git-svn.perlon commit git-svn: correctly track revisions made to deleted branches (e5a0b24)
   1#!/usr/bin/env perl
   2# Copyright (C) 2006, Eric Wong <normalperson@yhbt.net>
   3# License: GPL v2 or later
   4use warnings;
   5use strict;
   6use vars qw/    $AUTHOR $VERSION
   7                $SVN_URL
   8                $GIT_SVN_INDEX $GIT_SVN
   9                $GIT_DIR $GIT_SVN_DIR $REVDB
  10                $_follow_parent $sha1 $sha1_short $_revision
  11                $_cp_remote $_upgrade $_rmdir $_q $_cp_similarity
  12                $_find_copies_harder $_l $_authors %users/;
  13$AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
  14$VERSION = '@@GIT_VERSION@@';
  15
  16$ENV{GIT_DIR} ||= '.git';
  17$Git::SVN::default_repo_id = 'git-svn';
  18$Git::SVN::default_ref_id = $ENV{GIT_SVN_ID} || 'git-svn';
  19
  20my $LC_ALL = $ENV{LC_ALL};
  21$Git::SVN::Log::TZ = $ENV{TZ};
  22# make sure the svn binary gives consistent output between locales and TZs:
  23$ENV{TZ} = 'UTC';
  24$ENV{LC_ALL} = 'C';
  25$| = 1; # unbuffer STDOUT
  26
  27sub fatal (@) { print STDERR @_; exit 1 }
  28require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
  29require SVN::Ra;
  30require SVN::Delta;
  31if ($SVN::Core::VERSION lt '1.1.0') {
  32        fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)\n";
  33}
  34push @Git::SVN::Ra::ISA, 'SVN::Ra';
  35push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
  36push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
  37use Carp qw/croak/;
  38use IO::File qw//;
  39use File::Basename qw/dirname basename/;
  40use File::Path qw/mkpath/;
  41use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev pass_through/;
  42use IPC::Open3;
  43use Git;
  44
  45BEGIN {
  46        my $s;
  47        foreach (qw/command command_oneline command_noisy command_output_pipe
  48                    command_input_pipe command_close_pipe/) {
  49                $s .= "*SVN::Git::Editor::$_ = *SVN::Git::Fetcher::$_ = ".
  50                      "*Git::SVN::Migration::$_ = ".
  51                      "*Git::SVN::Log::$_ = *Git::SVN::$_ = *$_ = *Git::$_; ";
  52        }
  53        eval $s;
  54}
  55
  56my ($SVN);
  57
  58my $_optimize_commits = 1 unless $ENV{GIT_SVN_NO_OPTIMIZE_COMMITS};
  59$sha1 = qr/[a-f\d]{40}/;
  60$sha1_short = qr/[a-f\d]{4,40}/;
  61my ($_stdin, $_help, $_edit,
  62        $_repack, $_repack_nr, $_repack_flags,
  63        $_message, $_file, $_no_metadata,
  64        $_template, $_shared,
  65        $_version, $_upgrade,
  66        $_merge, $_strategy, $_dry_run,
  67        $_prefix);
  68
  69my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
  70                    'config-dir=s' => \$Git::SVN::Ra::config_dir,
  71                    'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache );
  72my %fc_opts = ( 'follow-parent|follow' => \$_follow_parent,
  73                'authors-file|A=s' => \$_authors,
  74                'repack:i' => \$_repack,
  75                'no-metadata' => \$_no_metadata,
  76                'quiet|q' => \$_q,
  77                'repack-flags|repack-args|repack-opts=s' => \$_repack_flags,
  78                %remote_opts );
  79
  80my ($_trunk, $_tags, $_branches);
  81my %multi_opts = ( 'trunk|T=s' => \$_trunk,
  82                'tags|t=s' => \$_tags,
  83                'branches|b=s' => \$_branches );
  84my %init_opts = ( 'template=s' => \$_template, 'shared' => \$_shared );
  85my %cmt_opts = ( 'edit|e' => \$_edit,
  86                'rmdir' => \$_rmdir,
  87                'find-copies-harder' => \$_find_copies_harder,
  88                'l=i' => \$_l,
  89                'copy-similarity|C=i'=> \$_cp_similarity
  90);
  91
  92my %cmd = (
  93        fetch => [ \&cmd_fetch, "Download new revisions from SVN",
  94                        { 'revision|r=s' => \$_revision, %fc_opts } ],
  95        init => [ \&cmd_init, "Initialize a repo for tracking" .
  96                          " (requires URL argument)",
  97                          \%init_opts ],
  98        dcommit => [ \&cmd_dcommit,
  99                     'Commit several diffs to merge with upstream',
 100                        { 'merge|m|M' => \$_merge,
 101                          'strategy|s=s' => \$_strategy,
 102                          'dry-run|n' => \$_dry_run,
 103                        %cmt_opts, %fc_opts } ],
 104        'set-tree' => [ \&cmd_set_tree,
 105                        "Set an SVN repository to a git tree-ish",
 106                        { 'stdin|' => \$_stdin, %cmt_opts, %fc_opts, } ],
 107        'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
 108                        { 'revision|r=i' => \$_revision } ],
 109        rebuild => [ \&cmd_rebuild, "Rebuild git-svn metadata (after git clone)",
 110                        { 'copy-remote|remote=s' => \$_cp_remote,
 111                          'upgrade' => \$_upgrade } ],
 112        'multi-init' => [ \&cmd_multi_init,
 113                        'Initialize multiple trees (like git-svnimport)',
 114                        { %multi_opts, %init_opts, %remote_opts,
 115                         'revision|r=i' => \$_revision,
 116                         'prefix=s' => \$_prefix,
 117                        } ],
 118        'multi-fetch' => [ \&cmd_multi_fetch,
 119                        'Fetch multiple trees (like git-svnimport)',
 120                        \%fc_opts ],
 121        'migrate' => [ sub { },
 122                       # no-op, we automatically run this anyways,
 123                       'Migrate configuration/metadata/layout from
 124                        previous versions of git-svn',
 125                        \%remote_opts ],
 126        'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
 127                        { 'limit=i' => \$Git::SVN::Log::limit,
 128                          'revision|r=s' => \$_revision,
 129                          'verbose|v' => \$Git::SVN::Log::verbose,
 130                          'incremental' => \$Git::SVN::Log::incremental,
 131                          'oneline' => \$Git::SVN::Log::oneline,
 132                          'show-commit' => \$Git::SVN::Log::show_commit,
 133                          'non-recursive' => \$Git::SVN::Log::non_recursive,
 134                          'authors-file|A=s' => \$_authors,
 135                          'color' => \$Git::SVN::Log::color,
 136                          'pager=s' => \$Git::SVN::Log::pager,
 137                        } ],
 138        'commit-diff' => [ \&cmd_commit_diff,
 139                           'Commit a diff between two trees',
 140                        { 'message|m=s' => \$_message,
 141                          'file|F=s' => \$_file,
 142                          'revision|r=s' => \$_revision,
 143                        %cmt_opts } ],
 144);
 145
 146my $cmd;
 147for (my $i = 0; $i < @ARGV; $i++) {
 148        if (defined $cmd{$ARGV[$i]}) {
 149                $cmd = $ARGV[$i];
 150                splice @ARGV, $i, 1;
 151                last;
 152        }
 153};
 154
 155my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
 156
 157read_repo_config(\%opts);
 158my $rv = GetOptions(%opts, 'help|H|h' => \$_help,
 159                                'version|V' => \$_version,
 160                                'minimize-connections' =>
 161                                  \$Git::SVN::Migration::_minimize,
 162                                'id|i=s' => \$Git::SVN::default_ref_id);
 163exit 1 if (!$rv && $cmd ne 'log');
 164
 165usage(0) if $_help;
 166version() if $_version;
 167usage(1) unless defined $cmd;
 168load_authors() if $_authors;
 169unless ($cmd =~ /^(?:init|rebuild|multi-init|commit-diff)$/) {
 170        Git::SVN::Migration::migration_check();
 171}
 172eval {
 173        Git::SVN::verify_remotes_sanity();
 174        $cmd{$cmd}->[0]->(@ARGV);
 175};
 176fatal $@ if $@;
 177exit 0;
 178
 179####################### primary functions ######################
 180sub usage {
 181        my $exit = shift || 0;
 182        my $fd = $exit ? \*STDERR : \*STDOUT;
 183        print $fd <<"";
 184git-svn - bidirectional operations between a single Subversion tree and git
 185Usage: $0 <command> [options] [arguments]\n
 186
 187        print $fd "Available commands:\n" unless $cmd;
 188
 189        foreach (sort keys %cmd) {
 190                next if $cmd && $cmd ne $_;
 191                print $fd '  ',pack('A17',$_),$cmd{$_}->[1],"\n";
 192                foreach (keys %{$cmd{$_}->[2]}) {
 193                        # prints out arguments as they should be passed:
 194                        my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
 195                        print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
 196                                                        "--$_" : "-$_" }
 197                                                split /\|/,$_)," $x\n";
 198                }
 199        }
 200        print $fd <<"";
 201\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
 202arbitrary identifier if you're tracking multiple SVN branches/repositories in
 203one git repository and want to keep them separate.  See git-svn(1) for more
 204information.
 205
 206        exit $exit;
 207}
 208
 209sub version {
 210        print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
 211        exit 0;
 212}
 213
 214sub cmd_rebuild {
 215        my $url = shift;
 216        my $gs = $url ? Git::SVN->init($url)
 217                      : eval { Git::SVN->new };
 218        $gs ||= Git::SVN->_new;
 219        if (!verify_ref($gs->refname.'^0')) {
 220                $gs->copy_remote_ref;
 221        }
 222
 223        my ($rev_list, $ctx) = command_output_pipe("rev-list", $gs->refname);
 224        my $latest;
 225        my $svn_uuid;
 226        while (<$rev_list>) {
 227                chomp;
 228                my $c = $_;
 229                fatal "Non-SHA1: $c\n" unless $c =~ /^$sha1$/o;
 230                my ($url, $rev, $uuid) = cmt_metadata($c);
 231
 232                # ignore merges (from set-tree)
 233                next if (!defined $rev || !$uuid);
 234
 235                # if we merged or otherwise started elsewhere, this is
 236                # how we break out of it
 237                if ((defined $svn_uuid && ($uuid ne $svn_uuid)) ||
 238                    ($gs->{url} && $url && ($url ne $gs->{url}))) {
 239                        next;
 240                }
 241
 242                unless (defined $latest) {
 243                        if (!$gs->{url} && !$url) {
 244                                fatal "SVN repository location required\n";
 245                        }
 246                        $gs = Git::SVN->init($url);
 247                        $latest = $rev;
 248                }
 249                $gs->rev_db_set($rev, $c);
 250                print "r$rev = $c\n";
 251        }
 252        command_close_pipe($rev_list, $ctx);
 253}
 254
 255sub do_git_init_db {
 256        unless (-d $ENV{GIT_DIR}) {
 257                my @init_db = ('init');
 258                push @init_db, "--template=$_template" if defined $_template;
 259                push @init_db, "--shared" if defined $_shared;
 260                command_noisy(@init_db);
 261        }
 262}
 263
 264sub cmd_init {
 265        my $url = shift or die "SVN repository location required " .
 266                                "as a command-line argument\n";
 267        if (my $repo_path = shift) {
 268                unless (-d $repo_path) {
 269                        mkpath([$repo_path]);
 270                }
 271                chdir $repo_path or croak $!;
 272                $ENV{GIT_DIR} = $repo_path . "/.git";
 273        }
 274        do_git_init_db();
 275
 276        Git::SVN->init($url);
 277}
 278
 279sub cmd_fetch {
 280        if (@_) {
 281                die "Additional fetch arguments are no longer supported.\n",
 282                    "Use --follow-parent if you have moved/copied directories
 283                    instead.\n";
 284        }
 285        my $gs = Git::SVN->new;
 286        $gs->fetch;
 287        if ($gs->{last_commit} && !verify_ref('refs/heads/master^0')) {
 288                command_noisy(qw(update-ref refs/heads/master),
 289                              $gs->{last_commit});
 290        }
 291}
 292
 293sub cmd_set_tree {
 294        my (@commits) = @_;
 295        if ($_stdin || !@commits) {
 296                print "Reading from stdin...\n";
 297                @commits = ();
 298                while (<STDIN>) {
 299                        if (/\b($sha1_short)\b/o) {
 300                                unshift @commits, $1;
 301                        }
 302                }
 303        }
 304        my @revs;
 305        foreach my $c (@commits) {
 306                my @tmp = command('rev-parse',$c);
 307                if (scalar @tmp == 1) {
 308                        push @revs, $tmp[0];
 309                } elsif (scalar @tmp > 1) {
 310                        push @revs, reverse(command('rev-list',@tmp));
 311                } else {
 312                        fatal "Failed to rev-parse $c\n";
 313                }
 314        }
 315        my $gs = Git::SVN->new;
 316        my ($r_last, $cmt_last) = $gs->last_rev_commit;
 317        $gs->fetch;
 318        if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
 319                fatal "There are new revisions that were fetched ",
 320                      "and need to be merged (or acknowledged) ",
 321                      "before committing.\nlast rev: $r_last\n",
 322                      " current: $gs->{last_rev}\n";
 323        }
 324        $gs->set_tree($_) foreach @revs;
 325        print "Done committing ",scalar @revs," revisions to SVN\n";
 326}
 327
 328sub cmd_dcommit {
 329        my $head = shift;
 330        my $gs = Git::SVN->new;
 331        $head ||= 'HEAD';
 332        my @refs = command(qw/rev-list --no-merges/, $gs->refname."..$head");
 333        my $last_rev;
 334        foreach my $d (reverse @refs) {
 335                if (!verify_ref("$d~1")) {
 336                        fatal "Commit $d\n",
 337                              "has no parent commit, and therefore ",
 338                              "nothing to diff against.\n",
 339                              "You should be working from a repository ",
 340                              "originally created by git-svn\n";
 341                }
 342                unless (defined $last_rev) {
 343                        (undef, $last_rev, undef) = cmt_metadata("$d~1");
 344                        unless (defined $last_rev) {
 345                                fatal "Unable to extract revision information ",
 346                                      "from commit $d~1\n";
 347                        }
 348                }
 349                if ($_dry_run) {
 350                        print "diff-tree $d~1 $d\n";
 351                } else {
 352                        my $log = get_commit_entry($d)->{log};
 353                        my $ra = $gs->ra;
 354                        my $pool = SVN::Pool->new;
 355                        my %ed_opts = ( r => $last_rev,
 356                                        ra => $ra->dup,
 357                                        svn_path => $ra->{svn_path} );
 358                        my $ed = SVN::Git::Editor->new(\%ed_opts,
 359                                         $ra->get_commit_editor($log,
 360                                         sub { print "Committed r$_[0]\n";
 361                                               $last_rev = $_[0]; }),
 362                                         $pool);
 363                        my $mods = $ed->apply_diff("$d~1", $d);
 364                        if (@$mods == 0) {
 365                                print "No changes\n$d~1 == $d\n";
 366                        }
 367                }
 368        }
 369        return if $_dry_run;
 370        $gs->fetch;
 371        # we always want to rebase against the current HEAD, not any
 372        # head that was passed to us
 373        my @diff = command('diff-tree', 'HEAD', $gs->refname, '--');
 374        my @finish;
 375        if (@diff) {
 376                @finish = qw/rebase/;
 377                push @finish, qw/--merge/ if $_merge;
 378                push @finish, "--strategy=$_strategy" if $_strategy;
 379                print STDERR "W: HEAD and ", $gs->refname, " differ, ",
 380                             "using @finish:\n", "@diff";
 381        } else {
 382                print "No changes between current HEAD and ",
 383                      $gs->refname, "\nResetting to the latest ",
 384                      $gs->refname, "\n";
 385                @finish = qw/reset --mixed/;
 386        }
 387        command_noisy(@finish, $gs->refname);
 388}
 389
 390sub cmd_show_ignore {
 391        my $gs = Git::SVN->new;
 392        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 393        $gs->traverse_ignore(\*STDOUT, '', $r);
 394}
 395
 396sub cmd_multi_init {
 397        my $url = shift;
 398        unless (defined $_trunk || defined $_branches || defined $_tags) {
 399                usage(1);
 400        }
 401        do_git_init_db();
 402        $_prefix = '' unless defined $_prefix;
 403        $url =~ s#/+$## if defined $url;
 404        if (defined $_trunk) {
 405                my $trunk_ref = $_prefix . 'trunk';
 406                # try both old-style and new-style lookups:
 407                my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
 408                unless ($gs_trunk) {
 409                        my ($trunk_url, $trunk_path) =
 410                                              complete_svn_url($url, $_trunk);
 411                        $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
 412                                                   undef, $trunk_ref);
 413                }
 414        }
 415        return unless defined $_branches || defined $_tags;
 416        my $ra = $url ? Git::SVN::Ra->new($url) : undef;
 417        complete_url_ls_init($ra, $_branches, '--branches/-b', $_prefix);
 418        complete_url_ls_init($ra, $_tags, '--tags/-t', $_prefix . 'tags/');
 419}
 420
 421sub cmd_multi_fetch {
 422        my @gs;
 423        foreach (command(qw/config -l/)) {
 424                next unless m!^svn-remote\.(.+)\.fetch=
 425                              \s*(.*)\s*:\s*refs/remotes/(.+)\s*$!x;
 426                my ($repo_id, $path, $ref_id) = ($1, $2, $3);
 427                push @gs, Git::SVN->new($ref_id, $repo_id, $path);
 428        }
 429        foreach (@gs) {
 430                $_->fetch;
 431        }
 432}
 433
 434# this command is special because it requires no metadata
 435sub cmd_commit_diff {
 436        my ($ta, $tb, $url) = @_;
 437        my $usage = "Usage: $0 commit-diff -r<revision> ".
 438                    "<tree-ish> <tree-ish> [<URL>]\n";
 439        fatal($usage) if (!defined $ta || !defined $tb);
 440        if (!defined $url) {
 441                my $gs = eval { Git::SVN->new };
 442                if (!$gs) {
 443                        fatal("Needed URL or usable git-svn --id in ",
 444                              "the command-line\n", $usage);
 445                }
 446                $url = $gs->{url};
 447        }
 448        unless (defined $_revision) {
 449                fatal("-r|--revision is a required argument\n", $usage);
 450        }
 451        if (defined $_message && defined $_file) {
 452                fatal("Both --message/-m and --file/-F specified ",
 453                      "for the commit message.\n",
 454                      "I have no idea what you mean\n");
 455        }
 456        if (defined $_file) {
 457                $_message = file_to_s($_file);
 458        } else {
 459                $_message ||= get_commit_entry($tb)->{log};
 460        }
 461        my $ra ||= Git::SVN::Ra->new($url);
 462        my $r = $_revision;
 463        if ($r eq 'HEAD') {
 464                $r = $ra->get_latest_revnum;
 465        } elsif ($r !~ /^\d+$/) {
 466                die "revision argument: $r not understood by git-svn\n";
 467        }
 468        my $pool = SVN::Pool->new;
 469        my %ed_opts = ( r => $r,
 470                        ra => $ra->dup,
 471                        svn_path => $ra->{svn_path} );
 472        my $ed = SVN::Git::Editor->new(\%ed_opts,
 473                                       $ra->get_commit_editor($_message,
 474                                         sub { print "Committed r$_[0]\n" }),
 475                                       $pool);
 476        my $mods = $ed->apply_diff($ta, $tb);
 477        if (@$mods == 0) {
 478                print "No changes\n$ta == $tb\n";
 479        }
 480        $pool->clear;
 481}
 482
 483########################### utility functions #########################
 484
 485sub complete_svn_url {
 486        my ($url, $path) = @_;
 487        $path =~ s#/+$##;
 488        if ($path !~ m#^[a-z\+]+://#) {
 489                if (!defined $url || $url !~ m#^[a-z\+]+://#) {
 490                        fatal("E: '$path' is not a complete URL ",
 491                              "and a separate URL is not specified\n");
 492                }
 493                return ($url, $path);
 494        }
 495        return ($path, '');
 496}
 497
 498sub complete_url_ls_init {
 499        my ($ra, $repo_path, $switch, $pfx) = @_;
 500        unless ($repo_path) {
 501                print STDERR "W: $switch not specified\n";
 502                return;
 503        }
 504        $repo_path =~ s#/+$##;
 505        if ($repo_path =~ m#^[a-z\+]+://#) {
 506                $ra = Git::SVN::Ra->new($repo_path);
 507                $repo_path = '';
 508        } else {
 509                $repo_path =~ s#^/+##;
 510                unless ($ra) {
 511                        fatal("E: '$repo_path' is not a complete URL ",
 512                              "and a separate URL is not specified\n");
 513                }
 514        }
 515        my $r = defined $_revision ? $_revision : $ra->get_latest_revnum;
 516        my ($dirent, undef, undef) = $ra->get_dir($repo_path, $r);
 517        my $url = $ra->{url};
 518        foreach my $d (sort keys %$dirent) {
 519                next if ($dirent->{$d}->kind != $SVN::Node::dir);
 520                my $path =  "$repo_path/$d";
 521                my $ref = "$pfx$d";
 522                my $gs = eval { Git::SVN->new($ref) };
 523                # don't try to init already existing refs
 524                unless ($gs) {
 525                        print "init $url/$path => $ref\n";
 526                        Git::SVN->init($url, $path, undef, $ref);
 527                }
 528        }
 529}
 530
 531sub verify_ref {
 532        my ($ref) = @_;
 533        eval { command_oneline([ 'rev-parse', '--verify', $ref ],
 534                               { STDERR => 0 }); };
 535}
 536
 537sub get_tree_from_treeish {
 538        my ($treeish) = @_;
 539        # $treeish can be a symbolic ref, too:
 540        my $type = command_oneline(qw/cat-file -t/, $treeish);
 541        my $expected;
 542        while ($type eq 'tag') {
 543                ($treeish, $type) = command(qw/cat-file tag/, $treeish);
 544        }
 545        if ($type eq 'commit') {
 546                $expected = (grep /^tree /, command(qw/cat-file commit/,
 547                                                    $treeish))[0];
 548                ($expected) = ($expected =~ /^tree ($sha1)$/o);
 549                die "Unable to get tree from $treeish\n" unless $expected;
 550        } elsif ($type eq 'tree') {
 551                $expected = $treeish;
 552        } else {
 553                die "$treeish is a $type, expected tree, tag or commit\n";
 554        }
 555        return $expected;
 556}
 557
 558sub get_commit_entry {
 559        my ($treeish) = shift;
 560        my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
 561        my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
 562        my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
 563        open my $log_fh, '>', $commit_editmsg or croak $!;
 564
 565        my $type = command_oneline(qw/cat-file -t/, $treeish);
 566        if ($type eq 'commit' || $type eq 'tag') {
 567                my ($msg_fh, $ctx) = command_output_pipe('cat-file',
 568                                                         $type, $treeish);
 569                my $in_msg = 0;
 570                while (<$msg_fh>) {
 571                        if (!$in_msg) {
 572                                $in_msg = 1 if (/^\s*$/);
 573                        } elsif (/^git-svn-id: /) {
 574                                # skip this for now, we regenerate the
 575                                # correct one on re-fetch anyways
 576                                # TODO: set *:merge properties or like...
 577                        } else {
 578                                print $log_fh $_ or croak $!;
 579                        }
 580                }
 581                command_close_pipe($msg_fh, $ctx);
 582        }
 583        close $log_fh or croak $!;
 584
 585        if ($_edit || ($type eq 'tree')) {
 586                my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
 587                # TODO: strip out spaces, comments, like git-commit.sh
 588                system($editor, $commit_editmsg);
 589        }
 590        rename $commit_editmsg, $commit_msg or croak $!;
 591        open $log_fh, '<', $commit_msg or croak $!;
 592        { local $/; chomp($log_entry{log} = <$log_fh>); }
 593        close $log_fh or croak $!;
 594        unlink $commit_msg;
 595        \%log_entry;
 596}
 597
 598sub s_to_file {
 599        my ($str, $file, $mode) = @_;
 600        open my $fd,'>',$file or croak $!;
 601        print $fd $str,"\n" or croak $!;
 602        close $fd or croak $!;
 603        chmod ($mode &~ umask, $file) if (defined $mode);
 604}
 605
 606sub file_to_s {
 607        my $file = shift;
 608        open my $fd,'<',$file or croak "$!: file: $file\n";
 609        local $/;
 610        my $ret = <$fd>;
 611        close $fd or croak $!;
 612        $ret =~ s/\s*$//s;
 613        return $ret;
 614}
 615
 616# '<svn username> = real-name <email address>' mapping based on git-svnimport:
 617sub load_authors {
 618        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
 619        my $log = $cmd eq 'log';
 620        while (<$authors>) {
 621                chomp;
 622                next unless /^(\S+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
 623                my ($user, $name, $email) = ($1, $2, $3);
 624                if ($log) {
 625                        $Git::SVN::Log::rusers{"$name <$email>"} = $user;
 626                } else {
 627                        $users{$user} = [$name, $email];
 628                }
 629        }
 630        close $authors or croak $!;
 631}
 632
 633# convert GetOpt::Long specs for use by git-config
 634sub read_repo_config {
 635        return unless -d $ENV{GIT_DIR};
 636        my $opts = shift;
 637        foreach my $o (keys %$opts) {
 638                my $v = $opts->{$o};
 639                my ($key) = ($o =~ /^([a-z\-]+)/);
 640                $key =~ s/-//g;
 641                my $arg = 'git-config';
 642                $arg .= ' --int' if ($o =~ /[:=]i$/);
 643                $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
 644                if (ref $v eq 'ARRAY') {
 645                        chomp(my @tmp = `$arg --get-all svn.$key`);
 646                        @$v = @tmp if @tmp;
 647                } else {
 648                        chomp(my $tmp = `$arg --get svn.$key`);
 649                        if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
 650                                $$v = $tmp;
 651                        }
 652                }
 653        }
 654}
 655
 656sub extract_metadata {
 657        my $id = shift or return (undef, undef, undef);
 658        my ($url, $rev, $uuid) = ($id =~ /^git-svn-id:\s(\S+?)\@(\d+)
 659                                                        \s([a-f\d\-]+)$/x);
 660        if (!defined $rev || !$uuid || !$url) {
 661                # some of the original repositories I made had
 662                # identifiers like this:
 663                ($rev, $uuid) = ($id =~/^git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
 664        }
 665        return ($url, $rev, $uuid);
 666}
 667
 668sub cmt_metadata {
 669        return extract_metadata((grep(/^git-svn-id: /,
 670                command(qw/cat-file commit/, shift)))[-1]);
 671}
 672
 673sub get_commit_time {
 674        my $cmt = shift;
 675        my $fh = command_output_pipe(qw/rev-list --pretty=raw -n1/, $cmt);
 676        while (<$fh>) {
 677                /^committer\s(?:.+) (\d+) ([\-\+]?\d+)$/ or next;
 678                my ($s, $tz) = ($1, $2);
 679                if ($tz =~ s/^\+//) {
 680                        $s += tz_to_s_offset($tz);
 681                } elsif ($tz =~ s/^\-//) {
 682                        $s -= tz_to_s_offset($tz);
 683                }
 684                close $fh;
 685                return $s;
 686        }
 687        die "Can't get commit time for commit: $cmt\n";
 688}
 689
 690sub tz_to_s_offset {
 691        my ($tz) = @_;
 692        $tz =~ s/(\d\d)$//;
 693        return ($1 * 60) + ($tz * 3600);
 694}
 695
 696package Git::SVN;
 697use strict;
 698use warnings;
 699use vars qw/$default_repo_id $default_ref_id/;
 700use Carp qw/croak/;
 701use File::Path qw/mkpath/;
 702use IPC::Open3;
 703
 704# properties that we do not log:
 705my %SKIP_PROP;
 706BEGIN {
 707        %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
 708                                        svn:special svn:executable
 709                                        svn:entry:committed-rev
 710                                        svn:entry:last-author
 711                                        svn:entry:uuid
 712                                        svn:entry:committed-date/;
 713}
 714
 715sub read_all_remotes {
 716        my $r = {};
 717        foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
 718                if (m!^(.+)\.fetch=\s*(.*)\s*:\s*refs/remotes/(.+)\s*$!) {
 719                        $r->{$1}->{fetch}->{$2} = $3;
 720                } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
 721                        $r->{$1}->{url} = $2;
 722                }
 723        }
 724        $r;
 725}
 726
 727sub verify_remotes_sanity {
 728        return unless -d $ENV{GIT_DIR};
 729        my %seen;
 730        foreach (command(qw/config -l/)) {
 731                if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
 732                        if ($seen{$1}) {
 733                                die "Remote ref refs/remote/$1 is tracked by",
 734                                    "\n  \"$_\"\nand\n  \"$seen{$1}\"\n",
 735                                    "Please resolve this ambiguity in ",
 736                                    "your git configuration file before ",
 737                                    "continuing\n";
 738                        }
 739                        $seen{$1} = $_;
 740                }
 741        }
 742}
 743
 744# we allow more chars than remotes2config.sh...
 745sub sanitize_remote_name {
 746        my ($name) = @_;
 747        $name =~ tr{A-Za-z0-9:,/+-}{.}c;
 748        $name;
 749}
 750
 751sub find_existing_remote {
 752        my ($url, $remotes) = @_;
 753        my $existing;
 754        foreach my $repo_id (keys %$remotes) {
 755                my $u = $remotes->{$repo_id}->{url} or next;
 756                next if $u ne $url;
 757                $existing = $repo_id;
 758                last;
 759        }
 760        $existing;
 761}
 762
 763sub init_remote_config {
 764        my ($self, $url) = @_;
 765        $url =~ s!/+$!!; # strip trailing slash
 766        my $r = read_all_remotes();
 767        my $existing = find_existing_remote($url, $r);
 768        if ($existing) {
 769                print STDERR "Using existing ",
 770                             "[svn-remote \"$existing\"]\n";
 771                $self->{repo_id} = $existing;
 772        } else {
 773                my $min_url = Git::SVN::Ra->new($url)->minimize_url;
 774                $existing = find_existing_remote($min_url, $r);
 775                if ($existing) {
 776                        print STDERR "Using existing ",
 777                                     "[svn-remote \"$existing\"]\n";
 778                        $self->{repo_id} = $existing;
 779                }
 780                if ($min_url ne $url) {
 781                        print STDERR "Using higher level of URL: ",
 782                                     "$url => $min_url\n";
 783                        my $old_path = $self->{path};
 784                        $self->{path} = $url;
 785                        $self->{path} =~ s!^\Q$min_url\E/*!!;
 786                        if (length $old_path) {
 787                                $self->{path} .= "/$old_path";
 788                        }
 789                        $url = $min_url;
 790                }
 791        }
 792        my $orig_url;
 793        if (!$existing) {
 794                # verify that we aren't overwriting anything:
 795                $orig_url = eval {
 796                        command_oneline('config', '--get',
 797                                        "svn-remote.$self->{repo_id}.url")
 798                };
 799                if ($orig_url && ($orig_url ne $url)) {
 800                        die "svn-remote.$self->{repo_id}.url already set: ",
 801                            "$orig_url\nwanted to set to: $url\n";
 802                }
 803        }
 804        my ($xrepo_id, $xpath) = find_ref($self->refname);
 805        if (defined $xpath) {
 806                die "svn-remote.$xrepo_id.fetch already set to track ",
 807                    "$xpath:refs/remotes/", $self->refname, "\n";
 808        }
 809        command_noisy('config',
 810                      "svn-remote.$self->{repo_id}.url", $url);
 811        command_noisy('config', '--add',
 812                      "svn-remote.$self->{repo_id}.fetch",
 813                      "$self->{path}:".$self->refname);
 814        $self->{url} = $url;
 815}
 816
 817sub init {
 818        my ($class, $url, $path, $repo_id, $ref_id) = @_;
 819        my $self = _new($class, $repo_id, $ref_id, $path);
 820        if (defined $url) {
 821                $self->init_remote_config($url);
 822        }
 823        $self;
 824}
 825
 826sub find_ref {
 827        my ($ref_id) = @_;
 828        foreach (command(qw/config -l/)) {
 829                next unless m!^svn-remote\.(.+)\.fetch=
 830                              \s*(.*)\s*:\s*refs/remotes/(.+)\s*$!x;
 831                my ($repo_id, $path, $ref) = ($1, $2, $3);
 832                if ($ref eq $ref_id) {
 833                        $path = '' if ($path =~ m#^\./?#);
 834                        return ($repo_id, $path);
 835                }
 836        }
 837        (undef, undef, undef);
 838}
 839
 840sub new {
 841        my ($class, $ref_id, $repo_id, $path) = @_;
 842        if (defined $ref_id && !defined $repo_id && !defined $path) {
 843                ($repo_id, $path) = find_ref($ref_id);
 844                if (!defined $repo_id) {
 845                        die "Could not find a \"svn-remote.*.fetch\" key ",
 846                            "in the repository configuration matching: ",
 847                            "refs/remotes/$ref_id\n";
 848                }
 849        }
 850        my $self = _new($class, $repo_id, $ref_id, $path);
 851        if (!defined $self->{path} || !length $self->{path}) {
 852                my $fetch = command_oneline('config', '--get',
 853                                            "svn-remote.$repo_id.fetch",
 854                                            ":refs/remotes/$ref_id\$") or
 855                     die "Failed to read \"svn-remote.$repo_id.fetch\" ",
 856                         "\":refs/remotes/$ref_id\$\" in config\n";
 857                ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
 858        }
 859        $self->{url} = command_oneline('config', '--get',
 860                                       "svn-remote.$repo_id.url") or
 861                  die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
 862        $self;
 863}
 864
 865sub refname { "refs/remotes/$_[0]->{ref_id}" }
 866
 867sub ra {
 868        my ($self) = shift;
 869        $self->{ra} ||= Git::SVN::Ra->new($self->{url});
 870}
 871
 872sub rel_path {
 873        my ($self) = @_;
 874        my $repos_root = $self->ra->{repos_root};
 875        return $self->{path} if ($self->{url} eq $repos_root);
 876        my $url = $self->{url} .
 877                  (length $self->{path} ? "/$self->{path}" : $self->{path});
 878        $url =~ s!^\Q$repos_root\E/*!!g;
 879        $url;
 880}
 881
 882sub copy_remote_ref {
 883        my ($self) = @_;
 884        my $origin = $::_cp_remote ? $::_cp_remote : 'origin';
 885        my $ref = $self->refname;
 886        if (command('ls-remote', $origin, $ref)) {
 887                command_noisy('fetch', $origin, "$ref:$ref");
 888        } elsif ($::_cp_remote && !$::_upgrade) {
 889                die "Unable to find remote reference: $ref on $origin\n";
 890        }
 891}
 892
 893sub traverse_ignore {
 894        my ($self, $fh, $path, $r) = @_;
 895        $path =~ s#^/+##g;
 896        my ($dirent, undef, $props) = $self->ra->get_dir($path, $r);
 897        my $p = $path;
 898        $p =~ s#^\Q$self->{ra}->{svn_path}\E/##;
 899        print $fh length $p ? "\n# $p\n" : "\n# /\n";
 900        if (my $s = $props->{'svn:ignore'}) {
 901                $s =~ s/[\r\n]+/\n/g;
 902                chomp $s;
 903                if (length $p == 0) {
 904                        $s =~ s#\n#\n/$p#g;
 905                        print $fh "/$s\n";
 906                } else {
 907                        $s =~ s#\n#\n/$p/#g;
 908                        print $fh "/$p/$s\n";
 909                }
 910        }
 911        foreach (sort keys %$dirent) {
 912                next if $dirent->{$_}->kind != $SVN::Node::dir;
 913                $self->traverse_ignore($fh, "$path/$_", $r);
 914        }
 915}
 916
 917# returns the newest SVN revision number and newest commit SHA1
 918sub last_rev_commit {
 919        my ($self) = @_;
 920        if (defined $self->{last_rev} && defined $self->{last_commit}) {
 921                return ($self->{last_rev}, $self->{last_commit});
 922        }
 923        my $c = ::verify_ref($self->refname.'^0');
 924        if ($c) {
 925                my $rev = (::cmt_metadata($c))[1];
 926                if (defined $rev) {
 927                        ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
 928                        return ($rev, $c);
 929                }
 930        }
 931        my $offset = -41; # from tail
 932        my $rl;
 933        open my $fh, '<', $self->{db_path} or
 934                                 croak "$self->{db_path} not readable: $!\n";
 935        seek $fh, $offset, 2;
 936        $rl = readline $fh;
 937        defined $rl or return (undef, undef);
 938        chomp $rl;
 939        while ($c ne $rl && tell $fh != 0) {
 940                $offset -= 41;
 941                seek $fh, $offset, 2;
 942                $rl = readline $fh;
 943                defined $rl or return (undef, undef);
 944                chomp $rl;
 945        }
 946        my $rev = tell $fh;
 947        croak $! if ($rev < 0);
 948        $rev =  ($rev - 41) / 41;
 949        close $fh or croak $!;
 950        ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
 951        return ($rev, $c);
 952}
 953
 954sub parse_revision {
 955        my ($self, $base) = @_;
 956        my $head = $self->ra->get_latest_revnum;
 957        if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
 958                return ($base + 1, $head) if (defined $base);
 959                return (0, $head);
 960        }
 961        return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
 962        return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
 963        if ($::_revision =~ /^BASE:(\d+)$/) {
 964                return ($base + 1, $1) if (defined $base);
 965                return (0, $head);
 966        }
 967        return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
 968        die "revision argument: $::_revision not understood by git-svn\n",
 969                "Try using the command-line svn client instead\n";
 970}
 971
 972sub tmp_index_do {
 973        my ($self, $sub) = @_;
 974        my $old_index = $ENV{GIT_INDEX_FILE};
 975        $ENV{GIT_INDEX_FILE} = $self->{index};
 976        my @ret = &$sub;
 977        if ($old_index) {
 978                $ENV{GIT_INDEX_FILE} = $old_index;
 979        } else {
 980                delete $ENV{GIT_INDEX_FILE};
 981        }
 982        wantarray ? @ret : $ret[0];
 983}
 984
 985sub assert_index_clean {
 986        my ($self, $treeish) = @_;
 987
 988        $self->tmp_index_do(sub {
 989                command_noisy('read-tree', $treeish) unless -e $self->{index};
 990                my $x = command_oneline('write-tree');
 991                my ($y) = (command(qw/cat-file commit/, $treeish) =~
 992                           /^tree ($::sha1)/mo);
 993                if ($y ne $x) {
 994                        unlink $self->{index} or croak $!;
 995                        command_noisy('read-tree', $treeish);
 996                }
 997                $x = command_oneline('write-tree');
 998                if ($y ne $x) {
 999                        ::fatal "trees ($treeish) $y != $x\n",
1000                                "Something is seriously wrong...\n";
1001                }
1002        });
1003}
1004
1005sub get_commit_parents {
1006        my ($self, $log_entry, @parents) = @_;
1007        my (%seen, @ret, @tmp);
1008        # commit parents can be conditionally bound to a particular
1009        # svn revision via: "svn_revno=commit_sha1", filter them out here:
1010        foreach my $p (@parents) {
1011                next unless defined $p;
1012                if ($p =~ /^(\d+)=($::sha1_short)$/o) {
1013                        push @tmp, $2 if $1 == $log_entry->{revision};
1014                } else {
1015                        push @tmp, $p if $p =~ /^$::sha1_short$/o;
1016                }
1017        }
1018        if (my $cur = ::verify_ref($self->refname.'^0')) {
1019                push @tmp, $cur;
1020        }
1021        push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
1022        while (my $p = shift @tmp) {
1023                next if $seen{$p};
1024                $seen{$p} = 1;
1025                push @ret, $p;
1026                # MAXPARENT is defined to 16 in commit-tree.c:
1027                last if @ret >= 16;
1028        }
1029        if (@tmp) {
1030                die "r$log_entry->{revision}: No room for parents:\n\t",
1031                    join("\n\t", @tmp), "\n";
1032        }
1033        @ret;
1034}
1035
1036sub full_url {
1037        my ($self) = @_;
1038        $self->ra->{url} . (length $self->{path} ? '/' . $self->{path} : '');
1039}
1040
1041sub do_git_commit {
1042        my ($self, $log_entry, @parents) = @_;
1043        if (my $c = $self->rev_db_get($log_entry->{revision})) {
1044                croak "$log_entry->{revision} = $c already exists! ",
1045                      "Why are we refetching it?\n";
1046        }
1047        my $author = $log_entry->{author};
1048        my ($name, $email) = (defined $::users{$author} ? @{$::users{$author}}
1049                           : ($author, "$author\@".$self->ra->uuid));
1050        $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} = $name;
1051        $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} = $email;
1052        $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
1053
1054        my $tree = $log_entry->{tree};
1055        if (!defined $tree) {
1056                $tree = $self->tmp_index_do(sub {
1057                                            command_oneline('write-tree') });
1058        }
1059        die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
1060
1061        my @exec = ('git-commit-tree', $tree);
1062        foreach ($self->get_commit_parents($log_entry, @parents)) {
1063                push @exec, '-p', $_;
1064        }
1065        defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
1066                                                                   or croak $!;
1067        print $msg_fh $log_entry->{log} or croak $!;
1068        print $msg_fh "\ngit-svn-id: ", $self->full_url, '@',
1069                      $log_entry->{revision}, ' ',
1070                      $self->ra->uuid, "\n" or croak $!;
1071        $msg_fh->flush == 0 or croak $!;
1072        close $msg_fh or croak $!;
1073        chomp(my $commit = do { local $/; <$out_fh> });
1074        close $out_fh or croak $!;
1075        waitpid $pid, 0;
1076        croak $? if $?;
1077        if ($commit !~ /^$::sha1$/o) {
1078                die "Failed to commit, invalid sha1: $commit\n";
1079        }
1080
1081        command_noisy('update-ref',$self->refname, $commit);
1082        $self->rev_db_set($log_entry->{revision}, $commit);
1083
1084        $self->{last_rev} = $log_entry->{revision};
1085        $self->{last_commit} = $commit;
1086        print "r$log_entry->{revision} = $commit\n";
1087        return $commit;
1088}
1089
1090sub revisions_eq {
1091        my ($self, $r0, $r1) = @_;
1092        return 1 if $r0 == $r1;
1093        my $nr = 0;
1094        $self->ra->get_log([$self->{path}], $r0, $r1,
1095                           0, 0, 1, sub { $nr++ });
1096        return 0 if ($nr > 1);
1097        return 1;
1098}
1099
1100sub find_parent_branch {
1101        my ($self, $paths, $rev) = @_;
1102        return undef unless $::_follow_parent;
1103        unless (defined $paths) {
1104                $self->ra->get_log([''], $rev, $rev, 0, 1, 1,
1105                                   sub { $paths = $_[0] });
1106        }
1107        return undef unless defined $paths;
1108
1109        # look for a parent from another branch:
1110        my @b_path_components = split m#/#, $self->rel_path;
1111        my @a_path_components;
1112        my $i;
1113        while (@b_path_components) {
1114                $i = $paths->{'/'.join('/', @b_path_components)};
1115                last if $i;
1116                unshift(@a_path_components, pop(@b_path_components));
1117        }
1118        goto not_found unless defined $i;
1119        my $branch_from = $i->copyfrom_path or goto not_found;
1120        if (@a_path_components) {
1121                print STDERR "branch_from: $branch_from => ";
1122                $branch_from .= '/'.join('/', @a_path_components);
1123                print STDERR $branch_from, "\n";
1124        }
1125        my $r = $i->copyfrom_rev;
1126        my $repos_root = $self->ra->{repos_root};
1127        my $url = $self->ra->{url};
1128        my $new_url = $repos_root . $branch_from;
1129        print STDERR  "Found possible branch point: ",
1130                      "$new_url => ", $self->full_url, ", $r\n";
1131        $branch_from =~ s#^/##;
1132        my $remotes = read_all_remotes();
1133        my $gs;
1134        foreach my $repo_id (keys %$remotes) {
1135                my $u = $remotes->{$repo_id}->{url} or next;
1136                next if $url ne $u;
1137                my $fetch = $remotes->{$repo_id}->{fetch};
1138                foreach my $f (keys %$fetch) {
1139                        next if $f ne $branch_from;
1140                        $gs = Git::SVN->new($fetch->{$f}, $repo_id, $f);
1141                        last;
1142                }
1143                last if $gs;
1144        }
1145        unless ($gs) {
1146                my $ref_id = $branch_from;
1147                $ref_id .= "\@$r" if find_ref($ref_id);
1148                # just grow a tail if we're not unique enough :x
1149                $ref_id .= '-' while find_ref($ref_id);
1150                $gs = Git::SVN->init($new_url, '', $ref_id, $ref_id);
1151        }
1152        my ($r0, $parent) = $gs->find_rev_before($r, 1);
1153        if ($::_follow_parent && (!defined $r0 || !defined $parent)) {
1154                foreach (1 .. $r) {
1155                        if (my $log_entry = $gs->do_fetch(undef, $_)) {
1156                                $gs->do_git_commit($log_entry);
1157                        }
1158                }
1159                ($r0, $parent) = $gs->last_rev_commit;
1160        }
1161        if (defined $r0 && defined $parent && $gs->revisions_eq($r0, $r)) {
1162                print STDERR "Found branch parent: ($self->{ref_id}) $parent\n";
1163                $self->assert_index_clean($parent);
1164                my $ed;
1165                if ($self->ra->can_do_switch) {
1166                        print STDERR "Following parent with do_switch\n";
1167                        # do_switch works with svn/trunk >= r22312, but that
1168                        # is not included with SVN 1.4.2 (the latest version
1169                        # at the moment), so we can't rely on it
1170                        $self->{last_commit} = $parent;
1171                        $ed = SVN::Git::Fetcher->new($self);
1172                        $gs->ra->gs_do_switch($r0, $rev, $gs->{path}, 1,
1173                                              $self->full_url, $ed)
1174                          or die "SVN connection failed somewhere...\n";
1175                } else {
1176                        print STDERR "Following parent with do_update\n";
1177                        $ed = SVN::Git::Fetcher->new($self);
1178                        $self->ra->gs_do_update($rev, $rev, $self->{path},
1179                                                1, $ed)
1180                          or die "SVN connection failed somewhere...\n";
1181                }
1182                return $self->make_log_entry($rev, [$parent], $ed);
1183        }
1184not_found:
1185        print STDERR "Branch parent for path: '/",
1186                     $self->rel_path, "' @ $rev not found\n";
1187        print STDERR '  ', $_, "\n" foreach (sort keys %$paths);
1188        return undef;
1189}
1190
1191sub do_fetch {
1192        my ($self, $paths, $rev) = @_;
1193        my $ed;
1194        my ($last_rev, @parents);
1195        if ($self->{last_commit}) {
1196                $ed = SVN::Git::Fetcher->new($self);
1197                $last_rev = $self->{last_rev};
1198                $ed->{c} = $self->{last_commit};
1199                @parents = ($self->{last_commit});
1200        } else {
1201                $last_rev = $rev;
1202                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
1203                        return $log_entry;
1204                }
1205                $ed = SVN::Git::Fetcher->new($self);
1206        }
1207        unless ($self->ra->gs_do_update($last_rev, $rev,
1208                                        $self->{path}, 1, $ed)) {
1209                die "SVN connection failed somewhere...\n";
1210        }
1211        $self->make_log_entry($rev, \@parents, $ed);
1212}
1213
1214sub get_untracked {
1215        my ($self, $ed) = @_;
1216        my @out;
1217        my $h = $ed->{empty};
1218        foreach (sort keys %$h) {
1219                my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
1220                push @out, "  $act: " . uri_encode($_);
1221                warn "W: $act: $_\n";
1222        }
1223        foreach my $t (qw/dir_prop file_prop/) {
1224                $h = $ed->{$t} or next;
1225                foreach my $path (sort keys %$h) {
1226                        my $ppath = $path eq '' ? '.' : $path;
1227                        foreach my $prop (sort keys %{$h->{$path}}) {
1228                                next if $SKIP_PROP{$prop};
1229                                my $v = $h->{$path}->{$prop};
1230                                my $t_ppath_prop = "$t: " .
1231                                                    uri_encode($ppath) . ' ' .
1232                                                    uri_encode($prop);
1233                                if (defined $v) {
1234                                        push @out, "  +$t_ppath_prop " .
1235                                                   uri_encode($v);
1236                                } else {
1237                                        push @out, "  -$t_ppath_prop";
1238                                }
1239                        }
1240                }
1241        }
1242        foreach my $t (qw/absent_file absent_directory/) {
1243                $h = $ed->{$t} or next;
1244                foreach my $parent (sort keys %$h) {
1245                        foreach my $path (sort @{$h->{$parent}}) {
1246                                push @out, "  $t: " .
1247                                           uri_encode("$parent/$path");
1248                                warn "W: $t: $parent/$path ",
1249                                     "Insufficient permissions?\n";
1250                        }
1251                }
1252        }
1253        \@out;
1254}
1255
1256sub parse_svn_date {
1257        my $date = shift || return '+0000 1970-01-01 00:00:00';
1258        my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
1259                                            (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x) or
1260                                         croak "Unable to parse date: $date\n";
1261        "+0000 $Y-$m-$d $H:$M:$S";
1262}
1263
1264sub check_author {
1265        my ($author) = @_;
1266        if (!defined $author || length $author == 0) {
1267                $author = '(no author)';
1268        }
1269        if (defined $::_authors && ! defined $::users{$author}) {
1270                die "Author: $author not defined in $::_authors file\n";
1271        }
1272        $author;
1273}
1274
1275sub make_log_entry {
1276        my ($self, $rev, $parents, $ed) = @_;
1277        my $untracked = $self->get_untracked($ed);
1278
1279        return undef if ($ed->{nr} == 0 && scalar @$untracked == 0);
1280
1281        open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
1282        print $un "r$rev\n" or croak $!;
1283        print $un $_, "\n" foreach @$untracked;
1284        my %log_entry = ( parents => $parents || [], revision => $rev,
1285                          log => '');
1286        my $rp = $self->ra->rev_proplist($rev);
1287        foreach (sort keys %$rp) {
1288                my $v = $rp->{$_};
1289                if (/^svn:(author|date|log)$/) {
1290                        $log_entry{$1} = $v;
1291                } else {
1292                        print $un "  rev_prop: ", uri_encode($_), ' ',
1293                                  uri_encode($v), "\n";
1294                }
1295        }
1296        close $un or croak $!;
1297
1298        $log_entry{date} = parse_svn_date($log_entry{date});
1299        $log_entry{author} = check_author($log_entry{author});
1300        $log_entry{log} .= "\n";
1301        \%log_entry;
1302}
1303
1304sub fetch {
1305        my ($self, @parents) = @_;
1306        my ($last_rev, $last_commit) = $self->last_rev_commit;
1307        my ($base, $head) = $self->parse_revision($last_rev);
1308        return if ($base > $head);
1309        if (defined $last_commit) {
1310                $self->assert_index_clean($last_commit);
1311        }
1312        my $inc = 1000;
1313        my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
1314        my $err_handler = $SVN::Error::handler;
1315        my $err;
1316        $SVN::Error::handler = sub { ($err) = @_; skip_unknown_revs($err); } ;
1317        while (1) {
1318                my @revs;
1319                $self->ra->get_log([$self->{path}], $min, $max, 0, 1, 1, sub {
1320                        my ($paths, $rev, $author, $date, $log) = @_;
1321                        push @revs, [ $paths, $rev ] });
1322                if (! @revs && $err) {
1323                        print STDERR "Branch probably deleted:\n  ",
1324                                     $err->expanded_message,
1325                                     "\nWill attempt to follow revisions ",
1326                                     "committed before the deletion\n";
1327                        @revs = map { [ undef, $_ ] } ($min .. $max);
1328                }
1329                foreach (@revs) {
1330                        if (my $log_entry = $self->do_fetch(@$_)) {
1331                                $self->do_git_commit($log_entry, @parents);
1332                        }
1333                }
1334                last if $max >= $head || $err;
1335                $min = $max + 1;
1336                $max += $inc;
1337                $max = $head if ($max > $head);
1338        }
1339        $SVN::Error::handler = $err_handler;
1340}
1341
1342sub set_tree_cb {
1343        my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
1344        # TODO: enable and test optimized commits:
1345        if (0 && $rev == ($self->{last_rev} + 1)) {
1346                $log_entry->{revision} = $rev;
1347                $log_entry->{author} = $author;
1348                $self->do_git_commit($log_entry, "$rev=$tree");
1349        } else {
1350                $self->fetch("$rev=$tree");
1351        }
1352}
1353
1354sub set_tree {
1355        my ($self, $tree) = (shift, shift);
1356        my $log_entry = ::get_commit_entry($tree);
1357        unless ($self->{last_rev}) {
1358                fatal("Must have an existing revision to commit\n");
1359        }
1360        my $pool = SVN::Pool->new;
1361        my $ed = SVN::Git::Editor->new({ r => $self->{last_rev},
1362                                         ra => $self->ra->dup,
1363                                         svn_path => $self->ra->{svn_path}
1364                                       },
1365                                       $self->ra->get_commit_editor(
1366                                         $log_entry->{log}, sub {
1367                                           $self->set_tree_cb($log_entry,
1368                                                              $tree, @_);
1369                                       }),
1370                                       $pool);
1371        my $mods = $ed->apply_diff($self->{last_commit}, $tree);
1372        if (@$mods == 0) {
1373                print "No changes\nr$self->{last_rev} = $tree\n";
1374        }
1375        $pool->clear;
1376}
1377
1378sub skip_unknown_revs {
1379        my ($err) = @_;
1380        my $errno = $err->apr_err();
1381        # Maybe the branch we're tracking didn't
1382        # exist when the repo started, so it's
1383        # not an error if it doesn't, just continue
1384        #
1385        # Wonderfully consistent library, eh?
1386        # 160013 - svn:// and file://
1387        # 175002 - http(s)://
1388        # 175007 - http(s):// (this repo required authorization, too...)
1389        #   More codes may be discovered later...
1390        if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
1391                return;
1392        }
1393        croak "Error from SVN, ($errno): ", $err->expanded_message,"\n";
1394}
1395
1396# rev_db:
1397# Tie::File seems to be prone to offset errors if revisions get sparse,
1398# it's not that fast, either.  Tie::File is also not in Perl 5.6.  So
1399# one of my favorite modules is out :<  Next up would be one of the DBM
1400# modules, but I'm not sure which is most portable...  So I'll just
1401# go with something that's plain-text, but still capable of
1402# being randomly accessed.  So here's my ultra-simple fixed-width
1403# database.  All records are 40 characters + "\n", so it's easy to seek
1404# to a revision: (41 * rev) is the byte offset.
1405# A record of 40 0s denotes an empty revision.
1406# And yes, it's still pretty fast (faster than Tie::File).
1407
1408sub rev_db_set {
1409        my ($self, $rev, $commit) = @_;
1410        length $commit == 40 or croak "arg3 must be a full SHA1 hexsum\n";
1411        open my $fh, '+<', $self->{db_path} or croak $!;
1412        my $offset = $rev * 41;
1413        # assume that append is the common case:
1414        seek $fh, 0, 2 or croak $!;
1415        my $pos = tell $fh;
1416        if ($pos < $offset) {
1417                print $fh (('0' x 40),"\n") x (($offset - $pos) / 41)
1418                  or croak $!;
1419        }
1420        seek $fh, $offset, 0 or croak $!;
1421        print $fh $commit,"\n" or croak $!;
1422        close $fh or croak $!;
1423}
1424
1425sub rev_db_get {
1426        my ($self, $rev) = @_;
1427        my $ret;
1428        my $offset = $rev * 41;
1429        open my $fh, '<', $self->{db_path} or croak $!;
1430        if (seek $fh, $offset, 0) {
1431                $ret = readline $fh;
1432                if (defined $ret) {
1433                        chomp $ret;
1434                        $ret = undef if ($ret =~ /^0{40}$/);
1435                }
1436        }
1437        close $fh or croak $!;
1438        $ret;
1439}
1440
1441sub find_rev_before {
1442        my ($self, $rev, $eq_ok) = @_;
1443        --$rev unless $eq_ok;
1444        while ($rev > 0) {
1445                if (my $c = $self->rev_db_get($rev)) {
1446                        return ($rev, $c);
1447                }
1448                --$rev;
1449        }
1450        return (undef, undef);
1451}
1452
1453sub _new {
1454        my ($class, $repo_id, $ref_id, $path) = @_;
1455        unless (defined $repo_id && length $repo_id) {
1456                $repo_id = $Git::SVN::default_repo_id;
1457        }
1458        unless (defined $ref_id && length $ref_id) {
1459                $_[2] = $ref_id = $Git::SVN::default_ref_id;
1460        }
1461        $_[1] = $repo_id = sanitize_remote_name($repo_id);
1462        my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
1463        $_[3] = $path = '' unless (defined $path);
1464        mkpath([$dir]);
1465        unless (-f "$dir/.rev_db") {
1466                open my $fh, '>>', "$dir/.rev_db" or croak $!;
1467                close $fh or croak $!;
1468        }
1469        bless { ref_id => $ref_id, dir => $dir, index => "$dir/index",
1470                path => $path,
1471                db_path => "$dir/.rev_db", repo_id => $repo_id }, $class;
1472}
1473
1474sub uri_encode {
1475        my ($f) = @_;
1476        $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
1477        $f
1478}
1479
1480package Git::SVN::Prompt;
1481use strict;
1482use warnings;
1483require SVN::Core;
1484use vars qw/$_no_auth_cache $_username/;
1485
1486sub simple {
1487        my ($cred, $realm, $default_username, $may_save, $pool) = @_;
1488        $may_save = undef if $_no_auth_cache;
1489        $default_username = $_username if defined $_username;
1490        if (defined $default_username && length $default_username) {
1491                if (defined $realm && length $realm) {
1492                        print STDERR "Authentication realm: $realm\n";
1493                        STDERR->flush;
1494                }
1495                $cred->username($default_username);
1496        } else {
1497                username($cred, $realm, $may_save, $pool);
1498        }
1499        $cred->password(_read_password("Password for '" .
1500                                       $cred->username . "': ", $realm));
1501        $cred->may_save($may_save);
1502        $SVN::_Core::SVN_NO_ERROR;
1503}
1504
1505sub ssl_server_trust {
1506        my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
1507        $may_save = undef if $_no_auth_cache;
1508        print STDERR "Error validating server certificate for '$realm':\n";
1509        if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
1510                print STDERR " - The certificate is not issued by a trusted ",
1511                      "authority. Use the\n",
1512                      "   fingerprint to validate the certificate manually!\n";
1513        }
1514        if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
1515                print STDERR " - The certificate hostname does not match.\n";
1516        }
1517        if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
1518                print STDERR " - The certificate is not yet valid.\n";
1519        }
1520        if ($failures & $SVN::Auth::SSL::EXPIRED) {
1521                print STDERR " - The certificate has expired.\n";
1522        }
1523        if ($failures & $SVN::Auth::SSL::OTHER) {
1524                print STDERR " - The certificate has an unknown error.\n";
1525        }
1526        printf STDERR
1527                "Certificate information:\n".
1528                " - Hostname: %s\n".
1529                " - Valid: from %s until %s\n".
1530                " - Issuer: %s\n".
1531                " - Fingerprint: %s\n",
1532                map $cert_info->$_, qw(hostname valid_from valid_until
1533                                       issuer_dname fingerprint);
1534        my $choice;
1535prompt:
1536        print STDERR $may_save ?
1537              "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
1538              "(R)eject or accept (t)emporarily? ";
1539        STDERR->flush;
1540        $choice = lc(substr(<STDIN> || 'R', 0, 1));
1541        if ($choice =~ /^t$/i) {
1542                $cred->may_save(undef);
1543        } elsif ($choice =~ /^r$/i) {
1544                return -1;
1545        } elsif ($may_save && $choice =~ /^p$/i) {
1546                $cred->may_save($may_save);
1547        } else {
1548                goto prompt;
1549        }
1550        $cred->accepted_failures($failures);
1551        $SVN::_Core::SVN_NO_ERROR;
1552}
1553
1554sub ssl_client_cert {
1555        my ($cred, $realm, $may_save, $pool) = @_;
1556        $may_save = undef if $_no_auth_cache;
1557        print STDERR "Client certificate filename: ";
1558        STDERR->flush;
1559        chomp(my $filename = <STDIN>);
1560        $cred->cert_file($filename);
1561        $cred->may_save($may_save);
1562        $SVN::_Core::SVN_NO_ERROR;
1563}
1564
1565sub ssl_client_cert_pw {
1566        my ($cred, $realm, $may_save, $pool) = @_;
1567        $may_save = undef if $_no_auth_cache;
1568        $cred->password(_read_password("Password: ", $realm));
1569        $cred->may_save($may_save);
1570        $SVN::_Core::SVN_NO_ERROR;
1571}
1572
1573sub username {
1574        my ($cred, $realm, $may_save, $pool) = @_;
1575        $may_save = undef if $_no_auth_cache;
1576        if (defined $realm && length $realm) {
1577                print STDERR "Authentication realm: $realm\n";
1578        }
1579        my $username;
1580        if (defined $_username) {
1581                $username = $_username;
1582        } else {
1583                print STDERR "Username: ";
1584                STDERR->flush;
1585                chomp($username = <STDIN>);
1586        }
1587        $cred->username($username);
1588        $cred->may_save($may_save);
1589        $SVN::_Core::SVN_NO_ERROR;
1590}
1591
1592sub _read_password {
1593        my ($prompt, $realm) = @_;
1594        print STDERR $prompt;
1595        STDERR->flush;
1596        require Term::ReadKey;
1597        Term::ReadKey::ReadMode('noecho');
1598        my $password = '';
1599        while (defined(my $key = Term::ReadKey::ReadKey(0))) {
1600                last if $key =~ /[\012\015]/; # \n\r
1601                $password .= $key;
1602        }
1603        Term::ReadKey::ReadMode('restore');
1604        print STDERR "\n";
1605        STDERR->flush;
1606        $password;
1607}
1608
1609package main;
1610
1611sub uri_encode {
1612        my ($f) = @_;
1613        $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
1614        $f
1615}
1616
1617sub uri_decode {
1618        my ($f) = @_;
1619        $f =~ tr/+/ /;
1620        $f =~ s/%([A-F0-9]{2})/chr hex($1)/ge;
1621        $f
1622}
1623
1624{
1625        my $kill_stupid_warnings = $SVN::Node::none.$SVN::Node::file.
1626                                $SVN::Node::dir.$SVN::Node::unknown.
1627                                $SVN::Node::none.$SVN::Node::file.
1628                                $SVN::Node::dir.$SVN::Node::unknown.
1629                                $SVN::Auth::SSL::CNMISMATCH.
1630                                $SVN::Auth::SSL::NOTYETVALID.
1631                                $SVN::Auth::SSL::EXPIRED.
1632                                $SVN::Auth::SSL::UNKNOWNCA.
1633                                $SVN::Auth::SSL::OTHER;
1634}
1635
1636package SVN::Git::Fetcher;
1637use vars qw/@ISA/;
1638use strict;
1639use warnings;
1640use Carp qw/croak/;
1641use IO::File qw//;
1642
1643# file baton members: path, mode_a, mode_b, pool, fh, blob, base
1644sub new {
1645        my ($class, $git_svn) = @_;
1646        my $self = SVN::Delta::Editor->new;
1647        bless $self, $class;
1648        $self->{c} = $git_svn->{last_commit} if exists $git_svn->{last_commit};
1649        $self->{empty} = {};
1650        $self->{dir_prop} = {};
1651        $self->{file_prop} = {};
1652        $self->{absent_dir} = {};
1653        $self->{absent_file} = {};
1654        $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
1655        require Digest::MD5;
1656        $self;
1657}
1658
1659sub set_path_strip {
1660        my ($self, $path) = @_;
1661        $self->{path_strip} = qr/^\Q$path\E\/?/;
1662}
1663
1664sub open_root {
1665        { path => '' };
1666}
1667
1668sub open_directory {
1669        my ($self, $path, $pb, $rev) = @_;
1670        { path => $path };
1671}
1672
1673sub git_path {
1674        my ($self, $path) = @_;
1675        $path =~ s!$self->{path_strip}!! if $self->{path_strip};
1676        $path;
1677}
1678
1679sub delete_entry {
1680        my ($self, $path, $rev, $pb) = @_;
1681
1682        my $gpath = $self->git_path($path);
1683        # remove entire directories.
1684        if (command('ls-tree', $self->{c}, '--', $gpath) =~ /^040000 tree/) {
1685                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
1686                                                     -r --name-only -z/,
1687                                                     $self->{c}, '--', $gpath);
1688                local $/ = "\0";
1689                while (<$ls>) {
1690                        chomp;
1691                        $self->{gii}->remove($_);
1692                        print "\tD\t$_\n" unless $self->{q};
1693                }
1694                print "\tD\t$gpath/\n" unless $self->{q};
1695                command_close_pipe($ls, $ctx);
1696                $self->{empty}->{$path} = 0
1697        } else {
1698                $self->{gii}->remove($gpath);
1699                print "\tD\t$gpath\n" unless $self->{q};
1700        }
1701        undef;
1702}
1703
1704sub open_file {
1705        my ($self, $path, $pb, $rev) = @_;
1706        my $gpath = $self->git_path($path);
1707        my ($mode, $blob) = (command('ls-tree', $self->{c}, '--', $gpath)
1708                             =~ /^(\d{6}) blob ([a-f\d]{40})\t/);
1709        unless (defined $mode && defined $blob) {
1710                die "$path was not found in commit $self->{c} (r$rev)\n";
1711        }
1712        { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
1713          pool => SVN::Pool->new, action => 'M' };
1714}
1715
1716sub add_file {
1717        my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
1718        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
1719        delete $self->{empty}->{$dir};
1720        { path => $path, mode_a => 100644, mode_b => 100644,
1721          pool => SVN::Pool->new, action => 'A' };
1722}
1723
1724sub add_directory {
1725        my ($self, $path, $cp_path, $cp_rev) = @_;
1726        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
1727        delete $self->{empty}->{$dir};
1728        $self->{empty}->{$path} = 1;
1729        { path => $path };
1730}
1731
1732sub change_dir_prop {
1733        my ($self, $db, $prop, $value) = @_;
1734        $self->{dir_prop}->{$db->{path}} ||= {};
1735        $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
1736        undef;
1737}
1738
1739sub absent_directory {
1740        my ($self, $path, $pb) = @_;
1741        $self->{absent_dir}->{$pb->{path}} ||= [];
1742        push @{$self->{absent_dir}->{$pb->{path}}}, $path;
1743        undef;
1744}
1745
1746sub absent_file {
1747        my ($self, $path, $pb) = @_;
1748        $self->{absent_file}->{$pb->{path}} ||= [];
1749        push @{$self->{absent_file}->{$pb->{path}}}, $path;
1750        undef;
1751}
1752
1753sub change_file_prop {
1754        my ($self, $fb, $prop, $value) = @_;
1755        if ($prop eq 'svn:executable') {
1756                if ($fb->{mode_b} != 120000) {
1757                        $fb->{mode_b} = defined $value ? 100755 : 100644;
1758                }
1759        } elsif ($prop eq 'svn:special') {
1760                $fb->{mode_b} = defined $value ? 120000 : 100644;
1761        } else {
1762                $self->{file_prop}->{$fb->{path}} ||= {};
1763                $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
1764        }
1765        undef;
1766}
1767
1768sub apply_textdelta {
1769        my ($self, $fb, $exp) = @_;
1770        my $fh = IO::File->new_tmpfile;
1771        $fh->autoflush(1);
1772        # $fh gets auto-closed() by SVN::TxDelta::apply(),
1773        # (but $base does not,) so dup() it for reading in close_file
1774        open my $dup, '<&', $fh or croak $!;
1775        my $base = IO::File->new_tmpfile;
1776        $base->autoflush(1);
1777        if ($fb->{blob}) {
1778                defined (my $pid = fork) or croak $!;
1779                if (!$pid) {
1780                        open STDOUT, '>&', $base or croak $!;
1781                        print STDOUT 'link ' if ($fb->{mode_a} == 120000);
1782                        exec qw/git-cat-file blob/, $fb->{blob} or croak $!;
1783                }
1784                waitpid $pid, 0;
1785                croak $? if $?;
1786
1787                if (defined $exp) {
1788                        seek $base, 0, 0 or croak $!;
1789                        my $md5 = Digest::MD5->new;
1790                        $md5->addfile($base);
1791                        my $got = $md5->hexdigest;
1792                        die "Checksum mismatch: $fb->{path} $fb->{blob}\n",
1793                            "expected: $exp\n",
1794                            "     got: $got\n" if ($got ne $exp);
1795                }
1796        }
1797        seek $base, 0, 0 or croak $!;
1798        $fb->{fh} = $dup;
1799        $fb->{base} = $base;
1800        [ SVN::TxDelta::apply($base, $fh, undef, $fb->{path}, $fb->{pool}) ];
1801}
1802
1803sub close_file {
1804        my ($self, $fb, $exp) = @_;
1805        my $hash;
1806        my $path = $self->git_path($fb->{path});
1807        if (my $fh = $fb->{fh}) {
1808                seek($fh, 0, 0) or croak $!;
1809                my $md5 = Digest::MD5->new;
1810                $md5->addfile($fh);
1811                my $got = $md5->hexdigest;
1812                die "Checksum mismatch: $path\n",
1813                    "expected: $exp\n    got: $got\n" if ($got ne $exp);
1814                seek($fh, 0, 0) or croak $!;
1815                if ($fb->{mode_b} == 120000) {
1816                        read($fh, my $buf, 5) == 5 or croak $!;
1817                        $buf eq 'link ' or die "$path has mode 120000",
1818                                               "but is not a link\n";
1819                }
1820                defined(my $pid = open my $out,'-|') or die "Can't fork: $!\n";
1821                if (!$pid) {
1822                        open STDIN, '<&', $fh or croak $!;
1823                        exec qw/git-hash-object -w --stdin/ or croak $!;
1824                }
1825                chomp($hash = do { local $/; <$out> });
1826                close $out or croak $!;
1827                close $fh or croak $!;
1828                $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
1829                close $fb->{base} or croak $!;
1830        } else {
1831                $hash = $fb->{blob} or die "no blob information\n";
1832        }
1833        $fb->{pool}->clear;
1834        $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
1835        print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $self->{q};
1836        undef;
1837}
1838
1839sub abort_edit {
1840        my $self = shift;
1841        $self->{nr} = $self->{gii}->{nr};
1842        delete $self->{gii};
1843        $self->SUPER::abort_edit(@_);
1844}
1845
1846sub close_edit {
1847        my $self = shift;
1848        $self->{git_commit_ok} = 1;
1849        $self->{nr} = $self->{gii}->{nr};
1850        delete $self->{gii};
1851        $self->SUPER::close_edit(@_);
1852}
1853
1854package SVN::Git::Editor;
1855use vars qw/@ISA/;
1856use strict;
1857use warnings;
1858use Carp qw/croak/;
1859use IO::File;
1860
1861sub new {
1862        my $class = shift;
1863        my $git_svn = shift;
1864        my $self = SVN::Delta::Editor->new(@_);
1865        bless $self, $class;
1866        foreach (qw/svn_path r ra/) {
1867                die "$_ required!\n" unless (defined $git_svn->{$_});
1868                $self->{$_} = $git_svn->{$_};
1869        }
1870        $self->{pool} = SVN::Pool->new;
1871        $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
1872        $self->{rm} = { };
1873        require Digest::MD5;
1874        return $self;
1875}
1876
1877sub split_path {
1878        return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
1879}
1880
1881sub repo_path {
1882        (defined $_[1] && length $_[1]) ? $_[1] : ''
1883}
1884
1885sub url_path {
1886        my ($self, $path) = @_;
1887        $self->{ra}->{url} . '/' . $self->repo_path($path);
1888}
1889
1890sub rmdirs {
1891        my ($self, $tree_b) = @_;
1892        my $rm = $self->{rm};
1893        delete $rm->{''}; # we never delete the url we're tracking
1894        return unless %$rm;
1895
1896        foreach (keys %$rm) {
1897                my @d = split m#/#, $_;
1898                my $c = shift @d;
1899                $rm->{$c} = 1;
1900                while (@d) {
1901                        $c .= '/' . shift @d;
1902                        $rm->{$c} = 1;
1903                }
1904        }
1905        delete $rm->{$self->{svn_path}};
1906        delete $rm->{''}; # we never delete the url we're tracking
1907        return unless %$rm;
1908
1909        my ($fh, $ctx) = command_output_pipe(
1910                                   qw/ls-tree --name-only -r -z/, $tree_b);
1911        local $/ = "\0";
1912        while (<$fh>) {
1913                chomp;
1914                my @dn = split m#/#, $_;
1915                while (pop @dn) {
1916                        delete $rm->{join '/', @dn};
1917                }
1918                unless (%$rm) {
1919                        close $fh;
1920                        return;
1921                }
1922        }
1923        command_close_pipe($fh, $ctx);
1924
1925        my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
1926        foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
1927                $self->close_directory($bat->{$d}, $p);
1928                my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
1929                print "\tD+\t$d/\n" unless $::_q;
1930                $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
1931                delete $bat->{$d};
1932        }
1933}
1934
1935sub open_or_add_dir {
1936        my ($self, $full_path, $baton) = @_;
1937        my $t = $self->{ra}->check_path($full_path, $self->{r});
1938        if ($t == $SVN::Node::none) {
1939                return $self->add_directory($full_path, $baton,
1940                                                undef, -1, $self->{pool});
1941        } elsif ($t == $SVN::Node::dir) {
1942                return $self->open_directory($full_path, $baton,
1943                                                $self->{r}, $self->{pool});
1944        }
1945        print STDERR "$full_path already exists in repository at ",
1946                "r$self->{r} and it is not a directory (",
1947                ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
1948        exit 1;
1949}
1950
1951sub ensure_path {
1952        my ($self, $path) = @_;
1953        my $bat = $self->{bat};
1954        $path = $self->repo_path($path);
1955        return $bat->{''} unless (length $path);
1956        my @p = split m#/+#, $path;
1957        my $c = shift @p;
1958        $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
1959        while (@p) {
1960                my $c0 = $c;
1961                $c .= '/' . shift @p;
1962                $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
1963        }
1964        return $bat->{$c};
1965}
1966
1967sub A {
1968        my ($self, $m) = @_;
1969        my ($dir, $file) = split_path($m->{file_b});
1970        my $pbat = $self->ensure_path($dir);
1971        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
1972                                        undef, -1);
1973        print "\tA\t$m->{file_b}\n" unless $::_q;
1974        $self->chg_file($fbat, $m);
1975        $self->close_file($fbat,undef,$self->{pool});
1976}
1977
1978sub C {
1979        my ($self, $m) = @_;
1980        my ($dir, $file) = split_path($m->{file_b});
1981        my $pbat = $self->ensure_path($dir);
1982        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
1983                                $self->url_path($m->{file_a}), $self->{r});
1984        print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
1985        $self->chg_file($fbat, $m);
1986        $self->close_file($fbat,undef,$self->{pool});
1987}
1988
1989sub delete_entry {
1990        my ($self, $path, $pbat) = @_;
1991        my $rpath = $self->repo_path($path);
1992        my ($dir, $file) = split_path($rpath);
1993        $self->{rm}->{$dir} = 1;
1994        $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
1995}
1996
1997sub R {
1998        my ($self, $m) = @_;
1999        my ($dir, $file) = split_path($m->{file_b});
2000        my $pbat = $self->ensure_path($dir);
2001        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2002                                $self->url_path($m->{file_a}), $self->{r});
2003        print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
2004        $self->chg_file($fbat, $m);
2005        $self->close_file($fbat,undef,$self->{pool});
2006
2007        ($dir, $file) = split_path($m->{file_a});
2008        $pbat = $self->ensure_path($dir);
2009        $self->delete_entry($m->{file_a}, $pbat);
2010}
2011
2012sub M {
2013        my ($self, $m) = @_;
2014        my ($dir, $file) = split_path($m->{file_b});
2015        my $pbat = $self->ensure_path($dir);
2016        my $fbat = $self->open_file($self->repo_path($m->{file_b}),
2017                                $pbat,$self->{r},$self->{pool});
2018        print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
2019        $self->chg_file($fbat, $m);
2020        $self->close_file($fbat,undef,$self->{pool});
2021}
2022
2023sub T { shift->M(@_) }
2024
2025sub change_file_prop {
2026        my ($self, $fbat, $pname, $pval) = @_;
2027        $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
2028}
2029
2030sub chg_file {
2031        my ($self, $fbat, $m) = @_;
2032        if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
2033                $self->change_file_prop($fbat,'svn:executable','*');
2034        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
2035                $self->change_file_prop($fbat,'svn:executable',undef);
2036        }
2037        my $fh = IO::File->new_tmpfile or croak $!;
2038        if ($m->{mode_b} =~ /^120/) {
2039                print $fh 'link ' or croak $!;
2040                $self->change_file_prop($fbat,'svn:special','*');
2041        } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
2042                $self->change_file_prop($fbat,'svn:special',undef);
2043        }
2044        defined(my $pid = fork) or croak $!;
2045        if (!$pid) {
2046                open STDOUT, '>&', $fh or croak $!;
2047                exec qw/git-cat-file blob/, $m->{sha1_b} or croak $!;
2048        }
2049        waitpid $pid, 0;
2050        croak $? if $?;
2051        $fh->flush == 0 or croak $!;
2052        seek $fh, 0, 0 or croak $!;
2053
2054        my $md5 = Digest::MD5->new;
2055        $md5->addfile($fh) or croak $!;
2056        seek $fh, 0, 0 or croak $!;
2057
2058        my $exp = $md5->hexdigest;
2059        my $pool = SVN::Pool->new;
2060        my $atd = $self->apply_textdelta($fbat, undef, $pool);
2061        my $got = SVN::TxDelta::send_stream($fh, @$atd, $pool);
2062        die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
2063        $pool->clear;
2064
2065        close $fh or croak $!;
2066}
2067
2068sub D {
2069        my ($self, $m) = @_;
2070        my ($dir, $file) = split_path($m->{file_b});
2071        my $pbat = $self->ensure_path($dir);
2072        print "\tD\t$m->{file_b}\n" unless $::_q;
2073        $self->delete_entry($m->{file_b}, $pbat);
2074}
2075
2076sub close_edit {
2077        my ($self) = @_;
2078        my ($p,$bat) = ($self->{pool}, $self->{bat});
2079        foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
2080                $self->close_directory($bat->{$_}, $p);
2081        }
2082        $self->SUPER::close_edit($p);
2083        $p->clear;
2084}
2085
2086sub abort_edit {
2087        my ($self) = @_;
2088        $self->SUPER::abort_edit($self->{pool});
2089        $self->{pool}->clear;
2090}
2091
2092# this drives the editor
2093sub apply_diff {
2094        my ($self, $tree_a, $tree_b) = @_;
2095        my @diff_tree = qw(diff-tree -z -r);
2096        if ($::_cp_similarity) {
2097                push @diff_tree, "-C$::_cp_similarity";
2098        } else {
2099                push @diff_tree, '-C';
2100        }
2101        push @diff_tree, '--find-copies-harder' if $::_find_copies_harder;
2102        push @diff_tree, "-l$::_l" if defined $::_l;
2103        push @diff_tree, $tree_a, $tree_b;
2104        my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
2105        my $nl = $/;
2106        local $/ = "\0";
2107        my $state = 'meta';
2108        my @mods;
2109        while (<$diff_fh>) {
2110                chomp $_; # this gets rid of the trailing "\0"
2111                if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
2112                                        $::sha1\s($::sha1)\s
2113                                        ([MTCRAD])\d*$/xo) {
2114                        push @mods, {   mode_a => $1, mode_b => $2,
2115                                        sha1_b => $3, chg => $4 };
2116                        if ($4 =~ /^(?:C|R)$/) {
2117                                $state = 'file_a';
2118                        } else {
2119                                $state = 'file_b';
2120                        }
2121                } elsif ($state eq 'file_a') {
2122                        my $x = $mods[$#mods] or croak "Empty array\n";
2123                        if ($x->{chg} !~ /^(?:C|R)$/) {
2124                                croak "Error parsing $_, $x->{chg}\n";
2125                        }
2126                        $x->{file_a} = $_;
2127                        $state = 'file_b';
2128                } elsif ($state eq 'file_b') {
2129                        my $x = $mods[$#mods] or croak "Empty array\n";
2130                        if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
2131                                croak "Error parsing $_, $x->{chg}\n";
2132                        }
2133                        if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
2134                                croak "Error parsing $_, $x->{chg}\n";
2135                        }
2136                        $x->{file_b} = $_;
2137                        $state = 'meta';
2138                } else {
2139                        croak "Error parsing $_\n";
2140                }
2141        }
2142        command_close_pipe($diff_fh, $ctx);
2143        $/ = $nl;
2144
2145        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
2146        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @mods) {
2147                my $f = $m->{chg};
2148                if (defined $o{$f}) {
2149                        $self->$f($m);
2150                } else {
2151                        fatal("Invalid change type: $f\n");
2152                }
2153        }
2154        $self->rmdirs($tree_b) if $::_rmdir;
2155        if (@mods == 0) {
2156                $self->abort_edit;
2157        } else {
2158                $self->close_edit;
2159        }
2160        \@mods;
2161}
2162
2163package Git::SVN::Ra;
2164use vars qw/@ISA $config_dir/;
2165use strict;
2166use warnings;
2167my ($can_do_switch);
2168my %RA;
2169
2170BEGIN {
2171        # enforce temporary pool usage for some simple functions
2172        my $e;
2173        foreach (qw/get_latest_revnum rev_proplist get_file
2174                    check_path get_dir get_uuid get_repos_root/) {
2175                $e .= "sub $_ {
2176                        my \$self = shift;
2177                        my \$pool = SVN::Pool->new;
2178                        my \@ret = \$self->SUPER::$_(\@_,\$pool);
2179                        \$pool->clear;
2180                        wantarray ? \@ret : \$ret[0]; }\n";
2181        }
2182        eval $e;
2183}
2184
2185sub new {
2186        my ($class, $url) = @_;
2187        $url =~ s!/+$!!;
2188        return $RA{$url} if $RA{$url};
2189
2190        SVN::_Core::svn_config_ensure($config_dir, undef);
2191        my ($baton, $callbacks) = SVN::Core::auth_open_helper([
2192            SVN::Client::get_simple_provider(),
2193            SVN::Client::get_ssl_server_trust_file_provider(),
2194            SVN::Client::get_simple_prompt_provider(
2195              \&Git::SVN::Prompt::simple, 2),
2196            SVN::Client::get_ssl_client_cert_prompt_provider(
2197              \&Git::SVN::Prompt::ssl_client_cert, 2),
2198            SVN::Client::get_ssl_client_cert_pw_prompt_provider(
2199              \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
2200            SVN::Client::get_username_provider(),
2201            SVN::Client::get_ssl_server_trust_prompt_provider(
2202              \&Git::SVN::Prompt::ssl_server_trust),
2203            SVN::Client::get_username_prompt_provider(
2204              \&Git::SVN::Prompt::username, 2),
2205          ]);
2206        my $config = SVN::Core::config_get_config($config_dir);
2207        my $self = SVN::Ra->new(url => $url, auth => $baton,
2208                              config => $config,
2209                              pool => SVN::Pool->new,
2210                              auth_provider_callbacks => $callbacks);
2211        $self->{svn_path} = $url;
2212        $self->{repos_root} = $self->get_repos_root;
2213        $self->{svn_path} =~ s#^\Q$self->{repos_root}\E/*##;
2214        $RA{$url} = bless $self, $class;
2215}
2216
2217sub DESTROY {
2218        # do not call the real DESTROY since we store ourselves in %RA
2219}
2220
2221sub dup {
2222        my ($self) = @_;
2223        my $dup = SVN::Ra->new(pool => SVN::Pool->new,
2224                                map { $_ => $self->{$_} } qw/config url
2225                     auth auth_provider_callbacks repos_root svn_path/);
2226        bless $dup, ref $self;
2227}
2228
2229sub get_log {
2230        my ($self, @args) = @_;
2231        my $pool = SVN::Pool->new;
2232        splice(@args, 3, 1) if ($SVN::Core::VERSION le '1.2.0');
2233        my $ret = $self->SUPER::get_log(@args, $pool);
2234        $pool->clear;
2235        $ret;
2236}
2237
2238sub get_commit_editor {
2239        my ($self, $log, $cb, $pool) = @_;
2240        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
2241        $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
2242}
2243
2244sub uuid {
2245        my ($self) = @_;
2246        $self->{uuid} ||= $self->get_uuid;
2247}
2248
2249sub gs_do_update {
2250        my ($self, $rev_a, $rev_b, $path, $recurse, $editor) = @_;
2251        my $pool = SVN::Pool->new;
2252        $editor->set_path_strip($path);
2253        my $reporter = $self->do_update($rev_b, $path, $recurse,
2254                                        $editor, $pool);
2255        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
2256        my $new = ($rev_a == $rev_b);
2257        $reporter->set_path('', $rev_a, $new, @lock, $pool);
2258        $reporter->finish_report($pool);
2259        $pool->clear;
2260        $editor->{git_commit_ok};
2261}
2262
2263sub gs_do_switch {
2264        my ($self, $rev_a, $rev_b, $path, $recurse, $url_b, $editor) = @_;
2265        my $pool = SVN::Pool->new;
2266        $editor->set_path_strip($path);
2267        my $reporter = $self->do_switch($rev_b, $path, $recurse,
2268                                        $url_b, $editor, $pool);
2269        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
2270        $reporter->set_path('', $rev_a, 0, @lock, $pool);
2271        $reporter->finish_report($pool);
2272        $pool->clear;
2273        $editor->{git_commit_ok};
2274}
2275
2276sub minimize_url {
2277        my ($self) = @_;
2278        return $self->{url} if ($self->{url} eq $self->{repos_root});
2279        my $url = $self->{repos_root};
2280        my @components = split(m!/!, $self->{svn_path});
2281        my $c = '';
2282        do {
2283                $url .= "/$c" if length $c;
2284                eval { (ref $self)->new($url)->get_latest_revnum };
2285        } while ($@ && ($c = shift @components));
2286        $url;
2287}
2288
2289sub can_do_switch {
2290        my $self = shift;
2291        unless (defined $can_do_switch) {
2292                my $pool = SVN::Pool->new;
2293                my $rep = eval {
2294                        $self->do_switch(1, '', 0, $self->{url},
2295                                         SVN::Delta::Editor->new, $pool);
2296                };
2297                if ($@) {
2298                        $can_do_switch = 0;
2299                } else {
2300                        $rep->abort_report($pool);
2301                        $can_do_switch = 1;
2302                }
2303                $pool->clear;
2304        }
2305        $can_do_switch;
2306}
2307
2308package Git::SVN::Log;
2309use strict;
2310use warnings;
2311use POSIX qw/strftime/;
2312use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
2313            %rusers $show_commit $incremental/;
2314my $l_fmt;
2315
2316sub cmt_showable {
2317        my ($c) = @_;
2318        return 1 if defined $c->{r};
2319        if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
2320                                $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
2321                my @log = command(qw/cat-file commit/, $c->{c});
2322                shift @log while ($log[0] ne "\n");
2323                shift @log;
2324                @{$c->{l}} = grep !/^git-svn-id: /, @log;
2325
2326                (undef, $c->{r}, undef) = ::extract_metadata(
2327                                (grep(/^git-svn-id: /, @log))[-1]);
2328        }
2329        return defined $c->{r};
2330}
2331
2332sub log_use_color {
2333        return 1 if $color;
2334        my ($dc, $dcvar);
2335        $dcvar = 'color.diff';
2336        $dc = `git-config --get $dcvar`;
2337        if ($dc eq '') {
2338                # nothing at all; fallback to "diff.color"
2339                $dcvar = 'diff.color';
2340                $dc = `git-config --get $dcvar`;
2341        }
2342        chomp($dc);
2343        if ($dc eq 'auto') {
2344                my $pc;
2345                $pc = `git-config --get color.pager`;
2346                if ($pc eq '') {
2347                        # does not have it -- fallback to pager.color
2348                        $pc = `git-config --bool --get pager.color`;
2349                }
2350                else {
2351                        $pc = `git-config --bool --get color.pager`;
2352                        if ($?) {
2353                                $pc = 'false';
2354                        }
2355                }
2356                chomp($pc);
2357                if (-t *STDOUT || (defined $pager && $pc eq 'true')) {
2358                        return ($ENV{TERM} && $ENV{TERM} ne 'dumb');
2359                }
2360                return 0;
2361        }
2362        return 0 if $dc eq 'never';
2363        return 1 if $dc eq 'always';
2364        chomp($dc = `git-config --bool --get $dcvar`);
2365        return ($dc eq 'true');
2366}
2367
2368sub git_svn_log_cmd {
2369        my ($r_min, $r_max) = @_;
2370        my $gs = Git::SVN->_new;
2371        my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
2372                   $gs->refname);
2373        push @cmd, '-r' unless $non_recursive;
2374        push @cmd, qw/--raw --name-status/ if $verbose;
2375        push @cmd, '--color' if log_use_color();
2376        return @cmd unless defined $r_max;
2377        if ($r_max == $r_min) {
2378                push @cmd, '--max-count=1';
2379                if (my $c = $gs->rev_db_get($r_max)) {
2380                        push @cmd, $c;
2381                }
2382        } else {
2383                my ($c_min, $c_max);
2384                $c_max = $gs->rev_db_get($r_max);
2385                $c_min = $gs->rev_db_get($r_min);
2386                if (defined $c_min && defined $c_max) {
2387                        if ($r_max > $r_max) {
2388                                push @cmd, "$c_min..$c_max";
2389                        } else {
2390                                push @cmd, "$c_max..$c_min";
2391                        }
2392                } elsif ($r_max > $r_min) {
2393                        push @cmd, $c_max;
2394                } else {
2395                        push @cmd, $c_min;
2396                }
2397        }
2398        return @cmd;
2399}
2400
2401# adapted from pager.c
2402sub config_pager {
2403        $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
2404        if (!defined $pager) {
2405                $pager = 'less';
2406        } elsif (length $pager == 0 || $pager eq 'cat') {
2407                $pager = undef;
2408        }
2409}
2410
2411sub run_pager {
2412        return unless -t *STDOUT;
2413        pipe my $rfd, my $wfd or return;
2414        defined(my $pid = fork) or ::fatal "Can't fork: $!\n";
2415        if (!$pid) {
2416                open STDOUT, '>&', $wfd or
2417                                     ::fatal "Can't redirect to stdout: $!\n";
2418                return;
2419        }
2420        open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!\n";
2421        $ENV{LESS} ||= 'FRSX';
2422        exec $pager or ::fatal "Can't run pager: $! ($pager)\n";
2423}
2424
2425sub get_author_info {
2426        my ($dest, $author, $t, $tz) = @_;
2427        $author =~ s/(?:^\s*|\s*$)//g;
2428        $dest->{a_raw} = $author;
2429        my $au;
2430        if ($::_authors) {
2431                $au = $rusers{$author} || undef;
2432        }
2433        if (!$au) {
2434                ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
2435        }
2436        $dest->{t} = $t;
2437        $dest->{tz} = $tz;
2438        $dest->{a} = $au;
2439        # Date::Parse isn't in the standard Perl distro :(
2440        if ($tz =~ s/^\+//) {
2441                $t += ::tz_to_s_offset($tz);
2442        } elsif ($tz =~ s/^\-//) {
2443                $t -= ::tz_to_s_offset($tz);
2444        }
2445        $dest->{t_utc} = $t;
2446}
2447
2448sub process_commit {
2449        my ($c, $r_min, $r_max, $defer) = @_;
2450        if (defined $r_min && defined $r_max) {
2451                if ($r_min == $c->{r} && $r_min == $r_max) {
2452                        show_commit($c);
2453                        return 0;
2454                }
2455                return 1 if $r_min == $r_max;
2456                if ($r_min < $r_max) {
2457                        # we need to reverse the print order
2458                        return 0 if (defined $limit && --$limit < 0);
2459                        push @$defer, $c;
2460                        return 1;
2461                }
2462                if ($r_min != $r_max) {
2463                        return 1 if ($r_min < $c->{r});
2464                        return 1 if ($r_max > $c->{r});
2465                }
2466        }
2467        return 0 if (defined $limit && --$limit < 0);
2468        show_commit($c);
2469        return 1;
2470}
2471
2472sub show_commit {
2473        my $c = shift;
2474        if ($oneline) {
2475                my $x = "\n";
2476                if (my $l = $c->{l}) {
2477                        while ($l->[0] =~ /^\s*$/) { shift @$l }
2478                        $x = $l->[0];
2479                }
2480                $l_fmt ||= 'A' . length($c->{r});
2481                print 'r',pack($l_fmt, $c->{r}),' | ';
2482                print "$c->{c} | " if $show_commit;
2483                print $x;
2484        } else {
2485                show_commit_normal($c);
2486        }
2487}
2488
2489sub show_commit_changed_paths {
2490        my ($c) = @_;
2491        return unless $c->{changed};
2492        print "Changed paths:\n", @{$c->{changed}};
2493}
2494
2495sub show_commit_normal {
2496        my ($c) = @_;
2497        print '-' x72, "\nr$c->{r} | ";
2498        print "$c->{c} | " if $show_commit;
2499        print "$c->{a} | ", strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)",
2500                                 localtime($c->{t_utc})), ' | ';
2501        my $nr_line = 0;
2502
2503        if (my $l = $c->{l}) {
2504                while ($l->[$#$l] eq "\n" && $#$l > 0
2505                                          && $l->[($#$l - 1)] eq "\n") {
2506                        pop @$l;
2507                }
2508                $nr_line = scalar @$l;
2509                if (!$nr_line) {
2510                        print "1 line\n\n\n";
2511                } else {
2512                        if ($nr_line == 1) {
2513                                $nr_line = '1 line';
2514                        } else {
2515                                $nr_line .= ' lines';
2516                        }
2517                        print $nr_line, "\n";
2518                        show_commit_changed_paths($c);
2519                        print "\n";
2520                        print $_ foreach @$l;
2521                }
2522        } else {
2523                print "1 line\n";
2524                show_commit_changed_paths($c);
2525                print "\n";
2526
2527        }
2528        foreach my $x (qw/raw diff/) {
2529                if ($c->{$x}) {
2530                        print "\n";
2531                        print $_ foreach @{$c->{$x}}
2532                }
2533        }
2534}
2535
2536sub cmd_show_log {
2537        my (@args) = @_;
2538        my ($r_min, $r_max);
2539        my $r_last = -1; # prevent dupes
2540        if (defined $TZ) {
2541                $ENV{TZ} = $TZ;
2542        } else {
2543                delete $ENV{TZ};
2544        }
2545        if (defined $::_revision) {
2546                if ($::_revision =~ /^(\d+):(\d+)$/) {
2547                        ($r_min, $r_max) = ($1, $2);
2548                } elsif ($::_revision =~ /^\d+$/) {
2549                        $r_min = $r_max = $::_revision;
2550                } else {
2551                        ::fatal "-r$::_revision is not supported, use ",
2552                                "standard \'git log\' arguments instead\n";
2553                }
2554        }
2555
2556        config_pager();
2557        @args = (git_svn_log_cmd($r_min, $r_max), @args);
2558        my $log = command_output_pipe(@args);
2559        run_pager();
2560        my (@k, $c, $d);
2561        my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
2562        while (<$log>) {
2563                if (/^${esc_color}commit ($::sha1_short)/o) {
2564                        my $cmt = $1;
2565                        if ($c && cmt_showable($c) && $c->{r} != $r_last) {
2566                                $r_last = $c->{r};
2567                                process_commit($c, $r_min, $r_max, \@k) or
2568                                                                goto out;
2569                        }
2570                        $d = undef;
2571                        $c = { c => $cmt };
2572                } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
2573                        get_author_info($c, $1, $2, $3);
2574                } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
2575                        # ignore
2576                } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
2577                        push @{$c->{raw}}, $_;
2578                } elsif (/^${esc_color}[ACRMDT]\t/) {
2579                        # we could add $SVN->{svn_path} here, but that requires
2580                        # remote access at the moment (repo_path_split)...
2581                        s#^(${esc_color})([ACRMDT])\t#$1   $2 #o;
2582                        push @{$c->{changed}}, $_;
2583                } elsif (/^${esc_color}diff /o) {
2584                        $d = 1;
2585                        push @{$c->{diff}}, $_;
2586                } elsif ($d) {
2587                        push @{$c->{diff}}, $_;
2588                } elsif (/^${esc_color}    (git-svn-id:.+)$/o) {
2589                        ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
2590                } elsif (s/^${esc_color}    //o) {
2591                        push @{$c->{l}}, $_;
2592                }
2593        }
2594        if ($c && defined $c->{r} && $c->{r} != $r_last) {
2595                $r_last = $c->{r};
2596                process_commit($c, $r_min, $r_max, \@k);
2597        }
2598        if (@k) {
2599                my $swap = $r_max;
2600                $r_max = $r_min;
2601                $r_min = $swap;
2602                process_commit($_, $r_min, $r_max) foreach reverse @k;
2603        }
2604out:
2605        close $log;
2606        print '-' x72,"\n" unless $incremental || $oneline;
2607}
2608
2609package Git::SVN::Migration;
2610# these version numbers do NOT correspond to actual version numbers
2611# of git nor git-svn.  They are just relative.
2612#
2613# v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
2614#
2615# v1 layout: .git/$id/info/url, refs/remotes/$id
2616#
2617# v2 layout: .git/svn/$id/info/url, refs/remotes/$id
2618#
2619# v3 layout: .git/svn/$id, refs/remotes/$id
2620#            - info/url may remain for backwards compatibility
2621#            - this is what we migrate up to this layout automatically,
2622#            - this will be used by git svn init on single branches
2623#
2624# v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
2625#            - this is only created for newly multi-init-ed
2626#              repositories.  Similar in spirit to the
2627#              --use-separate-remotes option in git-clone (now default)
2628#            - we do not automatically migrate to this (following
2629#              the example set by core git)
2630use strict;
2631use warnings;
2632use Carp qw/croak/;
2633use File::Path qw/mkpath/;
2634use File::Basename qw/dirname basename/;
2635use vars qw/$_minimize/;
2636
2637sub migrate_from_v0 {
2638        my $git_dir = $ENV{GIT_DIR};
2639        return undef unless -d $git_dir;
2640        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
2641        my $migrated = 0;
2642        while (<$fh>) {
2643                chomp;
2644                my ($id, $orig_ref) = ($_, $_);
2645                next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
2646                next unless -f "$git_dir/$id/info/url";
2647                my $new_ref = "refs/remotes/$id";
2648                if (::verify_ref("$new_ref^0")) {
2649                        print STDERR "W: $orig_ref is probably an old ",
2650                                     "branch used by an ancient version of ",
2651                                     "git-svn.\n",
2652                                     "However, $new_ref also exists.\n",
2653                                     "We will not be able ",
2654                                     "to use this branch until this ",
2655                                     "ambiguity is resolved.\n";
2656                        next;
2657                }
2658                print STDERR "Migrating from v0 layout...\n" if !$migrated;
2659                print STDERR "Renaming ref: $orig_ref => $new_ref\n";
2660                command_noisy('update-ref', $new_ref, $orig_ref);
2661                command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
2662                $migrated++;
2663        }
2664        command_close_pipe($fh, $ctx);
2665        print STDERR "Done migrating from v0 layout...\n" if $migrated;
2666        $migrated;
2667}
2668
2669sub migrate_from_v1 {
2670        my $git_dir = $ENV{GIT_DIR};
2671        my $migrated = 0;
2672        return $migrated unless -d $git_dir;
2673        my $svn_dir = "$git_dir/svn";
2674
2675        # just in case somebody used 'svn' as their $id at some point...
2676        return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
2677
2678        print STDERR "Migrating from a git-svn v1 layout...\n";
2679        mkpath([$svn_dir]);
2680        print STDERR "Data from a previous version of git-svn exists, but\n\t",
2681                     "$svn_dir\n\t(required for this version ",
2682                     "($::VERSION) of git-svn) does not. exist\n";
2683        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
2684        while (<$fh>) {
2685                my $x = $_;
2686                next unless $x =~ s#^refs/remotes/##;
2687                chomp $x;
2688                next unless -f "$git_dir/$x/info/url";
2689                my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
2690                next unless $u;
2691                my $dn = dirname("$git_dir/svn/$x");
2692                mkpath([$dn]) unless -d $dn;
2693                if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
2694                        mkpath(["$git_dir/svn/svn"]);
2695                        print STDERR " - $git_dir/$x/info => ",
2696                                        "$git_dir/svn/$x/info\n";
2697                        rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
2698                               croak "$!: $x";
2699                        # don't worry too much about these, they probably
2700                        # don't exist with repos this old (save for index,
2701                        # and we can easily regenerate that)
2702                        foreach my $f (qw/unhandled.log index .rev_db/) {
2703                                rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
2704                        }
2705                } else {
2706                        print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
2707                        rename "$git_dir/$x", "$git_dir/svn/$x" or
2708                               croak "$!: $x";
2709                }
2710                $migrated++;
2711        }
2712        command_close_pipe($fh, $ctx);
2713        print STDERR "Done migrating from a git-svn v1 layout\n";
2714        $migrated;
2715}
2716
2717sub read_old_urls {
2718        my ($l_map, $pfx, $path) = @_;
2719        my @dir;
2720        foreach (<$path/*>) {
2721                if (-r "$_/info/url") {
2722                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
2723                        my $ref_id = $pfx . basename $_;
2724                        my $url = ::file_to_s("$_/info/url");
2725                        $l_map->{$ref_id} = $url;
2726                } elsif (-d $_) {
2727                        push @dir, $_;
2728                }
2729        }
2730        foreach (@dir) {
2731                my $x = $_;
2732                $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
2733                read_old_urls($l_map, $x, $_);
2734        }
2735}
2736
2737sub migrate_from_v2 {
2738        my @cfg = command(qw/config -l/);
2739        return if grep /^svn-remote\..+\.url=/, @cfg;
2740        my %l_map;
2741        read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
2742        my $migrated = 0;
2743
2744        foreach my $ref_id (sort keys %l_map) {
2745                Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
2746                $migrated++;
2747        }
2748        $migrated;
2749}
2750
2751sub minimize_connections {
2752        my $r = Git::SVN::read_all_remotes();
2753        my $new_urls = {};
2754        my $root_repos = {};
2755        foreach my $repo_id (keys %$r) {
2756                my $url = $r->{$repo_id}->{url} or next;
2757                my $fetch = $r->{$repo_id}->{fetch} or next;
2758                my $ra = Git::SVN::Ra->new($url);
2759
2760                # skip existing cases where we already connect to the root
2761                if (($ra->{url} eq $ra->{repos_root}) ||
2762                    (Git::SVN::sanitize_remote_name($ra->{repos_root}) eq
2763                     $repo_id)) {
2764                        $root_repos->{$ra->{url}} = $repo_id;
2765                        next;
2766                }
2767
2768                my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
2769                my $root_path = $ra->{url};
2770                $root_path =~ s#^\Q$ra->{repos_root}\E/*##;
2771                foreach my $path (keys %$fetch) {
2772                        my $ref_id = $fetch->{$path};
2773                        my $gs = Git::SVN->new($ref_id, $repo_id, $path);
2774
2775                        # make sure we can read when connecting to
2776                        # a higher level of a repository
2777                        my ($last_rev, undef) = $gs->last_rev_commit;
2778                        if (!defined $last_rev) {
2779                                $last_rev = eval {
2780                                        $root_ra->get_latest_revnum;
2781                                };
2782                                next if $@;
2783                        }
2784                        my $new = $root_path;
2785                        $new .= length $path ? "/$path" : '';
2786                        eval {
2787                                $root_ra->get_log([$new], $last_rev, $last_rev,
2788                                                  0, 0, 1, sub { });
2789                        };
2790                        next if $@;
2791                        $new_urls->{$ra->{repos_root}}->{$new} =
2792                                { ref_id => $ref_id,
2793                                  old_repo_id => $repo_id,
2794                                  old_path => $path };
2795                }
2796        }
2797
2798        my @emptied;
2799        foreach my $url (keys %$new_urls) {
2800                # see if we can re-use an existing [svn-remote "repo_id"]
2801                # instead of creating a(n ugly) new section:
2802                my $repo_id = $root_repos->{$url} ||
2803                              Git::SVN::sanitize_remote_name($url);
2804
2805                my $fetch = $new_urls->{$url};
2806                foreach my $path (keys %$fetch) {
2807                        my $x = $fetch->{$path};
2808                        Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
2809                        my $pfx = "svn-remote.$x->{old_repo_id}";
2810
2811                        my $old_fetch = quotemeta("$x->{old_path}:".
2812                                                  "refs/remotes/$x->{ref_id}");
2813                        command_noisy(qw/config --unset/,
2814                                      "$pfx.fetch", '^'. $old_fetch . '$');
2815                        delete $r->{$x->{old_repo_id}}->
2816                               {fetch}->{$x->{old_path}};
2817                        if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
2818                                command_noisy(qw/config --unset/,
2819                                              "$pfx.url");
2820                                push @emptied, $x->{old_repo_id}
2821                        }
2822                }
2823        }
2824        if (@emptied) {
2825                my $file = $ENV{GIT_CONFIG} || $ENV{GIT_CONFIG_LOCAL} ||
2826                           "$ENV{GIT_DIR}/config";
2827                print STDERR <<EOF;
2828The following [svn-remote] sections in your config file ($file) are empty
2829and can be safely removed:
2830EOF
2831                print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
2832        }
2833}
2834
2835sub migration_check {
2836        migrate_from_v0();
2837        migrate_from_v1();
2838        migrate_from_v2();
2839        minimize_connections() if $_minimize;
2840}
2841
2842package Git::IndexInfo;
2843use strict;
2844use warnings;
2845use Git qw/command_input_pipe command_close_pipe/;
2846
2847sub new {
2848        my ($class) = @_;
2849        my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
2850        bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
2851}
2852
2853sub remove {
2854        my ($self, $path) = @_;
2855        if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
2856                return ++$self->{nr};
2857        }
2858        undef;
2859}
2860
2861sub update {
2862        my ($self, $mode, $hash, $path) = @_;
2863        if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
2864                return ++$self->{nr};
2865        }
2866        undef;
2867}
2868
2869sub DESTROY {
2870        my ($self) = @_;
2871        command_close_pipe($self->{gui}, $self->{ctx});
2872}
2873
2874__END__
2875
2876Data structures:
2877
2878$log_entry hashref as returned by libsvn_log_entry()
2879{
2880        log => 'whitespace-formatted log entry
2881',                                              # trailing newline is preserved
2882        revision => '8',                        # integer
2883        date => '2004-02-24T17:01:44.108345Z',  # commit date
2884        author => 'committer name'
2885};
2886
2887@mods = array of diff-index line hashes, each element represents one line
2888        of diff-index output
2889
2890diff-index line ($m hash)
2891{
2892        mode_a => first column of diff-index output, no leading ':',
2893        mode_b => second column of diff-index output,
2894        sha1_b => sha1sum of the final blob,
2895        chg => change type [MCRADT],
2896        file_a => original file name of a file (iff chg is 'C' or 'R')
2897        file_b => new/current file name of a file (any chg)
2898}
2899;
2900
2901# retval of read_url_paths{,_all}();
2902$l_map = {
2903        # repository root url
2904        'https://svn.musicpd.org' => {
2905                # repository path               # GIT_SVN_ID
2906                'mpd/trunk'             =>      'trunk',
2907                'mpd/tags/0.11.5'       =>      'tags/0.11.5',
2908        },
2909}
2910
2911Notes:
2912        I don't trust the each() function on unless I created %hash myself
2913        because the internal iterator may not have started at base.