {
2607        my ($last_commit, $paths, $rev, $author, $date, $msg) = @_;
2608        my $ed = SVN::Git::Fetcher->new({ c => $last_commit, q => $_q });
2609        my (undef, $last_rev, undef) = cmt_metadata($last_commit);
2610        unless ($SVN->gs_do_update($last_rev, $rev, '', 1, $ed)) {
2611                die "SVN connection failed somewhere...\n";
2612        }
2613        libsvn_log_entry($rev, $author, $date, $msg, [$last_commit], $ed);
2614}
2615
2616sub svn_grab_base_rev {
2617        my $c = eval { command_oneline([qw/rev-parse --verify/,
2618                                        "refs/remotes/$GIT_SVN^0"],
2619                                        { STDERR => 0 }) };
2620        if (defined $c && length $c) {
2621                my ($url, $rev, $uuid) = cmt_metadata($c);
2622                return ($rev, $c) if defined $rev;
2623        }
2624        if ($_no_metadata) {
2625                my $offset = -41; # from tail
2626                my $rl;
2627                open my $fh, '<', $REVDB or
2628                        die "--no-metadata specified and $REVDB not readable\n";
2629                seek $fh, $offset, 2;
2630                $rl = readline $fh;
2631                defined $rl or return (undef, undef);
2632                chomp $rl;
2633                while ($c ne $rl && tell $fh != 0) {
2634                        $offset -= 41;
2635                        seek $fh, $offset, 2;
2636                        $rl = readline $fh;
2637                        defined $rl or return (undef, undef);
2638                        chomp $rl;
2639                }
2640                my $rev = tell $fh;
2641                croak $! if ($rev < -1);
2642                $rev =  ($rev - 41) / 41;
2643                close $fh or croak $!;
2644                return ($rev, $c);
2645        }
2646        return (undef, undef);
2647}
2648
2649sub libsvn_parse_revision {
2650        my $base = shift;
2651        my $head = $SVN->get_latest_revnum();
2652        if (!defined $_revision || $_revision eq 'BASE:HEAD') {
2653                return ($base + 1, $head) if (defined $base);
2654                return (0, $head);
2655        }
2656        return ($1, $2) if ($_revision =~ /^(\d+):(\d+)$/);
2657        return ($_revision, $_revision) if ($_revision =~ /^\d+$/);
2658        if ($_revision =~ /^BASE:(\d+)$/) {
2659                return ($base + 1, $1) if (defined $base);
2660                return (0, $head);
2661        }
2662        return ($1, $head) if ($_revision =~ /^(\d+):HEAD$/);
2663        die "revision argument: $_revision not understood by git-svn\n",
2664                "Try using the command-line svn client instead\n";
2665}
2666
2667sub libsvn_traverse_ignore {
2668        my ($fh, $path, $r) = @_;
2669        $path =~ s#^/+##g;
2670        my ($dirent, undef, $props) = $SVN->get_dir($path, $r);
2671        my $p = $path;
2672        $p =~ s#^\Q$SVN->{svn_path}\E/##;
2673        print $fh length $p ? "\n# $p\n" : "\n# /\n";
2674        if (my $s = $props->{'svn:ignore'}) {
2675                $s =~ s/[\r\n]+/\n/g;
2676                chomp $s;
2677                if (length $p == 0) {
2678                        $s =~ s#\n#\n/$p#g;
2679                        print $fh "/$s\n";
2680                } else {
2681                        $s =~ s#\n#\n/$p/#g;
2682                        print $fh "/$p/$s\n";
2683                }
2684        }
2685        foreach (sort keys %$dirent) {
2686                next if $dirent->{$_}->kind != $SVN::Node::dir;
2687                libsvn_traverse_ignore($fh, "$path/$_", $r);
2688        }
2689}
2690
2691sub revisions_eq {
2692        my ($path, $r0, $r1) = @_;
2693        return 1 if $r0 == $r1;
2694        my $nr = 0;
2695        # should be OK to use Pool here (r1 - r0) should be small
2696        $SVN->get_log([$path], $r0, $r1, 0, 0, 1, sub {$nr++});
2697        return 0 if ($nr > 1);
2698        return 1;
2699}
2700
2701sub libsvn_find_parent_branch {
2702        my ($paths, $rev, $author, $date, $msg) = @_;
2703        my $svn_path = '/'.$SVN->{svn_path};
2704
2705        # look for a parent from another branch:
2706        my $i = $paths->{$svn_path} or return;
2707        my $branch_from = $i->copyfrom_path or return;
2708        my $r = $i->copyfrom_rev;
2709        print STDERR  "Found possible branch point: ",
2710                                "$branch_from => $svn_path, $r\n";
2711        $branch_from =~ s#^/##;
2712        my $l_map = {};
2713        read_url_paths_all($l_map, '', "$GIT_DIR/svn");
2714        my $url = $SVN->{repos_root};
2715        defined $l_map->{$url} or return;
2716        my $id = $l_map->{$url}->{$branch_from};
2717        if (!defined $id && $_follow_parent) {
2718                print STDERR "Following parent: $branch_from\@$r\n";
2719                # auto create a new branch and follow it
2720                $id = basename($branch_from);
2721                $id .= '@'.$r if -r "$GIT_DIR/svn/$id";
2722                while (-r "$GIT_DIR/svn/$id") {
2723                        # just grow a tail if we're not unique enough :x
2724                        $id .= '-';
2725                }
2726        }
2727        return unless defined $id;
2728
2729        my ($r0, $parent) = find_rev_before($r,$id,1);
2730        if ($_follow_parent && (!defined $r0 || !defined $parent)) {
2731                defined(my $pid = fork) or croak $!;
2732                if (!$pid) {
2733                        $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
2734                        init_vars();
2735                        $SVN_URL = "$url/$branch_from";
2736                        $SVN = undef;
2737                        setup_git_svn();
2738                        # we can't assume SVN_URL exists at r+1:
2739                        $_revision = "0:$r";
2740                        fetch_lib();
2741                        exit 0;
2742                }
2743                waitpid $pid, 0;
2744                croak $? if $?;
2745                ($r0, $parent) = find_rev_before($r,$id,1);
2746        }
2747        return unless (defined $r0 && defined $parent);
2748        if (revisions_eq($branch_from, $r0, $r)) {
2749                unlink $GIT_SVN_INDEX;
2750                print STDERR "Found branch parent: ($GIT_SVN) $parent\n";
2751                command_noisy('read-tree', $parent);
2752                unless ($SVN->can_do_switch) {
2753                        return _libsvn_new_tree($paths, $rev, $author, $date,
2754                                                $msg, [$parent]);
2755                }
2756                # do_switch works with svn/trunk >= r22312, but that is not
2757                # included with SVN 1.4.2 (the latest version at the moment),
2758                # so we can't rely on it.
2759                my $ra = Git::SVN::Ra->new("$url/$branch_from");
2760                my $ed = SVN::Git::Fetcher->new({c => $parent, q => $_q });
2761                $ra->gs_do_switch($r0, $rev, '', 1, $SVN->{url}, $ed) or
2762                                   die "SVN connection failed somewhere...\n";
2763                return libsvn_log_entry($rev, $author, $date, $msg, [$parent]);
2764        }
2765        print STDERR "Nope, branch point not imported or unknown\n";
2766        return undef;
2767}
2768
2769sub libsvn_new_tree {
2770        if (my $log_entry = libsvn_find_parent_branch(@_)) {
2771                return $log_entry;
2772        }
2773        my ($paths, $rev, $author, $date, $msg) = @_; # $pool is last
2774        _libsvn_new_tree($paths, $rev, $author, $date, $msg, []);
2775}
2776
2777sub _libsvn_new_tree {
2778        my ($paths, $rev, $author, $date, $msg, $parents) = @_;
2779        my $ed = SVN::Git::Fetcher->new({q => $_q});
2780        unless ($SVN->gs_do_update($rev, $rev, '', 1, $ed)) {
2781                die "SVN connection failed somewhere...\n";
2782        }
2783        libsvn_log_entry($rev, $author, $date, $msg, $parents, $ed);
2784}
2785
2786sub find_graft_path_commit {
2787        my ($tree_paths, $p1, $r1) = @_;
2788        foreach my $x (keys %$tree_paths) {
2789                next unless ($p1 =~ /^\Q$x\E/);
2790                my $i = $tree_paths->{$x};
2791                my ($r0, $parent) = find_rev_before($r1,$i,1);
2792                return $parent if (defined $r0 && $r0 == $r1);
2793                print STDERR "r$r1 of $i not imported\n";
2794                next;
2795        }
2796        return undef;
2797}
2798
2799sub find_graft_path_parents {
2800        my ($grafts, $tree_paths, $c, $p0, $r0) = @_;
2801        foreach my $x (keys %$tree_paths) {
2802                next unless ($p0 =~ /^\Q$x\E/);
2803                my $i = $tree_paths->{$x};
2804                my ($r, $parent) = find_rev_before($r0, $i, 1);
2805                if (defined $r && defined $parent && revisions_eq($x,$r,$r0)) {
2806                        my ($url_b, undef, $uuid_b) = cmt_metadata($c);
2807                        my ($url_a, undef, $uuid_a) = cmt_metadata($parent);
2808                        next if ($url_a && $url_b && $url_a eq $url_b &&
2809                                                        $uuid_b eq $uuid_a);
2810                        $grafts->{$c}->{$parent} = 1;
2811                }
2812        }
2813}
2814
2815sub libsvn_graft_file_copies {
2816        my ($grafts, $tree_paths, $path, $paths, $rev) = @_;
2817        foreach (keys %$paths) {
2818                my $i = $paths->{$_};
2819                my ($m, $p0, $r0) = ($i->action, $i->copyfrom_path,
2820                                        $i->copyfrom_rev);
2821                next unless (defined $p0 && defined $r0);
2822
2823                my $p1 = $_;
2824                $p1 =~ s#^/##;
2825                $p0 =~ s#^/##;
2826                my $c = find_graft_path_commit($tree_paths, $p1, $rev);
2827                next unless $c;
2828                find_graft_path_parents($grafts, $tree_paths, $c, $p0, $r0);
2829        }
2830}
2831
2832sub set_index {
2833        my $old = $ENV{GIT_INDEX_FILE};
2834        $ENV{GIT_INDEX_FILE} = shift;
2835        return $old;
2836}
2837
2838sub restore_index {
2839        my ($old) = @_;
2840        if (defined $old) {
2841                $ENV{GIT_INDEX_FILE} = $old;
2842        } else {
2843                delete $ENV{GIT_INDEX_FILE};
2844        }
2845}
2846
2847sub libsvn_commit_cb {
2848        my ($rev, $date, $committer, $c, $msg, $r_last, $cmt_last) = @_;
2849        if ($_optimize_commits && $rev == ($r_last + 1)) {
2850                my $log = libsvn_log_entry($rev,$committer,$date,$msg);
2851                $log->{tree} = get_tree_from_treeish($c);
2852                my $cmt = git_commit($log, $cmt_last, $c);
2853                my @diff = command('diff-tree', $cmt, $c);
2854                if (@diff) {
2855                        print STDERR "Trees differ: $cmt $c\n",
2856                                        join('',@diff),"\n";
2857                        exit 1;
2858                }
2859        } else {
2860                fetch("$rev=$c");
2861        }
2862}
2863
2864sub libsvn_skip_unknown_revs {
2865        my $err = shift;
2866        my $errno = $err->apr_err();
2867        # Maybe the branch we're tracking didn't
2868        # exist when the repo started, so it's
2869        # not an error if it doesn't, just continue
2870        #
2871        # Wonderfully consistent library, eh?
2872        # 160013 - svn:// and file://
2873        # 175002 - http(s)://
2874        # 175007 - http(s):// (this repo required authorization, too...)
2875        #   More codes may be discovered later...
2876        if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
2877                return;
2878        }
2879        croak "Error from SVN, ($errno): ", $err->expanded_message,"\n";
2880};
2881
2882# Tie::File seems to be prone to offset errors if revisions get sparse,
2883# it's not that fast, either.  Tie::File is also not in Perl 5.6.  So
2884# one of my favorite modules is out :<  Next up would be one of the DBM
2885# modules, but I'm not sure which is most portable...  So I'll just
2886# go with something that's plain-text, but still capable of
2887# being randomly accessed.  So here's my ultra-simple fixed-width
2888# database.  All records are 40 characters + "\n", so it's easy to seek
2889# to a revision: (41 * rev) is the byte offset.
2890# A record of 40 0s denotes an empty revision.
2891# And yes, it's still pretty fast (faster than Tie::File).
2892sub revdb_set {
2893        my ($file, $rev, $commit) = @_;
2894        length $commit == 40 or croak "arg3 must be a full SHA1 hexsum\n";
2895        open my $fh, '+<', $file or croak $!;
2896        my $offset = $rev * 41;
2897        # assume that append is the common case:
2898        seek $fh, 0, 2 or croak $!;
2899        my $pos = tell $fh;
2900        if ($pos < $offset) {
2901                print $fh (('0' x 40),"\n") x (($offset - $pos) / 41);
2902        }
2903        seek $fh, $offset, 0 or croak $!;
2904        print $fh $commit,"\n";
2905        close $fh or croak $!;
2906}
2907
2908sub revdb_get {
2909        my ($file, $rev) = @_;
2910        my $ret;
2911        my $offset = $rev * 41;
2912        open my $fh, '<', $file or croak $!;
2913        seek $fh, $offset, 0;
2914        if (tell $fh == $offset) {
2915                $ret = readline $fh;
2916                if (defined $ret) {
2917                        chomp $ret;
2918                        $ret = undef if ($ret =~ /^0{40}$/);
2919                }
2920        }
2921        close $fh or croak $!;
2922        return $ret;
2923}
2924
2925sub copy_remote_ref {
2926        my $origin = $_cp_remote ? $_cp_remote : 'origin';
2927        my $ref = "refs/remotes/$GIT_SVN";
2928        if (command('ls-remote', $origin, $ref)) {
2929                command_noisy('fetch', $origin, "$ref:$ref");
2930        } elsif ($_cp_remote && !$_upgrade) {
2931                die "Unable to find remote reference: ",
2932                                "refs/remotes/$GIT_SVN on $origin\n";
2933        }
2934}
2935
2936{
2937        my $kill_stupid_warnings = $SVN::Node::none.$SVN::Node::file.
2938                                $SVN::Node::dir.$SVN::Node::unknown.
2939                                $SVN::Node::none.$SVN::Node::file.
2940                                $SVN::Node::dir.$SVN::Node::unknown.
2941                                $SVN::Auth::SSL::CNMISMATCH.
2942                                $SVN::Auth::SSL::NOTYETVALID.
2943                                $SVN::Auth::SSL::EXPIRED.
2944                                $SVN::Auth::SSL::UNKNOWNCA.
2945                                $SVN::Auth::SSL::OTHER;
2946}
2947
2948package SVN::Git::Fetcher;
2949use vars qw/@ISA/;
2950use strict;
2951use warnings;
2952use Carp qw/croak/;
2953use IO::File qw//;
2954
2955# file baton members: path, mode_a, mode_b, pool, fh, blob, base
2956sub new {
2957        my ($class, $git_svn) = @_;
2958        my $self = SVN::Delta::Editor->new;
2959        bless $self, $class;
2960        $self->{c} = $git_svn->{c} if exists $git_svn->{c};
2961        $self->{q} = $git_svn->{q};
2962        $self->{empty} = {};
2963        $self->{dir_prop} = {};
2964        $self->{file_prop} = {};
2965        $self->{absent_dir} = {};
2966        $self->{absent_file} = {};
2967        ($self->{gui}, $self->{ctx}) = command_input_pipe(
2968                                             qw/update-index -z --index-info/);
2969        require Digest::MD5;
2970        $self;
2971}
2972
2973sub open_root {
2974        { path => '' };
2975}
2976
2977sub open_directory {
2978        my ($self, $path, $pb, $rev) = @_;
2979        { path => $path };
2980}
2981
2982sub delete_entry {
2983        my ($self, $path, $rev, $pb) = @_;
2984        my $gui = $self->{gui};
2985
2986        # remove entire directories.
2987        if (command('ls-tree', $self->{c}, '--', $path) =~ /^040000 tree/) {
2988                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
2989                                                     -r --name-only -z/,
2990                                                     $self->{c}, '--', $path);
2991                local $/ = "\0";
2992                while (<$ls>) {
2993                        print $gui '0 ',0 x 40,"\t",$_ or croak $!;
2994                        print "\tD\t$_\n" unless $self->{q};
2995                }
2996                print "\tD\t$path/\n" unless $self->{q};
2997                command_close_pipe($ls, $ctx);
2998                $self->{empty}->{$path} = 0
2999        } else {
3000                print $gui '0 ',0 x 40,"\t",$path,"\0" or croak $!;
3001                print "\tD\t$path\n" unless $self->{q};
3002        }
3003        undef;
3004}
3005
3006sub open_file {
3007        my ($self, $path, $pb, $rev) = @_;
3008        my ($mode, $blob) = (command('ls-tree', $self->{c}, '--',$path)
3009                             =~ /^(\d{6}) blob ([a-f\d]{40})\t/);
3010        unless (defined $mode && defined $blob) {
3011                die "$path was not found in commit $self->{c} (r$rev)\n";
3012        }
3013        { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
3014          pool => SVN::Pool->new, action => 'M' };
3015}
3016
3017sub add_file {
3018        my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
3019        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3020        delete $self->{empty}->{$dir};
3021        { path => $path, mode_a => 100644, mode_b => 100644,
3022          pool => SVN::Pool->new, action => 'A' };
3023}
3024
3025sub add_directory {
3026        my ($self, $path, $cp_path, $cp_rev) = @_;
3027        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3028        delete $self->{empty}->{$dir};
3029        $self->{empty}->{$path} = 1;
3030        { path => $path };
3031}
3032
3033sub change_dir_prop {
3034        my ($self, $db, $prop, $value) = @_;
3035        $self->{dir_prop}->{$db->{path}} ||= {};
3036        $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
3037        undef;
3038}
3039
3040sub absent_directory {
3041        my ($self, $path, $pb) = @_;
3042        $self->{absent_dir}->{$pb->{path}} ||= [];
3043        push @{$self->{absent_dir}->{$pb->{path}}}, $path;
3044        undef;
3045}
3046
3047sub absent_file {
3048        my ($self, $path, $pb) = @_;
3049        $self->{absent_file}->{$pb->{path}} ||= [];
3050        push @{$self->{absent_file}->{$pb->{path}}}, $path;
3051        undef;
3052}
3053
3054sub change_file_prop {
3055        my ($self, $fb, $prop, $value) = @_;
3056        if ($prop eq 'svn:executable') {
3057                if ($fb->{mode_b} != 120000) {
3058                        $fb->{mode_b} = defined $value ? 100755 : 100644;
3059                }
3060        } elsif ($prop eq 'svn:special') {
3061                $fb->{mode_b} = defined $value ? 120000 : 100644;
3062        } else {
3063                $self->{file_prop}->{$fb->{path}} ||= {};
3064                $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
3065        }
3066        undef;
3067}
3068
3069sub apply_textdelta {
3070        my ($self, $fb, $exp) = @_;
3071        my $fh = IO::File->new_tmpfile;
3072        $fh->autoflush(1);
3073        # $fh gets auto-closed() by SVN::TxDelta::apply(),
3074        # (but $base does not,) so dup() it for reading in close_file
3075        open my $dup, '<&', $fh or croak $!;
3076        my $base = IO::File->new_tmpfile;
3077        $base->autoflush(1);
3078        if ($fb->{blob}) {
3079                defined (my $pid = fork) or croak $!;
3080                if (!$pid) {
3081                        open STDOUT, '>&', $base or croak $!;
3082                        print STDOUT 'link ' if ($fb->{mode_a} == 120000);
3083                        exec qw/git-cat-file blob/, $fb->{blob} or croak $!;
3084                }
3085                waitpid $pid, 0;
3086                croak $? if $?;
3087
3088                if (defined $exp) {
3089                        seek $base, 0, 0 or croak $!;
3090                        my $md5 = Digest::MD5->new;
3091                        $md5->addfile($base);
3092                        my $got = $md5->hexdigest;
3093                        die "Checksum mismatch: $fb->{path} $fb->{blob}\n",
3094                            "expected: $exp\n",
3095                            "     got: $got\n" if ($got ne $exp);
3096                }
3097        }
3098        seek $base, 0, 0 or croak $!;
3099        $fb->{fh} = $dup;
3100        $fb->{base} = $base;
3101        [ SVN::TxDelta::apply($base, $fh, undef, $fb->{path}, $fb->{pool}) ];
3102}
3103
3104sub close_file {
3105        my ($self, $fb, $exp) = @_;
3106        my $hash;
3107        my $path = $fb->{path};
3108        if (my $fh = $fb->{fh}) {
3109                seek($fh, 0, 0) or croak $!;
3110                my $md5 = Digest::MD5->new;
3111                $md5->addfile($fh);
3112                my $got = $md5->hexdigest;
3113                die "Checksum mismatch: $path\n",
3114                    "expected: $exp\n    got: $got\n" if ($got ne $exp);
3115                seek($fh, 0, 0) or croak $!;
3116                if ($fb->{mode_b} == 120000) {
3117                        read($fh, my $buf, 5) == 5 or croak $!;
3118                        $buf eq 'link ' or die "$path has mode 120000",
3119                                               "but is not a link\n";
3120                }
3121                defined(my $pid = open my $out,'-|') or die "Can't fork: $!\n";
3122                if (!$pid) {
3123                        open STDIN, '<&', $fh or croak $!;
3124                        exec qw/git-hash-object -w --stdin/ or croak $!;
3125                }
3126                chomp($hash = do { local $/; <$out> });
3127                close $out or croak $!;
3128                close $fh or croak $!;
3129                $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
3130                close $fb->{base} or croak $!;
3131        } else {
3132                $hash = $fb->{blob} or die "no blob information\n";
3133        }
3134        $fb->{pool}->clear;
3135        my $gui = $self->{gui};
3136        print $gui "$fb->{mode_b} $hash\t$path\0" or croak $!;
3137        print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $self->{q};
3138        undef;
3139}
3140
3141sub abort_edit {
3142        my $self = shift;
3143        eval { command_close_pipe($self->{gui}, $self->{ctx}) };
3144        $self->SUPER::abort_edit(@_);
3145}
3146
3147sub close_edit {
3148        my $self = shift;
3149        command_close_pipe($self->{gui}, $self->{ctx});
3150        $self->{git_commit_ok} = 1;
3151        $self->SUPER::close_edit(@_);
3152}
3153
3154package SVN::Git::Editor;
3155use vars qw/@ISA/;
3156use strict;
3157use warnings;
3158use Carp qw/croak/;
3159use IO::File;
3160
3161sub new {
3162        my $class = shift;
3163        my $git_svn = shift;
3164        my $self = SVN::Delta::Editor->new(@_);
3165        bless $self, $class;
3166        foreach (qw/svn_path c r ra /) {
3167                die "$_ required!\n" unless (defined $git_svn->{$_});
3168                $self->{$_} = $git_svn->{$_};
3169        }
3170        $self->{pool} = SVN::Pool->new;
3171        $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
3172        $self->{rm} = { };
3173        require Digest::MD5;
3174        return $self;
3175}
3176
3177sub split_path {
3178        return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
3179}
3180
3181sub repo_path {
3182        (defined $_[1] && length $_[1]) ? $_[1] : ''
3183}
3184
3185sub url_path {
3186        my ($self, $path) = @_;
3187        $self->{ra}->{url} . '/' . $self->repo_path($path);
3188}
3189
3190sub rmdirs {
3191        my ($self, $q) = @_;
3192        my $rm = $self->{rm};
3193        delete $rm->{''}; # we never delete the url we're tracking
3194        return unless %$rm;
3195
3196        foreach (keys %$rm) {
3197                my @d = split m#/#, $_;
3198                my $c = shift @d;
3199                $rm->{$c} = 1;
3200                while (@d) {
3201                        $c .= '/' . shift @d;
3202                        $rm->{$c} = 1;
3203                }
3204        }
3205        delete $rm->{$self->{svn_path}};
3206        delete $rm->{''}; # we never delete the url we're tracking
3207        return unless %$rm;
3208
3209        my ($fh, $ctx) = command_output_pipe(
3210                                   qw/ls-tree --name-only -r -z/, $self->{c});
3211        local $/ = "\0";
3212        while (<$fh>) {
3213                chomp;
3214                my @dn = split m#/#, $_;
3215                while (pop @dn) {
3216                        delete $rm->{join '/', @dn};
3217                }
3218                unless (%$rm) {
3219                        close $fh;
3220                        return;
3221                }
3222        }
3223        command_close_pipe($fh, $ctx);
3224
3225        my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
3226        foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
3227                $self->close_directory($bat->{$d}, $p);
3228                my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
3229                print "\tD+\t$d/\n" unless $q;
3230                $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
3231                delete $bat->{$d};
3232        }
3233}
3234
3235sub open_or_add_dir {
3236        my ($self, $full_path, $baton) = @_;
3237        my $t = $self->{ra}->check_path($full_path, $self->{r});
3238        if ($t == $SVN::Node::none) {
3239                return $self->add_directory($full_path, $baton,
3240                                                undef, -1, $self->{pool});
3241        } elsif ($t == $SVN::Node::dir) {
3242                return $self->open_directory($full_path, $baton,
3243                                                $self->{r}, $self->{pool});
3244        }
3245        print STDERR "$full_path already exists in repository at ",
3246                "r$self->{r} and it is not a directory (",
3247                ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
3248        exit 1;
3249}
3250
3251sub ensure_path {
3252        my ($self, $path) = @_;
3253        my $bat = $self->{bat};
3254        $path = $self->repo_path($path);
3255        return $bat->{''} unless (length $path);
3256        my @p = split m#/+#, $path;
3257        my $c = shift @p;
3258        $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
3259        while (@p) {
3260                my $c0 = $c;
3261                $c .= '/' . shift @p;
3262                $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
3263        }
3264        return $bat->{$c};
3265}
3266
3267sub A {
3268        my ($self, $m, $q) = @_;
3269        my ($dir, $file) = split_path($m->{file_b});
3270        my $pbat = $self->ensure_path($dir);
3271        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3272                                        undef, -1);
3273        print "\tA\t$m->{file_b}\n" unless $q;
3274        $self->chg_file($fbat, $m);
3275        $self->close_file($fbat,undef,$self->{pool});
3276}
3277
3278sub C {
3279        my ($self, $m, $q) = @_;
3280        my ($dir, $file) = split_path($m->{file_b});
3281        my $pbat = $self->ensure_path($dir);
3282        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3283                                $self->url_path($m->{file_a}), $self->{r});
3284        print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $q;
3285        $self->chg_file($fbat, $m);
3286        $self->close_file($fbat,undef,$self->{pool});
3287}
3288
3289sub delete_entry {
3290        my ($self, $path, $pbat) = @_;
3291        my $rpath = $self->repo_path($path);
3292        my ($dir, $file) = split_path($rpath);
3293        $self->{rm}->{$dir} = 1;
3294        $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
3295}
3296
3297sub R {
3298        my ($self, $m, $q) = @_;
3299        my ($dir, $file) = split_path($m->{file_b});
3300        my $pbat = $self->ensure_path($dir);
3301        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3302                                $self->url_path($m->{file_a}), $self->{r});
3303        print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $q;
3304        $self->chg_file($fbat, $m);
3305        $self->close_file($fbat,undef,$self->{pool});
3306
3307        ($dir, $file) = split_path($m->{file_a});
3308        $pbat = $self->ensure_path($dir);
3309        $self->delete_entry($m->{file_a}, $pbat);
3310}
3311
3312sub M {
3313        my ($self, $m, $q) = @_;
3314        my ($dir, $file) = split_path($m->{file_b});
3315        my $pbat = $self->ensure_path($dir);
3316        my $fbat = $self->open_file($self->repo_path($m->{file_b}),
3317                                $pbat,$self->{r},$self->{pool});
3318        print "\t$m->{chg}\t$m->{file_b}\n" unless $q;
3319        $self->chg_file($fbat, $m);
3320        $self->close_file($fbat,undef,$self->{pool});
3321}
3322
3323sub T { shift->M(@_) }
3324
3325sub change_file_prop {
3326        my ($self, $fbat, $pname, $pval) = @_;
3327        $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
3328}
3329
3330sub chg_file {
3331        my ($self, $fbat, $m) = @_;
3332        if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
3333                $self->change_file_prop($fbat,'svn:executable','*');
3334        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
3335                $self->change_file_prop($fbat,'svn:executable',undef);
3336        }
3337        my $fh = IO::File->new_tmpfile or croak $!;
3338        if ($m->{mode_b} =~ /^120/) {
3339                print $fh 'link ' or croak $!;
3340                $self->change_file_prop($fbat,'svn:special','*');
3341        } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
3342                $self->change_file_prop($fbat,'svn:special',undef);
3343        }
3344        defined(my $pid = fork) or croak $!;
3345        if (!$pid) {
3346                open STDOUT, '>&', $fh or croak $!;
3347                exec qw/git-cat-file blob/, $m->{sha1_b} or croak $!;
3348        }
3349        waitpid $pid, 0;
3350        croak $? if $?;
3351        $fh->flush == 0 or croak $!;
3352        seek $fh, 0, 0 or croak $!;
3353
3354        my $md5 = Digest::MD5->new;
3355        $md5->addfile($fh) or croak $!;
3356        seek $fh, 0, 0 or croak $!;
3357
3358        my $exp = $md5->hexdigest;
3359        my $pool = SVN::Pool->new;
3360        my $atd = $self->apply_textdelta($fbat, undef, $pool);
3361        my $got = SVN::TxDelta::send_stream($fh, @$atd, $pool);
3362        die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
3363        $pool->clear;
3364
3365        close $fh or croak $!;
3366}
3367
3368sub D {
3369        my ($self, $m, $q) = @_;
3370        my ($dir, $file) = split_path($m->{file_b});
3371        my $pbat = $self->ensure_path($dir);
3372        print "\tD\t$m->{file_b}\n" unless $q;
3373        $self->delete_entry($m->{file_b}, $pbat);
3374}
3375
3376sub close_edit {
3377        my ($self) = @_;
3378        my ($p,$bat) = ($self->{pool}, $self->{bat});
3379        foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
3380                $self->close_directory($bat->{$_}, $p);
3381        }
3382        $self->SUPER::close_edit($p);
3383        $p->clear;
3384}
3385
3386sub abort_edit {
3387        my ($self) = @_;
3388        $self->SUPER::abort_edit($self->{pool});
3389        $self->{pool}->clear;
3390}
3391
3392package Git::SVN::Ra;
3393use vars qw/@ISA $config_dir/;
3394use strict;
3395use warnings;
3396my ($can_do_switch);
3397
3398BEGIN {
3399        # enforce temporary pool usage for some simple functions
3400        my $e;
3401        foreach (qw/get_latest_revnum rev_proplist get_file
3402                    check_path get_dir get_uuid get_repos_root/) {
3403                $e .= "sub $_ {
3404                        my \$self = shift;
3405                        my \$pool = SVN::Pool->new;
3406                        my \@ret = \$self->SUPER::$_(\@_,\$pool);
3407                        \$pool->clear;
3408                        wantarray ? \@ret : \$ret[0]; }\n";
3409        }
3410        eval $e;
3411}
3412
3413sub new {
3414        my ($class, $url) = @_;
3415        SVN::_Core::svn_config_ensure($config_dir, undef);
3416        my ($baton, $callbacks) = SVN::Core::auth_open_helper([
3417            SVN::Client::get_simple_provider(),
3418            SVN::Client::get_ssl_server_trust_file_provider(),
3419            SVN::Client::get_simple_prompt_provider(
3420              \&Git::SVN::Prompt::simple, 2),
3421            SVN::Client::get_ssl_client_cert_prompt_provider(
3422              \&Git::SVN::Prompt::ssl_client_cert, 2),
3423            SVN::Client::get_ssl_client_cert_pw_prompt_provider(
3424              \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
3425            SVN::Client::get_username_provider(),
3426            SVN::Client::get_ssl_server_trust_prompt_provider(
3427              \&Git::SVN::Prompt::ssl_server_trust),
3428            SVN::Client::get_username_prompt_provider(
3429              \&Git::SVN::Prompt::username, 2),
3430          ]);
3431        my $config = SVN::Core::config_get_config($config_dir);
3432        my $self = SVN::Ra->new(url => $url, auth => $baton,
3433                              config => $config,
3434                              pool => SVN::Pool->new,
3435                              auth_provider_callbacks => $callbacks);
3436        $self->{svn_path} = $url;
3437        $self->{repos_root} = $self->get_repos_root;
3438        $self->{svn_path} =~ s#^\Q$self->{repos_root}\E/*##;
3439        bless $self, $class;
3440}
3441
3442sub DESTROY {
3443        my $self = shift;
3444        $self->{pool}->clear if $self->{pool};
3445        $self->SUPER::DESTROY(@_);
3446}
3447
3448sub dup {
3449        my ($self) = @_;
3450        my $dup = SVN::Ra->new(pool => SVN::Pool->new,
3451                                map { $_ => $self->{$_} } qw/config url
3452                     auth auth_provider_callbacks repos_root svn_path/);
3453        bless $dup, ref $self;
3454}
3455
3456sub get_log {
3457        my ($self, @args) = @_;
3458        my $pool = SVN::Pool->new;
3459        $args[4]-- if $args[4] && ! $::_follow_parent;
3460        splice(@args, 3, 1) if ($SVN::Core::VERSION le '1.2.0');
3461        my $ret = $self->SUPER::get_log(@args, $pool);
3462        $pool->clear;
3463        $ret;
3464}
3465
3466sub get_commit_editor {
3467        my ($self, $msg, $cb, $pool) = @_;
3468        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
3469        $self->SUPER::get_commit_editor($msg, $cb, @lock, $pool);
3470}
3471
3472sub uuid {
3473        my ($self) = @_;
3474        $self->{uuid} ||= $self->get_uuid;
3475}
3476
3477sub gs_do_update {
3478        my ($self, $rev_a, $rev_b, $path, $recurse, $editor) = @_;
3479        my $pool = SVN::Pool->new;
3480        my $reporter = $self->do_update($rev_b, $path, $recurse,
3481                                        $editor, $pool);
3482        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
3483        my $new = ($rev_a == $rev_b);
3484        $reporter->set_path($path, $rev_a, $new, @lock, $pool);
3485        $reporter->finish_report($pool);
3486        $pool->clear;
3487        $editor->{git_commit_ok};
3488}
3489
3490sub gs_do_switch {
3491        my ($self, $rev_a, $rev_b, $path, $recurse, $url_b, $editor) = @_;
3492        my $pool = SVN::Pool->new;
3493        my $reporter = $self->do_switch($rev_b, $path, $recurse,
3494                                        $url_b, $editor, $pool);
3495        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
3496        $reporter->set_path($path, $rev_a, 0, @lock, $pool);
3497        $reporter->finish_report($pool);
3498        $pool->clear;
3499        $editor->{git_commit_ok};
3500}
3501
3502sub can_do_switch {
3503        my $self = shift;
3504        unless (defined $can_do_switch) {
3505                my $pool = SVN::Pool->new;
3506                my $rep = eval {
3507                        $self->do_switch(1, '', 0, $self->{url},
3508                                         SVN::Delta::Editor->new, $pool);
3509                };
3510                if ($@) {
3511                        $can_do_switch = 0;
3512                } else {
3513                        $rep->abort_report($pool);
3514                        $can_do_switch = 1;
3515                }
3516                $pool->clear;
3517        }
3518        $can_do_switch;
3519}
3520
3521__END__
3522
3523Data structures:
3524
3525$log_msg hashref as returned by libsvn_log_entry()
3526{
3527        msg => 'whitespace-formatted log entry
3528',                                              # trailing newline is preserved
3529        revision => '8',                        # integer
3530        date => '2004-02-24T17:01:44.108345Z',  # commit date
3531        author => 'committer name'
3532};
3533
3534@mods = array of diff-index line hashes, each element represents one line
3535        of diff-index output
3536
3537diff-index line ($m hash)
3538{
3539        mode_a => first column of diff-index output, no leading ':',
3540        mode_b => second column of diff-index output,
3541        sha1_b => sha1sum of the final blob,
3542        chg => change type [MCRADT],
3543        file_a => original file name of a file (iff chg is 'C' or 'R')
3544        file_b => new/current file name of a file (any chg)
3545}
3546;
3547
3548# retval of read_url_paths{,_all}();
3549$l_map = {
3550        # repository root url
3551        'https://svn.musicpd.org' => {
3552                # repository path               # GIT_SVN_ID
3553                'mpd/trunk'             =>      'trunk',
3554                'mpd/tags/0.11.5'       =>      'tags/0.11.5',
3555        },
3556}
3557
3558Notes:
3559        I don't trust the each() function on unless I created %hash myself
3560        because the internal iterator may not have started at base.