git-svn.perlon commit git-svn: fix segfaults from accessing svn_log_changed_path_t (3ebe8df)
   1#!/usr/bin/env perl
   2# Copyright (C) 2006, Eric Wong <normalperson@yhbt.net>
   3# License: GPL v2 or later
   4use warnings;
   5use strict;
   6use vars qw/    $AUTHOR $VERSION
   7                $SVN_URL
   8                $GIT_SVN_INDEX $GIT_SVN
   9                $GIT_DIR $GIT_SVN_DIR $REVDB
  10                $_follow_parent $sha1 $sha1_short $_revision
  11                $_cp_remote $_upgrade $_rmdir $_q $_cp_similarity
  12                $_find_copies_harder $_l $_authors %users/;
  13$AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
  14$VERSION = '@@GIT_VERSION@@';
  15
  16$ENV{GIT_DIR} ||= '.git';
  17$Git::SVN::default_repo_id = 'git-svn';
  18$Git::SVN::default_ref_id = $ENV{GIT_SVN_ID} || 'git-svn';
  19
  20my $LC_ALL = $ENV{LC_ALL};
  21$Git::SVN::Log::TZ = $ENV{TZ};
  22# make sure the svn binary gives consistent output between locales and TZs:
  23$ENV{TZ} = 'UTC';
  24$ENV{LC_ALL} = 'C';
  25$| = 1; # unbuffer STDOUT
  26
  27sub fatal (@) { print STDERR @_; exit 1 }
  28require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
  29require SVN::Ra;
  30require SVN::Delta;
  31if ($SVN::Core::VERSION lt '1.1.0') {
  32        fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)\n";
  33}
  34push @Git::SVN::Ra::ISA, 'SVN::Ra';
  35push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
  36push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
  37use Carp qw/croak/;
  38use IO::File qw//;
  39use File::Basename qw/dirname basename/;
  40use File::Path qw/mkpath/;
  41use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev pass_through/;
  42use IPC::Open3;
  43use Git;
  44
  45BEGIN {
  46        my $s;
  47        foreach (qw/command command_oneline command_noisy command_output_pipe
  48                    command_input_pipe command_close_pipe/) {
  49                $s .= "*SVN::Git::Editor::$_ = *SVN::Git::Fetcher::$_ = ".
  50                      "*Git::SVN::Migration::$_ = ".
  51                      "*Git::SVN::Log::$_ = *Git::SVN::$_ = *$_ = *Git::$_; ";
  52        }
  53        eval $s;
  54}
  55
  56my ($SVN);
  57
  58my $_optimize_commits = 1 unless $ENV{GIT_SVN_NO_OPTIMIZE_COMMITS};
  59$sha1 = qr/[a-f\d]{40}/;
  60$sha1_short = qr/[a-f\d]{4,40}/;
  61my ($_stdin, $_help, $_edit,
  62        $_repack, $_repack_nr, $_repack_flags,
  63        $_message, $_file, $_no_metadata,
  64        $_template, $_shared,
  65        $_version, $_upgrade,
  66        $_merge, $_strategy, $_dry_run,
  67        $_prefix);
  68
  69my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
  70                    'config-dir=s' => \$Git::SVN::Ra::config_dir,
  71                    'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache );
  72my %fc_opts = ( 'follow-parent|follow' => \$_follow_parent,
  73                'authors-file|A=s' => \$_authors,
  74                'repack:i' => \$_repack,
  75                'no-metadata' => \$_no_metadata,
  76                'quiet|q' => \$_q,
  77                'repack-flags|repack-args|repack-opts=s' => \$_repack_flags,
  78                %remote_opts );
  79
  80my ($_trunk, $_tags, $_branches);
  81my %multi_opts = ( 'trunk|T=s' => \$_trunk,
  82                'tags|t=s' => \$_tags,
  83                'branches|b=s' => \$_branches );
  84my %init_opts = ( 'template=s' => \$_template, 'shared' => \$_shared );
  85my %cmt_opts = ( 'edit|e' => \$_edit,
  86                'rmdir' => \$_rmdir,
  87                'find-copies-harder' => \$_find_copies_harder,
  88                'l=i' => \$_l,
  89                'copy-similarity|C=i'=> \$_cp_similarity
  90);
  91
  92my %cmd = (
  93        fetch => [ \&cmd_fetch, "Download new revisions from SVN",
  94                        { 'revision|r=s' => \$_revision, %fc_opts } ],
  95        init => [ \&cmd_init, "Initialize a repo for tracking" .
  96                          " (requires URL argument)",
  97                          \%init_opts ],
  98        dcommit => [ \&cmd_dcommit,
  99                     'Commit several diffs to merge with upstream',
 100                        { 'merge|m|M' => \$_merge,
 101                          'strategy|s=s' => \$_strategy,
 102                          'dry-run|n' => \$_dry_run,
 103                        %cmt_opts, %fc_opts } ],
 104        'set-tree' => [ \&cmd_set_tree,
 105                        "Set an SVN repository to a git tree-ish",
 106                        { 'stdin|' => \$_stdin, %cmt_opts, %fc_opts, } ],
 107        'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
 108                        { 'revision|r=i' => \$_revision } ],
 109        rebuild => [ \&cmd_rebuild, "Rebuild git-svn metadata (after git clone)",
 110                        { 'copy-remote|remote=s' => \$_cp_remote,
 111                          'upgrade' => \$_upgrade } ],
 112        'multi-init' => [ \&cmd_multi_init,
 113                        'Initialize multiple trees (like git-svnimport)',
 114                        { %multi_opts, %init_opts, %remote_opts,
 115                         'revision|r=i' => \$_revision,
 116                         'prefix=s' => \$_prefix,
 117                        } ],
 118        'multi-fetch' => [ \&cmd_multi_fetch,
 119                        'Fetch multiple trees (like git-svnimport)',
 120                        \%fc_opts ],
 121        'migrate' => [ sub { },
 122                       # no-op, we automatically run this anyways,
 123                       'Migrate configuration/metadata/layout from
 124                        previous versions of git-svn',
 125                        \%remote_opts ],
 126        'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
 127                        { 'limit=i' => \$Git::SVN::Log::limit,
 128                          'revision|r=s' => \$_revision,
 129                          'verbose|v' => \$Git::SVN::Log::verbose,
 130                          'incremental' => \$Git::SVN::Log::incremental,
 131                          'oneline' => \$Git::SVN::Log::oneline,
 132                          'show-commit' => \$Git::SVN::Log::show_commit,
 133                          'non-recursive' => \$Git::SVN::Log::non_recursive,
 134                          'authors-file|A=s' => \$_authors,
 135                          'color' => \$Git::SVN::Log::color,
 136                          'pager=s' => \$Git::SVN::Log::pager,
 137                        } ],
 138        'commit-diff' => [ \&cmd_commit_diff,
 139                           'Commit a diff between two trees',
 140                        { 'message|m=s' => \$_message,
 141                          'file|F=s' => \$_file,
 142                          'revision|r=s' => \$_revision,
 143                        %cmt_opts } ],
 144);
 145
 146my $cmd;
 147for (my $i = 0; $i < @ARGV; $i++) {
 148        if (defined $cmd{$ARGV[$i]}) {
 149                $cmd = $ARGV[$i];
 150                splice @ARGV, $i, 1;
 151                last;
 152        }
 153};
 154
 155my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
 156
 157read_repo_config(\%opts);
 158my $rv = GetOptions(%opts, 'help|H|h' => \$_help,
 159                                'version|V' => \$_version,
 160                                'minimize-connections' =>
 161                                  \$Git::SVN::Migration::_minimize,
 162                                'id|i=s' => \$Git::SVN::default_ref_id);
 163exit 1 if (!$rv && $cmd ne 'log');
 164
 165usage(0) if $_help;
 166version() if $_version;
 167usage(1) unless defined $cmd;
 168load_authors() if $_authors;
 169unless ($cmd =~ /^(?:init|rebuild|multi-init|commit-diff)$/) {
 170        Git::SVN::Migration::migration_check();
 171}
 172eval {
 173        Git::SVN::verify_remotes_sanity();
 174        $cmd{$cmd}->[0]->(@ARGV);
 175};
 176fatal $@ if $@;
 177exit 0;
 178
 179####################### primary functions ######################
 180sub usage {
 181        my $exit = shift || 0;
 182        my $fd = $exit ? \*STDERR : \*STDOUT;
 183        print $fd <<"";
 184git-svn - bidirectional operations between a single Subversion tree and git
 185Usage: $0 <command> [options] [arguments]\n
 186
 187        print $fd "Available commands:\n" unless $cmd;
 188
 189        foreach (sort keys %cmd) {
 190                next if $cmd && $cmd ne $_;
 191                print $fd '  ',pack('A17',$_),$cmd{$_}->[1],"\n";
 192                foreach (keys %{$cmd{$_}->[2]}) {
 193                        # prints out arguments as they should be passed:
 194                        my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
 195                        print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
 196                                                        "--$_" : "-$_" }
 197                                                split /\|/,$_)," $x\n";
 198                }
 199        }
 200        print $fd <<"";
 201\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
 202arbitrary identifier if you're tracking multiple SVN branches/repositories in
 203one git repository and want to keep them separate.  See git-svn(1) for more
 204information.
 205
 206        exit $exit;
 207}
 208
 209sub version {
 210        print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
 211        exit 0;
 212}
 213
 214sub cmd_rebuild {
 215        my $url = shift;
 216        my $gs = $url ? Git::SVN->init($url)
 217                      : eval { Git::SVN->new };
 218        $gs ||= Git::SVN->_new;
 219        if (!verify_ref($gs->refname.'^0')) {
 220                $gs->copy_remote_ref;
 221        }
 222
 223        my ($rev_list, $ctx) = command_output_pipe("rev-list", $gs->refname);
 224        my $latest;
 225        my $svn_uuid;
 226        while (<$rev_list>) {
 227                chomp;
 228                my $c = $_;
 229                fatal "Non-SHA1: $c\n" unless $c =~ /^$sha1$/o;
 230                my ($url, $rev, $uuid) = cmt_metadata($c);
 231
 232                # ignore merges (from set-tree)
 233                next if (!defined $rev || !$uuid);
 234
 235                # if we merged or otherwise started elsewhere, this is
 236                # how we break out of it
 237                if ((defined $svn_uuid && ($uuid ne $svn_uuid)) ||
 238                    ($gs->{url} && $url && ($url ne $gs->{url}))) {
 239                        next;
 240                }
 241
 242                unless (defined $latest) {
 243                        if (!$gs->{url} && !$url) {
 244                                fatal "SVN repository location required\n";
 245                        }
 246                        $gs = Git::SVN->init($url);
 247                        $latest = $rev;
 248                }
 249                $gs->rev_db_set($rev, $c);
 250                print "r$rev = $c\n";
 251        }
 252        command_close_pipe($rev_list, $ctx);
 253}
 254
 255sub do_git_init_db {
 256        unless (-d $ENV{GIT_DIR}) {
 257                my @init_db = ('init');
 258                push @init_db, "--template=$_template" if defined $_template;
 259                push @init_db, "--shared" if defined $_shared;
 260                command_noisy(@init_db);
 261        }
 262}
 263
 264sub cmd_init {
 265        my $url = shift or die "SVN repository location required " .
 266                                "as a command-line argument\n";
 267        if (my $repo_path = shift) {
 268                unless (-d $repo_path) {
 269                        mkpath([$repo_path]);
 270                }
 271                chdir $repo_path or croak $!;
 272                $ENV{GIT_DIR} = $repo_path . "/.git";
 273        }
 274        do_git_init_db();
 275
 276        Git::SVN->init($url);
 277}
 278
 279sub cmd_fetch {
 280        if (@_) {
 281                die "Additional fetch arguments are no longer supported.\n",
 282                    "Use --follow-parent if you have moved/copied directories
 283                    instead.\n";
 284        }
 285        my $gs = Git::SVN->new;
 286        $gs->fetch(parse_revision_argument());
 287        if ($gs->{last_commit} && !verify_ref('refs/heads/master^0')) {
 288                command_noisy(qw(update-ref refs/heads/master),
 289                              $gs->{last_commit});
 290        }
 291}
 292
 293sub cmd_set_tree {
 294        my (@commits) = @_;
 295        if ($_stdin || !@commits) {
 296                print "Reading from stdin...\n";
 297                @commits = ();
 298                while (<STDIN>) {
 299                        if (/\b($sha1_short)\b/o) {
 300                                unshift @commits, $1;
 301                        }
 302                }
 303        }
 304        my @revs;
 305        foreach my $c (@commits) {
 306                my @tmp = command('rev-parse',$c);
 307                if (scalar @tmp == 1) {
 308                        push @revs, $tmp[0];
 309                } elsif (scalar @tmp > 1) {
 310                        push @revs, reverse(command('rev-list',@tmp));
 311                } else {
 312                        fatal "Failed to rev-parse $c\n";
 313                }
 314        }
 315        my $gs = Git::SVN->new;
 316        my ($r_last, $cmt_last) = $gs->last_rev_commit;
 317        $gs->fetch;
 318        if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
 319                fatal "There are new revisions that were fetched ",
 320                      "and need to be merged (or acknowledged) ",
 321                      "before committing.\nlast rev: $r_last\n",
 322                      " current: $gs->{last_rev}\n";
 323        }
 324        $gs->set_tree($_) foreach @revs;
 325        print "Done committing ",scalar @revs," revisions to SVN\n";
 326}
 327
 328sub cmd_dcommit {
 329        my $head = shift;
 330        my $gs = Git::SVN->new;
 331        $head ||= 'HEAD';
 332        my @refs = command(qw/rev-list --no-merges/, $gs->refname."..$head");
 333        my $last_rev;
 334        foreach my $d (reverse @refs) {
 335                if (!verify_ref("$d~1")) {
 336                        fatal "Commit $d\n",
 337                              "has no parent commit, and therefore ",
 338                              "nothing to diff against.\n",
 339                              "You should be working from a repository ",
 340                              "originally created by git-svn\n";
 341                }
 342                unless (defined $last_rev) {
 343                        (undef, $last_rev, undef) = cmt_metadata("$d~1");
 344                        unless (defined $last_rev) {
 345                                fatal "Unable to extract revision information ",
 346                                      "from commit $d~1\n";
 347                        }
 348                }
 349                if ($_dry_run) {
 350                        print "diff-tree $d~1 $d\n";
 351                } else {
 352                        my $log = get_commit_entry($d)->{log};
 353                        my $ra = $gs->ra;
 354                        my $pool = SVN::Pool->new;
 355                        my %ed_opts = ( r => $last_rev,
 356                                        ra => $ra->dup,
 357                                        svn_path => $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 parse_revision_argument {
 486        if (!defined $_revision || $_revision eq 'BASE:HEAD') {
 487                return (undef, undef);
 488        }
 489        return ($1, $2) if ($_revision =~ /^(\d+):(\d+)$/);
 490        return ($_revision, $_revision) if ($_revision =~ /^\d+$/);
 491        return (undef, $1) if ($_revision =~ /^BASE:(\d+)$/);
 492        return ($1, undef) if ($_revision =~ /^(\d+):HEAD$/);
 493        die "revision argument: $_revision not understood by git-svn\n",
 494            "Try using the command-line svn client instead\n";
 495}
 496
 497sub complete_svn_url {
 498        my ($url, $path) = @_;
 499        $path =~ s#/+$##;
 500        if ($path !~ m#^[a-z\+]+://#) {
 501                if (!defined $url || $url !~ m#^[a-z\+]+://#) {
 502                        fatal("E: '$path' is not a complete URL ",
 503                              "and a separate URL is not specified\n");
 504                }
 505                return ($url, $path);
 506        }
 507        return ($path, '');
 508}
 509
 510sub complete_url_ls_init {
 511        my ($ra, $repo_path, $switch, $pfx) = @_;
 512        unless ($repo_path) {
 513                print STDERR "W: $switch not specified\n";
 514                return;
 515        }
 516        $repo_path =~ s#/+$##;
 517        if ($repo_path =~ m#^[a-z\+]+://#) {
 518                $ra = Git::SVN::Ra->new($repo_path);
 519                $repo_path = '';
 520        } else {
 521                $repo_path =~ s#^/+##;
 522                unless ($ra) {
 523                        fatal("E: '$repo_path' is not a complete URL ",
 524                              "and a separate URL is not specified\n");
 525                }
 526        }
 527        my $r = defined $_revision ? $_revision : $ra->get_latest_revnum;
 528        my ($dirent, undef, undef) = $ra->get_dir($repo_path, $r);
 529        my $url = $ra->{url};
 530        foreach my $d (sort keys %$dirent) {
 531                next if ($dirent->{$d}->kind != $SVN::Node::dir);
 532                my $path =  "$repo_path/$d";
 533                my $ref = "$pfx$d";
 534                my $gs = eval { Git::SVN->new($ref) };
 535                # don't try to init already existing refs
 536                unless ($gs) {
 537                        print "init $url/$path => $ref\n";
 538                        Git::SVN->init($url, $path, undef, $ref);
 539                }
 540        }
 541}
 542
 543sub verify_ref {
 544        my ($ref) = @_;
 545        eval { command_oneline([ 'rev-parse', '--verify', $ref ],
 546                               { STDERR => 0 }); };
 547}
 548
 549sub get_tree_from_treeish {
 550        my ($treeish) = @_;
 551        # $treeish can be a symbolic ref, too:
 552        my $type = command_oneline(qw/cat-file -t/, $treeish);
 553        my $expected;
 554        while ($type eq 'tag') {
 555                ($treeish, $type) = command(qw/cat-file tag/, $treeish);
 556        }
 557        if ($type eq 'commit') {
 558                $expected = (grep /^tree /, command(qw/cat-file commit/,
 559                                                    $treeish))[0];
 560                ($expected) = ($expected =~ /^tree ($sha1)$/o);
 561                die "Unable to get tree from $treeish\n" unless $expected;
 562        } elsif ($type eq 'tree') {
 563                $expected = $treeish;
 564        } else {
 565                die "$treeish is a $type, expected tree, tag or commit\n";
 566        }
 567        return $expected;
 568}
 569
 570sub get_commit_entry {
 571        my ($treeish) = shift;
 572        my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
 573        my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
 574        my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
 575        open my $log_fh, '>', $commit_editmsg or croak $!;
 576
 577        my $type = command_oneline(qw/cat-file -t/, $treeish);
 578        if ($type eq 'commit' || $type eq 'tag') {
 579                my ($msg_fh, $ctx) = command_output_pipe('cat-file',
 580                                                         $type, $treeish);
 581                my $in_msg = 0;
 582                while (<$msg_fh>) {
 583                        if (!$in_msg) {
 584                                $in_msg = 1 if (/^\s*$/);
 585                        } elsif (/^git-svn-id: /) {
 586                                # skip this for now, we regenerate the
 587                                # correct one on re-fetch anyways
 588                                # TODO: set *:merge properties or like...
 589                        } else {
 590                                print $log_fh $_ or croak $!;
 591                        }
 592                }
 593                command_close_pipe($msg_fh, $ctx);
 594        }
 595        close $log_fh or croak $!;
 596
 597        if ($_edit || ($type eq 'tree')) {
 598                my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
 599                # TODO: strip out spaces, comments, like git-commit.sh
 600                system($editor, $commit_editmsg);
 601        }
 602        rename $commit_editmsg, $commit_msg or croak $!;
 603        open $log_fh, '<', $commit_msg or croak $!;
 604        { local $/; chomp($log_entry{log} = <$log_fh>); }
 605        close $log_fh or croak $!;
 606        unlink $commit_msg;
 607        \%log_entry;
 608}
 609
 610sub s_to_file {
 611        my ($str, $file, $mode) = @_;
 612        open my $fd,'>',$file or croak $!;
 613        print $fd $str,"\n" or croak $!;
 614        close $fd or croak $!;
 615        chmod ($mode &~ umask, $file) if (defined $mode);
 616}
 617
 618sub file_to_s {
 619        my $file = shift;
 620        open my $fd,'<',$file or croak "$!: file: $file\n";
 621        local $/;
 622        my $ret = <$fd>;
 623        close $fd or croak $!;
 624        $ret =~ s/\s*$//s;
 625        return $ret;
 626}
 627
 628# '<svn username> = real-name <email address>' mapping based on git-svnimport:
 629sub load_authors {
 630        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
 631        my $log = $cmd eq 'log';
 632        while (<$authors>) {
 633                chomp;
 634                next unless /^(\S+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
 635                my ($user, $name, $email) = ($1, $2, $3);
 636                if ($log) {
 637                        $Git::SVN::Log::rusers{"$name <$email>"} = $user;
 638                } else {
 639                        $users{$user} = [$name, $email];
 640                }
 641        }
 642        close $authors or croak $!;
 643}
 644
 645# convert GetOpt::Long specs for use by git-config
 646sub read_repo_config {
 647        return unless -d $ENV{GIT_DIR};
 648        my $opts = shift;
 649        foreach my $o (keys %$opts) {
 650                my $v = $opts->{$o};
 651                my ($key) = ($o =~ /^([a-z\-]+)/);
 652                $key =~ s/-//g;
 653                my $arg = 'git-config';
 654                $arg .= ' --int' if ($o =~ /[:=]i$/);
 655                $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
 656                if (ref $v eq 'ARRAY') {
 657                        chomp(my @tmp = `$arg --get-all svn.$key`);
 658                        @$v = @tmp if @tmp;
 659                } else {
 660                        chomp(my $tmp = `$arg --get svn.$key`);
 661                        if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
 662                                $$v = $tmp;
 663                        }
 664                }
 665        }
 666}
 667
 668sub extract_metadata {
 669        my $id = shift or return (undef, undef, undef);
 670        my ($url, $rev, $uuid) = ($id =~ /^git-svn-id:\s(\S+?)\@(\d+)
 671                                                        \s([a-f\d\-]+)$/x);
 672        if (!defined $rev || !$uuid || !$url) {
 673                # some of the original repositories I made had
 674                # identifiers like this:
 675                ($rev, $uuid) = ($id =~/^git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
 676        }
 677        return ($url, $rev, $uuid);
 678}
 679
 680sub cmt_metadata {
 681        return extract_metadata((grep(/^git-svn-id: /,
 682                command(qw/cat-file commit/, shift)))[-1]);
 683}
 684
 685sub get_commit_time {
 686        my $cmt = shift;
 687        my $fh = command_output_pipe(qw/rev-list --pretty=raw -n1/, $cmt);
 688        while (<$fh>) {
 689                /^committer\s(?:.+) (\d+) ([\-\+]?\d+)$/ or next;
 690                my ($s, $tz) = ($1, $2);
 691                if ($tz =~ s/^\+//) {
 692                        $s += tz_to_s_offset($tz);
 693                } elsif ($tz =~ s/^\-//) {
 694                        $s -= tz_to_s_offset($tz);
 695                }
 696                close $fh;
 697                return $s;
 698        }
 699        die "Can't get commit time for commit: $cmt\n";
 700}
 701
 702sub tz_to_s_offset {
 703        my ($tz) = @_;
 704        $tz =~ s/(\d\d)$//;
 705        return ($1 * 60) + ($tz * 3600);
 706}
 707
 708package Git::SVN;
 709use strict;
 710use warnings;
 711use vars qw/$default_repo_id $default_ref_id/;
 712use Carp qw/croak/;
 713use File::Path qw/mkpath/;
 714use IPC::Open3;
 715
 716# properties that we do not log:
 717my %SKIP_PROP;
 718BEGIN {
 719        %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
 720                                        svn:special svn:executable
 721                                        svn:entry:committed-rev
 722                                        svn:entry:last-author
 723                                        svn:entry:uuid
 724                                        svn:entry:committed-date/;
 725}
 726
 727sub read_all_remotes {
 728        my $r = {};
 729        foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
 730                if (m!^(.+)\.fetch=\s*(.*)\s*:\s*refs/remotes/(.+)\s*$!) {
 731                        $r->{$1}->{fetch}->{$2} = $3;
 732                } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
 733                        $r->{$1}->{url} = $2;
 734                }
 735        }
 736        $r;
 737}
 738
 739sub verify_remotes_sanity {
 740        return unless -d $ENV{GIT_DIR};
 741        my %seen;
 742        foreach (command(qw/config -l/)) {
 743                if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
 744                        if ($seen{$1}) {
 745                                die "Remote ref refs/remote/$1 is tracked by",
 746                                    "\n  \"$_\"\nand\n  \"$seen{$1}\"\n",
 747                                    "Please resolve this ambiguity in ",
 748                                    "your git configuration file before ",
 749                                    "continuing\n";
 750                        }
 751                        $seen{$1} = $_;
 752                }
 753        }
 754}
 755
 756# we allow more chars than remotes2config.sh...
 757sub sanitize_remote_name {
 758        my ($name) = @_;
 759        $name =~ tr{A-Za-z0-9:,/+-}{.}c;
 760        $name;
 761}
 762
 763sub find_existing_remote {
 764        my ($url, $remotes) = @_;
 765        my $existing;
 766        foreach my $repo_id (keys %$remotes) {
 767                my $u = $remotes->{$repo_id}->{url} or next;
 768                next if $u ne $url;
 769                $existing = $repo_id;
 770                last;
 771        }
 772        $existing;
 773}
 774
 775sub init_remote_config {
 776        my ($self, $url) = @_;
 777        $url =~ s!/+$!!; # strip trailing slash
 778        my $r = read_all_remotes();
 779        my $existing = find_existing_remote($url, $r);
 780        if ($existing) {
 781                print STDERR "Using existing ",
 782                             "[svn-remote \"$existing\"]\n";
 783                $self->{repo_id} = $existing;
 784        } else {
 785                my $min_url = Git::SVN::Ra->new($url)->minimize_url;
 786                $existing = find_existing_remote($min_url, $r);
 787                if ($existing) {
 788                        print STDERR "Using existing ",
 789                                     "[svn-remote \"$existing\"]\n";
 790                        $self->{repo_id} = $existing;
 791                }
 792                if ($min_url ne $url) {
 793                        print STDERR "Using higher level of URL: ",
 794                                     "$url => $min_url\n";
 795                        my $old_path = $self->{path};
 796                        $self->{path} = $url;
 797                        $self->{path} =~ s!^\Q$min_url\E/*!!;
 798                        if (length $old_path) {
 799                                $self->{path} .= "/$old_path";
 800                        }
 801                        $url = $min_url;
 802                }
 803        }
 804        my $orig_url;
 805        if (!$existing) {
 806                # verify that we aren't overwriting anything:
 807                $orig_url = eval {
 808                        command_oneline('config', '--get',
 809                                        "svn-remote.$self->{repo_id}.url")
 810                };
 811                if ($orig_url && ($orig_url ne $url)) {
 812                        die "svn-remote.$self->{repo_id}.url already set: ",
 813                            "$orig_url\nwanted to set to: $url\n";
 814                }
 815        }
 816        my ($xrepo_id, $xpath) = find_ref($self->refname);
 817        if (defined $xpath) {
 818                die "svn-remote.$xrepo_id.fetch already set to track ",
 819                    "$xpath:refs/remotes/", $self->refname, "\n";
 820        }
 821        command_noisy('config',
 822                      "svn-remote.$self->{repo_id}.url", $url);
 823        command_noisy('config', '--add',
 824                      "svn-remote.$self->{repo_id}.fetch",
 825                      "$self->{path}:".$self->refname);
 826        $self->{url} = $url;
 827}
 828
 829sub init {
 830        my ($class, $url, $path, $repo_id, $ref_id) = @_;
 831        my $self = _new($class, $repo_id, $ref_id, $path);
 832        if (defined $url) {
 833                $self->init_remote_config($url);
 834        }
 835        $self;
 836}
 837
 838sub find_ref {
 839        my ($ref_id) = @_;
 840        foreach (command(qw/config -l/)) {
 841                next unless m!^svn-remote\.(.+)\.fetch=
 842                              \s*(.*)\s*:\s*refs/remotes/(.+)\s*$!x;
 843                my ($repo_id, $path, $ref) = ($1, $2, $3);
 844                if ($ref eq $ref_id) {
 845                        $path = '' if ($path =~ m#^\./?#);
 846                        return ($repo_id, $path);
 847                }
 848        }
 849        (undef, undef, undef);
 850}
 851
 852sub new {
 853        my ($class, $ref_id, $repo_id, $path) = @_;
 854        if (defined $ref_id && !defined $repo_id && !defined $path) {
 855                ($repo_id, $path) = find_ref($ref_id);
 856                if (!defined $repo_id) {
 857                        die "Could not find a \"svn-remote.*.fetch\" key ",
 858                            "in the repository configuration matching: ",
 859                            "refs/remotes/$ref_id\n";
 860                }
 861        }
 862        my $self = _new($class, $repo_id, $ref_id, $path);
 863        if (!defined $self->{path} || !length $self->{path}) {
 864                my $fetch = command_oneline('config', '--get',
 865                                            "svn-remote.$repo_id.fetch",
 866                                            ":refs/remotes/$ref_id\$") or
 867                     die "Failed to read \"svn-remote.$repo_id.fetch\" ",
 868                         "\":refs/remotes/$ref_id\$\" in config\n";
 869                ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
 870        }
 871        $self->{url} = command_oneline('config', '--get',
 872                                       "svn-remote.$repo_id.url") or
 873                  die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
 874        $self;
 875}
 876
 877sub refname { "refs/remotes/$_[0]->{ref_id}" }
 878
 879sub ra {
 880        my ($self) = shift;
 881        $self->{ra} ||= Git::SVN::Ra->new($self->{url});
 882}
 883
 884sub rel_path {
 885        my ($self) = @_;
 886        my $repos_root = $self->ra->{repos_root};
 887        return $self->{path} if ($self->{url} eq $repos_root);
 888        my $url = $self->{url} .
 889                  (length $self->{path} ? "/$self->{path}" : $self->{path});
 890        $url =~ s!^\Q$repos_root\E/*!!g;
 891        $url;
 892}
 893
 894sub copy_remote_ref {
 895        my ($self) = @_;
 896        my $origin = $::_cp_remote ? $::_cp_remote : 'origin';
 897        my $ref = $self->refname;
 898        if (command('ls-remote', $origin, $ref)) {
 899                command_noisy('fetch', $origin, "$ref:$ref");
 900        } elsif ($::_cp_remote && !$::_upgrade) {
 901                die "Unable to find remote reference: $ref on $origin\n";
 902        }
 903}
 904
 905sub traverse_ignore {
 906        my ($self, $fh, $path, $r) = @_;
 907        $path =~ s#^/+##g;
 908        my ($dirent, undef, $props) = $self->ra->get_dir($path, $r);
 909        my $p = $path;
 910        $p =~ s#^\Q$self->{ra}->{svn_path}\E/##;
 911        print $fh length $p ? "\n# $p\n" : "\n# /\n";
 912        if (my $s = $props->{'svn:ignore'}) {
 913                $s =~ s/[\r\n]+/\n/g;
 914                chomp $s;
 915                if (length $p == 0) {
 916                        $s =~ s#\n#\n/$p#g;
 917                        print $fh "/$s\n";
 918                } else {
 919                        $s =~ s#\n#\n/$p/#g;
 920                        print $fh "/$p/$s\n";
 921                }
 922        }
 923        foreach (sort keys %$dirent) {
 924                next if $dirent->{$_}->kind != $SVN::Node::dir;
 925                $self->traverse_ignore($fh, "$path/$_", $r);
 926        }
 927}
 928
 929sub last_rev { ($_[0]->last_rev_commit)[0] }
 930sub last_commit { ($_[0]->last_rev_commit)[1] }
 931
 932# returns the newest SVN revision number and newest commit SHA1
 933sub last_rev_commit {
 934        my ($self) = @_;
 935        if (defined $self->{last_rev} && defined $self->{last_commit}) {
 936                return ($self->{last_rev}, $self->{last_commit});
 937        }
 938        my $c = ::verify_ref($self->refname.'^0');
 939        if ($c) {
 940                my $rev = (::cmt_metadata($c))[1];
 941                if (defined $rev) {
 942                        ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
 943                        return ($rev, $c);
 944                }
 945        }
 946        my $offset = -41; # from tail
 947        my $rl;
 948        open my $fh, '<', $self->{db_path} or
 949                                 croak "$self->{db_path} not readable: $!\n";
 950        seek $fh, $offset, 2;
 951        $rl = readline $fh;
 952        defined $rl or return (undef, undef);
 953        chomp $rl;
 954        while ($c ne $rl && tell $fh != 0) {
 955                $offset -= 41;
 956                seek $fh, $offset, 2;
 957                $rl = readline $fh;
 958                defined $rl or return (undef, undef);
 959                chomp $rl;
 960        }
 961        my $rev = tell $fh;
 962        croak $! if ($rev < 0);
 963        $rev =  ($rev - 41) / 41;
 964        close $fh or croak $!;
 965        ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
 966        return ($rev, $c);
 967}
 968
 969sub get_fetch_range {
 970        my ($self, $min, $max) = @_;
 971        $max ||= $self->ra->get_latest_revnum;
 972        $min ||= $self->last_rev || 0;
 973        (++$min, $max);
 974}
 975
 976sub tmp_index_do {
 977        my ($self, $sub) = @_;
 978        my $old_index = $ENV{GIT_INDEX_FILE};
 979        $ENV{GIT_INDEX_FILE} = $self->{index};
 980        my @ret = &$sub;
 981        if ($old_index) {
 982                $ENV{GIT_INDEX_FILE} = $old_index;
 983        } else {
 984                delete $ENV{GIT_INDEX_FILE};
 985        }
 986        wantarray ? @ret : $ret[0];
 987}
 988
 989sub assert_index_clean {
 990        my ($self, $treeish) = @_;
 991
 992        $self->tmp_index_do(sub {
 993                command_noisy('read-tree', $treeish) unless -e $self->{index};
 994                my $x = command_oneline('write-tree');
 995                my ($y) = (command(qw/cat-file commit/, $treeish) =~
 996                           /^tree ($::sha1)/mo);
 997                if ($y ne $x) {
 998                        unlink $self->{index} or croak $!;
 999                        command_noisy('read-tree', $treeish);
1000                }
1001                $x = command_oneline('write-tree');
1002                if ($y ne $x) {
1003                        ::fatal "trees ($treeish) $y != $x\n",
1004                                "Something is seriously wrong...\n";
1005                }
1006        });
1007}
1008
1009sub get_commit_parents {
1010        my ($self, $log_entry, @parents) = @_;
1011        my (%seen, @ret, @tmp);
1012        # commit parents can be conditionally bound to a particular
1013        # svn revision via: "svn_revno=commit_sha1", filter them out here:
1014        foreach my $p (@parents) {
1015                next unless defined $p;
1016                if ($p =~ /^(\d+)=($::sha1_short)$/o) {
1017                        push @tmp, $2 if $1 == $log_entry->{revision};
1018                } else {
1019                        push @tmp, $p if $p =~ /^$::sha1_short$/o;
1020                }
1021        }
1022        if (my $cur = ::verify_ref($self->refname.'^0')) {
1023                push @tmp, $cur;
1024        }
1025        push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
1026        while (my $p = shift @tmp) {
1027                next if $seen{$p};
1028                $seen{$p} = 1;
1029                push @ret, $p;
1030                # MAXPARENT is defined to 16 in commit-tree.c:
1031                last if @ret >= 16;
1032        }
1033        if (@tmp) {
1034                die "r$log_entry->{revision}: No room for parents:\n\t",
1035                    join("\n\t", @tmp), "\n";
1036        }
1037        @ret;
1038}
1039
1040sub full_url {
1041        my ($self) = @_;
1042        $self->ra->{url} . (length $self->{path} ? '/' . $self->{path} : '');
1043}
1044
1045sub do_git_commit {
1046        my ($self, $log_entry, @parents) = @_;
1047        if (my $c = $self->rev_db_get($log_entry->{revision})) {
1048                croak "$log_entry->{revision} = $c already exists! ",
1049                      "Why are we refetching it?\n";
1050        }
1051        my $author = $log_entry->{author};
1052        my ($name, $email) = (defined $::users{$author} ? @{$::users{$author}}
1053                           : ($author, "$author\@".$self->ra->uuid));
1054        $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} = $name;
1055        $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} = $email;
1056        $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
1057
1058        my $tree = $log_entry->{tree};
1059        if (!defined $tree) {
1060                $tree = $self->tmp_index_do(sub {
1061                                            command_oneline('write-tree') });
1062        }
1063        die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
1064
1065        my @exec = ('git-commit-tree', $tree);
1066        foreach ($self->get_commit_parents($log_entry, @parents)) {
1067                push @exec, '-p', $_;
1068        }
1069        defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
1070                                                                   or croak $!;
1071        print $msg_fh $log_entry->{log} or croak $!;
1072        print $msg_fh "\ngit-svn-id: ", $self->full_url, '@',
1073                      $log_entry->{revision}, ' ',
1074                      $self->ra->uuid, "\n" or croak $!;
1075        $msg_fh->flush == 0 or croak $!;
1076        close $msg_fh or croak $!;
1077        chomp(my $commit = do { local $/; <$out_fh> });
1078        close $out_fh or croak $!;
1079        waitpid $pid, 0;
1080        croak $? if $?;
1081        if ($commit !~ /^$::sha1$/o) {
1082                die "Failed to commit, invalid sha1: $commit\n";
1083        }
1084
1085        command_noisy('update-ref',$self->refname, $commit);
1086        $self->rev_db_set($log_entry->{revision}, $commit);
1087
1088        $self->{last_rev} = $log_entry->{revision};
1089        $self->{last_commit} = $commit;
1090        print "r$log_entry->{revision} = $commit\n";
1091        return $commit;
1092}
1093
1094sub revisions_eq {
1095        my ($self, $r0, $r1) = @_;
1096        return 1 if $r0 == $r1;
1097        my $nr = 0;
1098        $self->ra->get_log([$self->{path}], $r0, $r1,
1099                           0, 0, 1, sub { $nr++ });
1100        return 0 if ($nr > 1);
1101        return 1;
1102}
1103
1104sub find_parent_branch {
1105        my ($self, $paths, $rev) = @_;
1106        return undef unless $::_follow_parent;
1107        unless (defined $paths) {
1108                $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1,
1109                                   sub { $paths = dup_changed_paths($_[0]) });
1110        }
1111        return undef unless defined $paths;
1112
1113        # look for a parent from another branch:
1114        my @b_path_components = split m#/#, $self->rel_path;
1115        my @a_path_components;
1116        my $i;
1117        while (@b_path_components) {
1118                $i = $paths->{'/'.join('/', @b_path_components)};
1119                last if $i;
1120                unshift(@a_path_components, pop(@b_path_components));
1121        }
1122        goto not_found unless defined $i;
1123        my $branch_from = $i->{copyfrom_path} or goto not_found;
1124        if (@a_path_components) {
1125                print STDERR "branch_from: $branch_from => ";
1126                $branch_from .= '/'.join('/', @a_path_components);
1127                print STDERR $branch_from, "\n";
1128        }
1129        my $r = $i->{copyfrom_rev};
1130        my $repos_root = $self->ra->{repos_root};
1131        my $url = $self->ra->{url};
1132        my $new_url = $repos_root . $branch_from;
1133        print STDERR  "Found possible branch point: ",
1134                      "$new_url => ", $self->full_url, ", $r\n";
1135        $branch_from =~ s#^/##;
1136        my $remotes = read_all_remotes();
1137        my $gs;
1138        foreach my $repo_id (keys %$remotes) {
1139                my $u = $remotes->{$repo_id}->{url} or next;
1140                next if $url ne $u;
1141                my $fetch = $remotes->{$repo_id}->{fetch};
1142                foreach my $f (keys %$fetch) {
1143                        next if $f ne $branch_from;
1144                        $gs = Git::SVN->new($fetch->{$f}, $repo_id, $f);
1145                        last;
1146                }
1147                last if $gs;
1148        }
1149        unless ($gs) {
1150                my $ref_id = $branch_from;
1151                $ref_id .= "\@$r" if find_ref($ref_id);
1152                # just grow a tail if we're not unique enough :x
1153                $ref_id .= '-' while find_ref($ref_id);
1154                $gs = Git::SVN->init($new_url, '', $ref_id, $ref_id);
1155        }
1156        my ($r0, $parent) = $gs->find_rev_before($r, 1);
1157        if ($::_follow_parent && (!defined $r0 || !defined $parent)) {
1158                $gs->fetch(0, $r);
1159                ($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                     $self->rel_path, "' @ r$rev not found:\n";
1187        return undef unless $paths;
1188        print STDERR "Changed paths:\n";
1189        foreach my $x (sort keys %$paths) {
1190                my $p = $paths->{$x};
1191                print STDERR "\t$p->{action}\t$x";
1192                if ($p->{copyfrom_path}) {
1193                        print STDERR "(from $p->{copyfrom_path}: ",
1194                                     "$p->{copyfrom_rev})";
1195                }
1196                print STDERR "\n";
1197        }
1198        print STDERR '-'x72, "\n";
1199        return undef;
1200}
1201
1202sub do_fetch {
1203        my ($self, $paths, $rev) = @_;
1204        my $ed;
1205        my ($last_rev, @parents);
1206        if ($self->{last_commit}) {
1207                $ed = SVN::Git::Fetcher->new($self);
1208                $last_rev = $self->{last_rev};
1209                $ed->{c} = $self->{last_commit};
1210                @parents = ($self->{last_commit});
1211        } else {
1212                $last_rev = $rev;
1213                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
1214                        return $log_entry;
1215                }
1216                $ed = SVN::Git::Fetcher->new($self);
1217        }
1218        unless ($self->ra->gs_do_update($last_rev, $rev,
1219                                        $self->{path}, 1, $ed)) {
1220                die "SVN connection failed somewhere...\n";
1221        }
1222        $self->make_log_entry($rev, \@parents, $ed);
1223}
1224
1225sub get_untracked {
1226        my ($self, $ed) = @_;
1227        my @out;
1228        my $h = $ed->{empty};
1229        foreach (sort keys %$h) {
1230                my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
1231                push @out, "  $act: " . uri_encode($_);
1232                warn "W: $act: $_\n";
1233        }
1234        foreach my $t (qw/dir_prop file_prop/) {
1235                $h = $ed->{$t} or next;
1236                foreach my $path (sort keys %$h) {
1237                        my $ppath = $path eq '' ? '.' : $path;
1238                        foreach my $prop (sort keys %{$h->{$path}}) {
1239                                next if $SKIP_PROP{$prop};
1240                                my $v = $h->{$path}->{$prop};
1241                                my $t_ppath_prop = "$t: " .
1242                                                    uri_encode($ppath) . ' ' .
1243                                                    uri_encode($prop);
1244                                if (defined $v) {
1245                                        push @out, "  +$t_ppath_prop " .
1246                                                   uri_encode($v);
1247                                } else {
1248                                        push @out, "  -$t_ppath_prop";
1249                                }
1250                        }
1251                }
1252        }
1253        foreach my $t (qw/absent_file absent_directory/) {
1254                $h = $ed->{$t} or next;
1255                foreach my $parent (sort keys %$h) {
1256                        foreach my $path (sort @{$h->{$parent}}) {
1257                                push @out, "  $t: " .
1258                                           uri_encode("$parent/$path");
1259                                warn "W: $t: $parent/$path ",
1260                                     "Insufficient permissions?\n";
1261                        }
1262                }
1263        }
1264        \@out;
1265}
1266
1267sub parse_svn_date {
1268        my $date = shift || return '+0000 1970-01-01 00:00:00';
1269        my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
1270                                            (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x) or
1271                                         croak "Unable to parse date: $date\n";
1272        "+0000 $Y-$m-$d $H:$M:$S";
1273}
1274
1275sub check_author {
1276        my ($author) = @_;
1277        if (!defined $author || length $author == 0) {
1278                $author = '(no author)';
1279        }
1280        if (defined $::_authors && ! defined $::users{$author}) {
1281                die "Author: $author not defined in $::_authors file\n";
1282        }
1283        $author;
1284}
1285
1286sub make_log_entry {
1287        my ($self, $rev, $parents, $ed) = @_;
1288        my $untracked = $self->get_untracked($ed);
1289
1290        return undef if ($ed->{nr} == 0 && scalar @$untracked == 0);
1291
1292        open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
1293        print $un "r$rev\n" or croak $!;
1294        print $un $_, "\n" foreach @$untracked;
1295        my %log_entry = ( parents => $parents || [], revision => $rev,
1296                          log => '');
1297        my $rp = $self->ra->rev_proplist($rev);
1298        foreach (sort keys %$rp) {
1299                my $v = $rp->{$_};
1300                if (/^svn:(author|date|log)$/) {
1301                        $log_entry{$1} = $v;
1302                } else {
1303                        print $un "  rev_prop: ", uri_encode($_), ' ',
1304                                  uri_encode($v), "\n";
1305                }
1306        }
1307        close $un or croak $!;
1308
1309        $log_entry{date} = parse_svn_date($log_entry{date});
1310        $log_entry{author} = check_author($log_entry{author});
1311        $log_entry{log} .= "\n";
1312        \%log_entry;
1313}
1314
1315sub fetch {
1316        my ($self, $min_rev, $max_rev, @parents) = @_;
1317        my ($last_rev, $last_commit) = $self->last_rev_commit;
1318        my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
1319        return if ($base > $head);
1320        if (defined $last_commit) {
1321                $self->assert_index_clean($last_commit);
1322        }
1323        my $inc = 1000;
1324        my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
1325        my $err_handler = $SVN::Error::handler;
1326        my $err;
1327        $SVN::Error::handler = sub { ($err) = @_; skip_unknown_revs($err); } ;
1328        while (1) {
1329                my @revs;
1330                $self->ra->get_log([$self->{path}], $min, $max, 0, 1, 1,
1331                    sub {
1332                        my ($paths, $rev) = @_;
1333                        push @revs, [ dup_changed_paths($paths), $rev ];
1334                        });
1335                if (! @revs && $err && $max >= $head) {
1336                        print STDERR "Branch probably deleted:\n  ",
1337                                     $err->expanded_message,
1338                                     "\nWill attempt to follow revisions ",
1339                                     "r$min .. r$max",
1340                                     "committed before the deletion\n";
1341                        @revs = map { [ undef, $_ ] } ($min .. $max);
1342                }
1343                foreach (@revs) {
1344                        if (my $log_entry = $self->do_fetch(@$_)) {
1345                                $self->do_git_commit($log_entry, @parents);
1346                        }
1347                }
1348                last if $max >= $head;
1349                $min = $max + 1;
1350                $max += $inc;
1351                $max = $head if ($max > $head);
1352        }
1353        $SVN::Error::handler = $err_handler;
1354}
1355
1356sub set_tree_cb {
1357        my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
1358        # TODO: enable and test optimized commits:
1359        if (0 && $rev == ($self->{last_rev} + 1)) {
1360                $log_entry->{revision} = $rev;
1361                $log_entry->{author} = $author;
1362                $self->do_git_commit($log_entry, "$rev=$tree");
1363        } else {
1364                $self->fetch(undef, undef, "$rev=$tree");
1365        }
1366}
1367
1368sub set_tree {
1369        my ($self, $tree) = (shift, shift);
1370        my $log_entry = ::get_commit_entry($tree);
1371        unless ($self->{last_rev}) {
1372                fatal("Must have an existing revision to commit\n");
1373        }
1374        my $pool = SVN::Pool->new;
1375        my $ed = SVN::Git::Editor->new({ r => $self->{last_rev},
1376                                         ra => $self->ra->dup,
1377                                         svn_path => $self->ra->{svn_path}
1378                                       },
1379                                       $self->ra->get_commit_editor(
1380                                         $log_entry->{log}, sub {
1381                                           $self->set_tree_cb($log_entry,
1382                                                              $tree, @_);
1383                                       }),
1384                                       $pool);
1385        my $mods = $ed->apply_diff($self->{last_commit}, $tree);
1386        if (@$mods == 0) {
1387                print "No changes\nr$self->{last_rev} = $tree\n";
1388        }
1389        $pool->clear;
1390}
1391
1392sub skip_unknown_revs {
1393        my ($err) = @_;
1394        my $errno = $err->apr_err();
1395        # Maybe the branch we're tracking didn't
1396        # exist when the repo started, so it's
1397        # not an error if it doesn't, just continue
1398        #
1399        # Wonderfully consistent library, eh?
1400        # 160013 - svn:// and file://
1401        # 175002 - http(s)://
1402        # 175007 - http(s):// (this repo required authorization, too...)
1403        #   More codes may be discovered later...
1404        if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
1405                return;
1406        }
1407        croak "Error from SVN, ($errno): ", $err->expanded_message,"\n";
1408}
1409
1410# svn_log_changed_path_t objects passed to get_log are likely to be
1411# overwritten even if only the refs are copied to an external variable,
1412# so we should dup the structures in their entirety.  Using an externally
1413# passed pool (instead of our temporary and quickly cleared pool in
1414# Git::SVN::Ra) does not help matters at all...
1415sub dup_changed_paths {
1416        my ($paths) = @_;
1417        return undef unless $paths;
1418        my %ret;
1419        foreach my $p (keys %$paths) {
1420                my $i = $paths->{$p};
1421                my %s = map { $_ => $i->$_ }
1422                              qw/copyfrom_path copyfrom_rev action/;
1423                $ret{$p} = \%s;
1424        }
1425        \%ret;
1426}
1427
1428# rev_db:
1429# Tie::File seems to be prone to offset errors if revisions get sparse,
1430# it's not that fast, either.  Tie::File is also not in Perl 5.6.  So
1431# one of my favorite modules is out :<  Next up would be one of the DBM
1432# modules, but I'm not sure which is most portable...  So I'll just
1433# go with something that's plain-text, but still capable of
1434# being randomly accessed.  So here's my ultra-simple fixed-width
1435# database.  All records are 40 characters + "\n", so it's easy to seek
1436# to a revision: (41 * rev) is the byte offset.
1437# A record of 40 0s denotes an empty revision.
1438# And yes, it's still pretty fast (faster than Tie::File).
1439
1440sub rev_db_set {
1441        my ($self, $rev, $commit) = @_;
1442        length $commit == 40 or croak "arg3 must be a full SHA1 hexsum\n";
1443        open my $fh, '+<', $self->{db_path} or croak $!;
1444        my $offset = $rev * 41;
1445        # assume that append is the common case:
1446        seek $fh, 0, 2 or croak $!;
1447        my $pos = tell $fh;
1448        if ($pos < $offset) {
1449                print $fh (('0' x 40),"\n") x (($offset - $pos) / 41)
1450                  or croak $!;
1451        }
1452        seek $fh, $offset, 0 or croak $!;
1453        print $fh $commit,"\n" or croak $!;
1454        close $fh or croak $!;
1455}
1456
1457sub rev_db_get {
1458        my ($self, $rev) = @_;
1459        my $ret;
1460        my $offset = $rev * 41;
1461        open my $fh, '<', $self->{db_path} or croak $!;
1462        if (seek $fh, $offset, 0) {
1463                $ret = readline $fh;
1464                if (defined $ret) {
1465                        chomp $ret;
1466                        $ret = undef if ($ret =~ /^0{40}$/);
1467                }
1468        }
1469        close $fh or croak $!;
1470        $ret;
1471}
1472
1473sub find_rev_before {
1474        my ($self, $rev, $eq_ok) = @_;
1475        --$rev unless $eq_ok;
1476        while ($rev > 0) {
1477                if (my $c = $self->rev_db_get($rev)) {
1478                        return ($rev, $c);
1479                }
1480                --$rev;
1481        }
1482        return (undef, undef);
1483}
1484
1485sub _new {
1486        my ($class, $repo_id, $ref_id, $path) = @_;
1487        unless (defined $repo_id && length $repo_id) {
1488                $repo_id = $Git::SVN::default_repo_id;
1489        }
1490        unless (defined $ref_id && length $ref_id) {
1491                $_[2] = $ref_id = $Git::SVN::default_ref_id;
1492        }
1493        $_[1] = $repo_id = sanitize_remote_name($repo_id);
1494        my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
1495        $_[3] = $path = '' unless (defined $path);
1496        mkpath([$dir]);
1497        unless (-f "$dir/.rev_db") {
1498                open my $fh, '>>', "$dir/.rev_db" or croak $!;
1499                close $fh or croak $!;
1500        }
1501        bless { ref_id => $ref_id, dir => $dir, index => "$dir/index",
1502                path => $path,
1503                db_path => "$dir/.rev_db", repo_id => $repo_id }, $class;
1504}
1505
1506sub uri_encode {
1507        my ($f) = @_;
1508        $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
1509        $f
1510}
1511
1512package Git::SVN::Prompt;
1513use strict;
1514use warnings;
1515require SVN::Core;
1516use vars qw/$_no_auth_cache $_username/;
1517
1518sub simple {
1519        my ($cred, $realm, $default_username, $may_save, $pool) = @_;
1520        $may_save = undef if $_no_auth_cache;
1521        $default_username = $_username if defined $_username;
1522        if (defined $default_username && length $default_username) {
1523                if (defined $realm && length $realm) {
1524                        print STDERR "Authentication realm: $realm\n";
1525                        STDERR->flush;
1526                }
1527                $cred->username($default_username);
1528        } else {
1529                username($cred, $realm, $may_save, $pool);
1530        }
1531        $cred->password(_read_password("Password for '" .
1532                                       $cred->username . "': ", $realm));
1533        $cred->may_save($may_save);
1534        $SVN::_Core::SVN_NO_ERROR;
1535}
1536
1537sub ssl_server_trust {
1538        my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
1539        $may_save = undef if $_no_auth_cache;
1540        print STDERR "Error validating server certificate for '$realm':\n";
1541        if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
1542                print STDERR " - The certificate is not issued by a trusted ",
1543                      "authority. Use the\n",
1544                      "   fingerprint to validate the certificate manually!\n";
1545        }
1546        if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
1547                print STDERR " - The certificate hostname does not match.\n";
1548        }
1549        if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
1550                print STDERR " - The certificate is not yet valid.\n";
1551        }
1552        if ($failures & $SVN::Auth::SSL::EXPIRED) {
1553                print STDERR " - The certificate has expired.\n";
1554        }
1555        if ($failures & $SVN::Auth::SSL::OTHER) {
1556                print STDERR " - The certificate has an unknown error.\n";
1557        }
1558        printf STDERR
1559                "Certificate information:\n".
1560                " - Hostname: %s\n".
1561                " - Valid: from %s until %s\n".
1562                " - Issuer: %s\n".
1563                " - Fingerprint: %s\n",
1564                map $cert_info->$_, qw(hostname valid_from valid_until
1565                                       issuer_dname fingerprint);
1566        my $choice;
1567prompt:
1568        print STDERR $may_save ?
1569              "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
1570              "(R)eject or accept (t)emporarily? ";
1571        STDERR->flush;
1572        $choice = lc(substr(<STDIN> || 'R', 0, 1));
1573        if ($choice =~ /^t$/i) {
1574                $cred->may_save(undef);
1575        } elsif ($choice =~ /^r$/i) {
1576                return -1;
1577        } elsif ($may_save && $choice =~ /^p$/i) {
1578                $cred->may_save($may_save);
1579        } else {
1580                goto prompt;
1581        }
1582        $cred->accepted_failures($failures);
1583        $SVN::_Core::SVN_NO_ERROR;
1584}
1585
1586sub ssl_client_cert {
1587        my ($cred, $realm, $may_save, $pool) = @_;
1588        $may_save = undef if $_no_auth_cache;
1589        print STDERR "Client certificate filename: ";
1590        STDERR->flush;
1591        chomp(my $filename = <STDIN>);
1592        $cred->cert_file($filename);
1593        $cred->may_save($may_save);
1594        $SVN::_Core::SVN_NO_ERROR;
1595}
1596
1597sub ssl_client_cert_pw {
1598        my ($cred, $realm, $may_save, $pool) = @_;
1599        $may_save = undef if $_no_auth_cache;
1600        $cred->password(_read_password("Password: ", $realm));
1601        $cred->may_save($may_save);
1602        $SVN::_Core::SVN_NO_ERROR;
1603}
1604
1605sub username {
1606        my ($cred, $realm, $may_save, $pool) = @_;
1607        $may_save = undef if $_no_auth_cache;
1608        if (defined $realm && length $realm) {
1609                print STDERR "Authentication realm: $realm\n";
1610        }
1611        my $username;
1612        if (defined $_username) {
1613                $username = $_username;
1614        } else {
1615                print STDERR "Username: ";
1616                STDERR->flush;
1617                chomp($username = <STDIN>);
1618        }
1619        $cred->username($username);
1620        $cred->may_save($may_save);
1621        $SVN::_Core::SVN_NO_ERROR;
1622}
1623
1624sub _read_password {
1625        my ($prompt, $realm) = @_;
1626        print STDERR $prompt;
1627        STDERR->flush;
1628        require Term::ReadKey;
1629        Term::ReadKey::ReadMode('noecho');
1630        my $password = '';
1631        while (defined(my $key = Term::ReadKey::ReadKey(0))) {
1632                last if $key =~ /[\012\015]/; # \n\r
1633                $password .= $key;
1634        }
1635        Term::ReadKey::ReadMode('restore');
1636        print STDERR "\n";
1637        STDERR->flush;
1638        $password;
1639}
1640
1641package main;
1642
1643sub uri_encode {
1644        my ($f) = @_;
1645        $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
1646        $f
1647}
1648
1649sub uri_decode {
1650        my ($f) = @_;
1651        $f =~ tr/+/ /;
1652        $f =~ s/%([A-F0-9]{2})/chr hex($1)/ge;
1653        $f
1654}
1655
1656{
1657        my $kill_stupid_warnings = $SVN::Node::none.$SVN::Node::file.
1658                                $SVN::Node::dir.$SVN::Node::unknown.
1659                                $SVN::Node::none.$SVN::Node::file.
1660                                $SVN::Node::dir.$SVN::Node::unknown.
1661                                $SVN::Auth::SSL::CNMISMATCH.
1662                                $SVN::Auth::SSL::NOTYETVALID.
1663                                $SVN::Auth::SSL::EXPIRED.
1664                                $SVN::Auth::SSL::UNKNOWNCA.
1665                                $SVN::Auth::SSL::OTHER;
1666}
1667
1668package SVN::Git::Fetcher;
1669use vars qw/@ISA/;
1670use strict;
1671use warnings;
1672use Carp qw/croak/;
1673use IO::File qw//;
1674
1675# file baton members: path, mode_a, mode_b, pool, fh, blob, base
1676sub new {
1677        my ($class, $git_svn) = @_;
1678        my $self = SVN::Delta::Editor->new;
1679        bless $self, $class;
1680        $self->{c} = $git_svn->{last_commit} if exists $git_svn->{last_commit};
1681        $self->{empty} = {};
1682        $self->{dir_prop} = {};
1683        $self->{file_prop} = {};
1684        $self->{absent_dir} = {};
1685        $self->{absent_file} = {};
1686        $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
1687        require Digest::MD5;
1688        $self;
1689}
1690
1691sub set_path_strip {
1692        my ($self, $path) = @_;
1693        $self->{path_strip} = qr/^\Q$path\E\/?/;
1694}
1695
1696sub open_root {
1697        { path => '' };
1698}
1699
1700sub open_directory {
1701        my ($self, $path, $pb, $rev) = @_;
1702        { path => $path };
1703}
1704
1705sub git_path {
1706        my ($self, $path) = @_;
1707        $path =~ s!$self->{path_strip}!! if $self->{path_strip};
1708        $path;
1709}
1710
1711sub delete_entry {
1712        my ($self, $path, $rev, $pb) = @_;
1713
1714        my $gpath = $self->git_path($path);
1715        # remove entire directories.
1716        if (command('ls-tree', $self->{c}, '--', $gpath) =~ /^040000 tree/) {
1717                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
1718                                                     -r --name-only -z/,
1719                                                     $self->{c}, '--', $gpath);
1720                local $/ = "\0";
1721                while (<$ls>) {
1722                        chomp;
1723                        $self->{gii}->remove($_);
1724                        print "\tD\t$_\n" unless $self->{q};
1725                }
1726                print "\tD\t$gpath/\n" unless $self->{q};
1727                command_close_pipe($ls, $ctx);
1728                $self->{empty}->{$path} = 0
1729        } else {
1730                $self->{gii}->remove($gpath);
1731                print "\tD\t$gpath\n" unless $self->{q};
1732        }
1733        undef;
1734}
1735
1736sub open_file {
1737        my ($self, $path, $pb, $rev) = @_;
1738        my $gpath = $self->git_path($path);
1739        my ($mode, $blob) = (command('ls-tree', $self->{c}, '--', $gpath)
1740                             =~ /^(\d{6}) blob ([a-f\d]{40})\t/);
1741        unless (defined $mode && defined $blob) {
1742                die "$path was not found in commit $self->{c} (r$rev)\n";
1743        }
1744        { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
1745          pool => SVN::Pool->new, action => 'M' };
1746}
1747
1748sub add_file {
1749        my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
1750        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
1751        delete $self->{empty}->{$dir};
1752        { path => $path, mode_a => 100644, mode_b => 100644,
1753          pool => SVN::Pool->new, action => 'A' };
1754}
1755
1756sub add_directory {
1757        my ($self, $path, $cp_path, $cp_rev) = @_;
1758        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
1759        delete $self->{empty}->{$dir};
1760        $self->{empty}->{$path} = 1;
1761        { path => $path };
1762}
1763
1764sub change_dir_prop {
1765        my ($self, $db, $prop, $value) = @_;
1766        $self->{dir_prop}->{$db->{path}} ||= {};
1767        $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
1768        undef;
1769}
1770
1771sub absent_directory {
1772        my ($self, $path, $pb) = @_;
1773        $self->{absent_dir}->{$pb->{path}} ||= [];
1774        push @{$self->{absent_dir}->{$pb->{path}}}, $path;
1775        undef;
1776}
1777
1778sub absent_file {
1779        my ($self, $path, $pb) = @_;
1780        $self->{absent_file}->{$pb->{path}} ||= [];
1781        push @{$self->{absent_file}->{$pb->{path}}}, $path;
1782        undef;
1783}
1784
1785sub change_file_prop {
1786        my ($self, $fb, $prop, $value) = @_;
1787        if ($prop eq 'svn:executable') {
1788                if ($fb->{mode_b} != 120000) {
1789                        $fb->{mode_b} = defined $value ? 100755 : 100644;
1790                }
1791        } elsif ($prop eq 'svn:special') {
1792                $fb->{mode_b} = defined $value ? 120000 : 100644;
1793        } else {
1794                $self->{file_prop}->{$fb->{path}} ||= {};
1795                $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
1796        }
1797        undef;
1798}
1799
1800sub apply_textdelta {
1801        my ($self, $fb, $exp) = @_;
1802        my $fh = IO::File->new_tmpfile;
1803        $fh->autoflush(1);
1804        # $fh gets auto-closed() by SVN::TxDelta::apply(),
1805        # (but $base does not,) so dup() it for reading in close_file
1806        open my $dup, '<&', $fh or croak $!;
1807        my $base = IO::File->new_tmpfile;
1808        $base->autoflush(1);
1809        if ($fb->{blob}) {
1810                defined (my $pid = fork) or croak $!;
1811                if (!$pid) {
1812                        open STDOUT, '>&', $base or croak $!;
1813                        print STDOUT 'link ' if ($fb->{mode_a} == 120000);
1814                        exec qw/git-cat-file blob/, $fb->{blob} or croak $!;
1815                }
1816                waitpid $pid, 0;
1817                croak $? if $?;
1818
1819                if (defined $exp) {
1820                        seek $base, 0, 0 or croak $!;
1821                        my $md5 = Digest::MD5->new;
1822                        $md5->addfile($base);
1823                        my $got = $md5->hexdigest;
1824                        die "Checksum mismatch: $fb->{path} $fb->{blob}\n",
1825                            "expected: $exp\n",
1826                            "     got: $got\n" if ($got ne $exp);
1827                }
1828        }
1829        seek $base, 0, 0 or croak $!;
1830        $fb->{fh} = $dup;
1831        $fb->{base} = $base;
1832        [ SVN::TxDelta::apply($base, $fh, undef, $fb->{path}, $fb->{pool}) ];
1833}
1834
1835sub close_file {
1836        my ($self, $fb, $exp) = @_;
1837        my $hash;
1838        my $path = $self->git_path($fb->{path});
1839        if (my $fh = $fb->{fh}) {
1840                seek($fh, 0, 0) or croak $!;
1841                my $md5 = Digest::MD5->new;
1842                $md5->addfile($fh);
1843                my $got = $md5->hexdigest;
1844                die "Checksum mismatch: $path\n",
1845                    "expected: $exp\n    got: $got\n" if ($got ne $exp);
1846                seek($fh, 0, 0) or croak $!;
1847                if ($fb->{mode_b} == 120000) {
1848                        read($fh, my $buf, 5) == 5 or croak $!;
1849                        $buf eq 'link ' or die "$path has mode 120000",
1850                                               "but is not a link\n";
1851                }
1852                defined(my $pid = open my $out,'-|') or die "Can't fork: $!\n";
1853                if (!$pid) {
1854                        open STDIN, '<&', $fh or croak $!;
1855                        exec qw/git-hash-object -w --stdin/ or croak $!;
1856                }
1857                chomp($hash = do { local $/; <$out> });
1858                close $out or croak $!;
1859                close $fh or croak $!;
1860                $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
1861                close $fb->{base} or croak $!;
1862        } else {
1863                $hash = $fb->{blob} or die "no blob information\n";
1864        }
1865        $fb->{pool}->clear;
1866        $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
1867        print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $self->{q};
1868        undef;
1869}
1870
1871sub abort_edit {
1872        my $self = shift;
1873        $self->{nr} = $self->{gii}->{nr};
1874        delete $self->{gii};
1875        $self->SUPER::abort_edit(@_);
1876}
1877
1878sub close_edit {
1879        my $self = shift;
1880        $self->{git_commit_ok} = 1;
1881        $self->{nr} = $self->{gii}->{nr};
1882        delete $self->{gii};
1883        $self->SUPER::close_edit(@_);
1884}
1885
1886package SVN::Git::Editor;
1887use vars qw/@ISA/;
1888use strict;
1889use warnings;
1890use Carp qw/croak/;
1891use IO::File;
1892
1893sub new {
1894        my $class = shift;
1895        my $git_svn = shift;
1896        my $self = SVN::Delta::Editor->new(@_);
1897        bless $self, $class;
1898        foreach (qw/svn_path r ra/) {
1899                die "$_ required!\n" unless (defined $git_svn->{$_});
1900                $self->{$_} = $git_svn->{$_};
1901        }
1902        $self->{pool} = SVN::Pool->new;
1903        $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
1904        $self->{rm} = { };
1905        require Digest::MD5;
1906        return $self;
1907}
1908
1909sub split_path {
1910        return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
1911}
1912
1913sub repo_path {
1914        (defined $_[1] && length $_[1]) ? $_[1] : ''
1915}
1916
1917sub url_path {
1918        my ($self, $path) = @_;
1919        $self->{ra}->{url} . '/' . $self->repo_path($path);
1920}
1921
1922sub rmdirs {
1923        my ($self, $tree_b) = @_;
1924        my $rm = $self->{rm};
1925        delete $rm->{''}; # we never delete the url we're tracking
1926        return unless %$rm;
1927
1928        foreach (keys %$rm) {
1929                my @d = split m#/#, $_;
1930                my $c = shift @d;
1931                $rm->{$c} = 1;
1932                while (@d) {
1933                        $c .= '/' . shift @d;
1934                        $rm->{$c} = 1;
1935                }
1936        }
1937        delete $rm->{$self->{svn_path}};
1938        delete $rm->{''}; # we never delete the url we're tracking
1939        return unless %$rm;
1940
1941        my ($fh, $ctx) = command_output_pipe(
1942                                   qw/ls-tree --name-only -r -z/, $tree_b);
1943        local $/ = "\0";
1944        while (<$fh>) {
1945                chomp;
1946                my @dn = split m#/#, $_;
1947                while (pop @dn) {
1948                        delete $rm->{join '/', @dn};
1949                }
1950                unless (%$rm) {
1951                        close $fh;
1952                        return;
1953                }
1954        }
1955        command_close_pipe($fh, $ctx);
1956
1957        my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
1958        foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
1959                $self->close_directory($bat->{$d}, $p);
1960                my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
1961                print "\tD+\t$d/\n" unless $::_q;
1962                $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
1963                delete $bat->{$d};
1964        }
1965}
1966
1967sub open_or_add_dir {
1968        my ($self, $full_path, $baton) = @_;
1969        my $t = $self->{ra}->check_path($full_path, $self->{r});
1970        if ($t == $SVN::Node::none) {
1971                return $self->add_directory($full_path, $baton,
1972                                                undef, -1, $self->{pool});
1973        } elsif ($t == $SVN::Node::dir) {
1974                return $self->open_directory($full_path, $baton,
1975                                                $self->{r}, $self->{pool});
1976        }
1977        print STDERR "$full_path already exists in repository at ",
1978                "r$self->{r} and it is not a directory (",
1979                ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
1980        exit 1;
1981}
1982
1983sub ensure_path {
1984        my ($self, $path) = @_;
1985        my $bat = $self->{bat};
1986        $path = $self->repo_path($path);
1987        return $bat->{''} unless (length $path);
1988        my @p = split m#/+#, $path;
1989        my $c = shift @p;
1990        $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
1991        while (@p) {
1992                my $c0 = $c;
1993                $c .= '/' . shift @p;
1994                $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
1995        }
1996        return $bat->{$c};
1997}
1998
1999sub A {
2000        my ($self, $m) = @_;
2001        my ($dir, $file) = split_path($m->{file_b});
2002        my $pbat = $self->ensure_path($dir);
2003        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2004                                        undef, -1);
2005        print "\tA\t$m->{file_b}\n" unless $::_q;
2006        $self->chg_file($fbat, $m);
2007        $self->close_file($fbat,undef,$self->{pool});
2008}
2009
2010sub C {
2011        my ($self, $m) = @_;
2012        my ($dir, $file) = split_path($m->{file_b});
2013        my $pbat = $self->ensure_path($dir);
2014        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2015                                $self->url_path($m->{file_a}), $self->{r});
2016        print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
2017        $self->chg_file($fbat, $m);
2018        $self->close_file($fbat,undef,$self->{pool});
2019}
2020
2021sub delete_entry {
2022        my ($self, $path, $pbat) = @_;
2023        my $rpath = $self->repo_path($path);
2024        my ($dir, $file) = split_path($rpath);
2025        $self->{rm}->{$dir} = 1;
2026        $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
2027}
2028
2029sub R {
2030        my ($self, $m) = @_;
2031        my ($dir, $file) = split_path($m->{file_b});
2032        my $pbat = $self->ensure_path($dir);
2033        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2034                                $self->url_path($m->{file_a}), $self->{r});
2035        print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
2036        $self->chg_file($fbat, $m);
2037        $self->close_file($fbat,undef,$self->{pool});
2038
2039        ($dir, $file) = split_path($m->{file_a});
2040        $pbat = $self->ensure_path($dir);
2041        $self->delete_entry($m->{file_a}, $pbat);
2042}
2043
2044sub M {
2045        my ($self, $m) = @_;
2046        my ($dir, $file) = split_path($m->{file_b});
2047        my $pbat = $self->ensure_path($dir);
2048        my $fbat = $self->open_file($self->repo_path($m->{file_b}),
2049                                $pbat,$self->{r},$self->{pool});
2050        print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
2051        $self->chg_file($fbat, $m);
2052        $self->close_file($fbat,undef,$self->{pool});
2053}
2054
2055sub T { shift->M(@_) }
2056
2057sub change_file_prop {
2058        my ($self, $fbat, $pname, $pval) = @_;
2059        $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
2060}
2061
2062sub chg_file {
2063        my ($self, $fbat, $m) = @_;
2064        if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
2065                $self->change_file_prop($fbat,'svn:executable','*');
2066        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
2067                $self->change_file_prop($fbat,'svn:executable',undef);
2068        }
2069        my $fh = IO::File->new_tmpfile or croak $!;
2070        if ($m->{mode_b} =~ /^120/) {
2071                print $fh 'link ' or croak $!;
2072                $self->change_file_prop($fbat,'svn:special','*');
2073        } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
2074                $self->change_file_prop($fbat,'svn:special',undef);
2075        }
2076        defined(my $pid = fork) or croak $!;
2077        if (!$pid) {
2078                open STDOUT, '>&', $fh or croak $!;
2079                exec qw/git-cat-file blob/, $m->{sha1_b} or croak $!;
2080        }
2081        waitpid $pid, 0;
2082        croak $? if $?;
2083        $fh->flush == 0 or croak $!;
2084        seek $fh, 0, 0 or croak $!;
2085
2086        my $md5 = Digest::MD5->new;
2087        $md5->addfile($fh) or croak $!;
2088        seek $fh, 0, 0 or croak $!;
2089
2090        my $exp = $md5->hexdigest;
2091        my $pool = SVN::Pool->new;
2092        my $atd = $self->apply_textdelta($fbat, undef, $pool);
2093        my $got = SVN::TxDelta::send_stream($fh, @$atd, $pool);
2094        die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
2095        $pool->clear;
2096
2097        close $fh or croak $!;
2098}
2099
2100sub D {
2101        my ($self, $m) = @_;
2102        my ($dir, $file) = split_path($m->{file_b});
2103        my $pbat = $self->ensure_path($dir);
2104        print "\tD\t$m->{file_b}\n" unless $::_q;
2105        $self->delete_entry($m->{file_b}, $pbat);
2106}
2107
2108sub close_edit {
2109        my ($self) = @_;
2110        my ($p,$bat) = ($self->{pool}, $self->{bat});
2111        foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
2112                $self->close_directory($bat->{$_}, $p);
2113        }
2114        $self->SUPER::close_edit($p);
2115        $p->clear;
2116}
2117
2118sub abort_edit {
2119        my ($self) = @_;
2120        $self->SUPER::abort_edit($self->{pool});
2121        $self->{pool}->clear;
2122}
2123
2124# this drives the editor
2125sub apply_diff {
2126        my ($self, $tree_a, $tree_b) = @_;
2127        my @diff_tree = qw(diff-tree -z -r);
2128        if ($::_cp_similarity) {
2129                push @diff_tree, "-C$::_cp_similarity";
2130        } else {
2131                push @diff_tree, '-C';
2132        }
2133        push @diff_tree, '--find-copies-harder' if $::_find_copies_harder;
2134        push @diff_tree, "-l$::_l" if defined $::_l;
2135        push @diff_tree, $tree_a, $tree_b;
2136        my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
2137        my $nl = $/;
2138        local $/ = "\0";
2139        my $state = 'meta';
2140        my @mods;
2141        while (<$diff_fh>) {
2142                chomp $_; # this gets rid of the trailing "\0"
2143                if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
2144                                        $::sha1\s($::sha1)\s
2145                                        ([MTCRAD])\d*$/xo) {
2146                        push @mods, {   mode_a => $1, mode_b => $2,
2147                                        sha1_b => $3, chg => $4 };
2148                        if ($4 =~ /^(?:C|R)$/) {
2149                                $state = 'file_a';
2150                        } else {
2151                                $state = 'file_b';
2152                        }
2153                } elsif ($state eq 'file_a') {
2154                        my $x = $mods[$#mods] or croak "Empty array\n";
2155                        if ($x->{chg} !~ /^(?:C|R)$/) {
2156                                croak "Error parsing $_, $x->{chg}\n";
2157                        }
2158                        $x->{file_a} = $_;
2159                        $state = 'file_b';
2160                } elsif ($state eq 'file_b') {
2161                        my $x = $mods[$#mods] or croak "Empty array\n";
2162                        if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
2163                                croak "Error parsing $_, $x->{chg}\n";
2164                        }
2165                        if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
2166                                croak "Error parsing $_, $x->{chg}\n";
2167                        }
2168                        $x->{file_b} = $_;
2169                        $state = 'meta';
2170                } else {
2171                        croak "Error parsing $_\n";
2172                }
2173        }
2174        command_close_pipe($diff_fh, $ctx);
2175        $/ = $nl;
2176
2177        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
2178        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @mods) {
2179                my $f = $m->{chg};
2180                if (defined $o{$f}) {
2181                        $self->$f($m);
2182                } else {
2183                        fatal("Invalid change type: $f\n");
2184                }
2185        }
2186        $self->rmdirs($tree_b) if $::_rmdir;
2187        if (@mods == 0) {
2188                $self->abort_edit;
2189        } else {
2190                $self->close_edit;
2191        }
2192        \@mods;
2193}
2194
2195package Git::SVN::Ra;
2196use vars qw/@ISA $config_dir/;
2197use strict;
2198use warnings;
2199my ($can_do_switch);
2200my %RA;
2201
2202BEGIN {
2203        # enforce temporary pool usage for some simple functions
2204        my $e;
2205        foreach (qw/get_latest_revnum rev_proplist get_file
2206                    check_path get_dir get_uuid get_repos_root/) {
2207                $e .= "sub $_ {
2208                        my \$self = shift;
2209                        my \$pool = SVN::Pool->new;
2210                        my \@ret = \$self->SUPER::$_(\@_,\$pool);
2211                        \$pool->clear;
2212                        wantarray ? \@ret : \$ret[0]; }\n";
2213        }
2214        eval $e;
2215}
2216
2217sub new {
2218        my ($class, $url) = @_;
2219        $url =~ s!/+$!!;
2220        return $RA{$url} if $RA{$url};
2221
2222        SVN::_Core::svn_config_ensure($config_dir, undef);
2223        my ($baton, $callbacks) = SVN::Core::auth_open_helper([
2224            SVN::Client::get_simple_provider(),
2225            SVN::Client::get_ssl_server_trust_file_provider(),
2226            SVN::Client::get_simple_prompt_provider(
2227              \&Git::SVN::Prompt::simple, 2),
2228            SVN::Client::get_ssl_client_cert_prompt_provider(
2229              \&Git::SVN::Prompt::ssl_client_cert, 2),
2230            SVN::Client::get_ssl_client_cert_pw_prompt_provider(
2231              \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
2232            SVN::Client::get_username_provider(),
2233            SVN::Client::get_ssl_server_trust_prompt_provider(
2234              \&Git::SVN::Prompt::ssl_server_trust),
2235            SVN::Client::get_username_prompt_provider(
2236              \&Git::SVN::Prompt::username, 2),
2237          ]);
2238        my $config = SVN::Core::config_get_config($config_dir);
2239        my $self = SVN::Ra->new(url => $url, auth => $baton,
2240                              config => $config,
2241                              pool => SVN::Pool->new,
2242                              auth_provider_callbacks => $callbacks);
2243        $self->{svn_path} = $url;
2244        $self->{repos_root} = $self->get_repos_root;
2245        $self->{svn_path} =~ s#^\Q$self->{repos_root}\E/*##;
2246        $RA{$url} = bless $self, $class;
2247}
2248
2249sub DESTROY {
2250        # do not call the real DESTROY since we store ourselves in %RA
2251}
2252
2253sub dup {
2254        my ($self) = @_;
2255        my $dup = SVN::Ra->new(pool => SVN::Pool->new,
2256                                map { $_ => $self->{$_} } qw/config url
2257                     auth auth_provider_callbacks repos_root svn_path/);
2258        bless $dup, ref $self;
2259}
2260
2261sub get_log {
2262        my ($self, @args) = @_;
2263        my $pool = SVN::Pool->new;
2264        splice(@args, 3, 1) if ($SVN::Core::VERSION le '1.2.0');
2265        my $ret = $self->SUPER::get_log(@args, $pool);
2266        $pool->clear;
2267        $ret;
2268}
2269
2270sub get_commit_editor {
2271        my ($self, $log, $cb, $pool) = @_;
2272        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
2273        $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
2274}
2275
2276sub uuid {
2277        my ($self) = @_;
2278        $self->{uuid} ||= $self->get_uuid;
2279}
2280
2281sub gs_do_update {
2282        my ($self, $rev_a, $rev_b, $path, $recurse, $editor) = @_;
2283        my $pool = SVN::Pool->new;
2284        $editor->set_path_strip($path);
2285        my $reporter = $self->do_update($rev_b, $path, $recurse,
2286                                        $editor, $pool);
2287        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
2288        my $new = ($rev_a == $rev_b);
2289        $reporter->set_path('', $rev_a, $new, @lock, $pool);
2290        $reporter->finish_report($pool);
2291        $pool->clear;
2292        $editor->{git_commit_ok};
2293}
2294
2295sub gs_do_switch {
2296        my ($self, $rev_a, $rev_b, $path, $recurse, $url_b, $editor) = @_;
2297        my $pool = SVN::Pool->new;
2298        $editor->set_path_strip($path);
2299        my $reporter = $self->do_switch($rev_b, $path, $recurse,
2300                                        $url_b, $editor, $pool);
2301        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
2302        $reporter->set_path('', $rev_a, 0, @lock, $pool);
2303        $reporter->finish_report($pool);
2304        $pool->clear;
2305        $editor->{git_commit_ok};
2306}
2307
2308sub minimize_url {
2309        my ($self) = @_;
2310        return $self->{url} if ($self->{url} eq $self->{repos_root});
2311        my $url = $self->{repos_root};
2312        my @components = split(m!/!, $self->{svn_path});
2313        my $c = '';
2314        do {
2315                $url .= "/$c" if length $c;
2316                eval { (ref $self)->new($url)->get_latest_revnum };
2317        } while ($@ && ($c = shift @components));
2318        $url;
2319}
2320
2321sub can_do_switch {
2322        my $self = shift;
2323        unless (defined $can_do_switch) {
2324                my $pool = SVN::Pool->new;
2325                my $rep = eval {
2326                        $self->do_switch(1, '', 0, $self->{url},
2327                                         SVN::Delta::Editor->new, $pool);
2328                };
2329                if ($@) {
2330                        $can_do_switch = 0;
2331                } else {
2332                        $rep->abort_report($pool);
2333                        $can_do_switch = 1;
2334                }
2335                $pool->clear;
2336        }
2337        $can_do_switch;
2338}
2339
2340package Git::SVN::Log;
2341use strict;
2342use warnings;
2343use POSIX qw/strftime/;
2344use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
2345            %rusers $show_commit $incremental/;
2346my $l_fmt;
2347
2348sub cmt_showable {
2349        my ($c) = @_;
2350        return 1 if defined $c->{r};
2351        if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
2352                                $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
2353                my @log = command(qw/cat-file commit/, $c->{c});
2354                shift @log while ($log[0] ne "\n");
2355                shift @log;
2356                @{$c->{l}} = grep !/^git-svn-id: /, @log;
2357
2358                (undef, $c->{r}, undef) = ::extract_metadata(
2359                                (grep(/^git-svn-id: /, @log))[-1]);
2360        }
2361        return defined $c->{r};
2362}
2363
2364sub log_use_color {
2365        return 1 if $color;
2366        my ($dc, $dcvar);
2367        $dcvar = 'color.diff';
2368        $dc = `git-config --get $dcvar`;
2369        if ($dc eq '') {
2370                # nothing at all; fallback to "diff.color"
2371                $dcvar = 'diff.color';
2372                $dc = `git-config --get $dcvar`;
2373        }
2374        chomp($dc);
2375        if ($dc eq 'auto') {
2376                my $pc;
2377                $pc = `git-config --get color.pager`;
2378                if ($pc eq '') {
2379                        # does not have it -- fallback to pager.color
2380                        $pc = `git-config --bool --get pager.color`;
2381                }
2382                else {
2383                        $pc = `git-config --bool --get color.pager`;
2384                        if ($?) {
2385                                $pc = 'false';
2386                        }
2387                }
2388                chomp($pc);
2389                if (-t *STDOUT || (defined $pager && $pc eq 'true')) {
2390                        return ($ENV{TERM} && $ENV{TERM} ne 'dumb');
2391                }
2392                return 0;
2393        }
2394        return 0 if $dc eq 'never';
2395        return 1 if $dc eq 'always';
2396        chomp($dc = `git-config --bool --get $dcvar`);
2397        return ($dc eq 'true');
2398}
2399
2400sub git_svn_log_cmd {
2401        my ($r_min, $r_max) = @_;
2402        my $gs = Git::SVN->_new;
2403        my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
2404                   $gs->refname);
2405        push @cmd, '-r' unless $non_recursive;
2406        push @cmd, qw/--raw --name-status/ if $verbose;
2407        push @cmd, '--color' if log_use_color();
2408        return @cmd unless defined $r_max;
2409        if ($r_max == $r_min) {
2410                push @cmd, '--max-count=1';
2411                if (my $c = $gs->rev_db_get($r_max)) {
2412                        push @cmd, $c;
2413                }
2414        } else {
2415                my ($c_min, $c_max);
2416                $c_max = $gs->rev_db_get($r_max);
2417                $c_min = $gs->rev_db_get($r_min);
2418                if (defined $c_min && defined $c_max) {
2419                        if ($r_max > $r_max) {
2420                                push @cmd, "$c_min..$c_max";
2421                        } else {
2422                                push @cmd, "$c_max..$c_min";
2423                        }
2424                } elsif ($r_max > $r_min) {
2425                        push @cmd, $c_max;
2426                } else {
2427                        push @cmd, $c_min;
2428                }
2429        }
2430        return @cmd;
2431}
2432
2433# adapted from pager.c
2434sub config_pager {
2435        $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
2436        if (!defined $pager) {
2437                $pager = 'less';
2438        } elsif (length $pager == 0 || $pager eq 'cat') {
2439                $pager = undef;
2440        }
2441}
2442
2443sub run_pager {
2444        return unless -t *STDOUT;
2445        pipe my $rfd, my $wfd or return;
2446        defined(my $pid = fork) or ::fatal "Can't fork: $!\n";
2447        if (!$pid) {
2448                open STDOUT, '>&', $wfd or
2449                                     ::fatal "Can't redirect to stdout: $!\n";
2450                return;
2451        }
2452        open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!\n";
2453        $ENV{LESS} ||= 'FRSX';
2454        exec $pager or ::fatal "Can't run pager: $! ($pager)\n";
2455}
2456
2457sub get_author_info {
2458        my ($dest, $author, $t, $tz) = @_;
2459        $author =~ s/(?:^\s*|\s*$)//g;
2460        $dest->{a_raw} = $author;
2461        my $au;
2462        if ($::_authors) {
2463                $au = $rusers{$author} || undef;
2464        }
2465        if (!$au) {
2466                ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
2467        }
2468        $dest->{t} = $t;
2469        $dest->{tz} = $tz;
2470        $dest->{a} = $au;
2471        # Date::Parse isn't in the standard Perl distro :(
2472        if ($tz =~ s/^\+//) {
2473                $t += ::tz_to_s_offset($tz);
2474        } elsif ($tz =~ s/^\-//) {
2475                $t -= ::tz_to_s_offset($tz);
2476        }
2477        $dest->{t_utc} = $t;
2478}
2479
2480sub process_commit {
2481        my ($c, $r_min, $r_max, $defer) = @_;
2482        if (defined $r_min && defined $r_max) {
2483                if ($r_min == $c->{r} && $r_min == $r_max) {
2484                        show_commit($c);
2485                        return 0;
2486                }
2487                return 1 if $r_min == $r_max;
2488                if ($r_min < $r_max) {
2489                        # we need to reverse the print order
2490                        return 0 if (defined $limit && --$limit < 0);
2491                        push @$defer, $c;
2492                        return 1;
2493                }
2494                if ($r_min != $r_max) {
2495                        return 1 if ($r_min < $c->{r});
2496                        return 1 if ($r_max > $c->{r});
2497                }
2498        }
2499        return 0 if (defined $limit && --$limit < 0);
2500        show_commit($c);
2501        return 1;
2502}
2503
2504sub show_commit {
2505        my $c = shift;
2506        if ($oneline) {
2507                my $x = "\n";
2508                if (my $l = $c->{l}) {
2509                        while ($l->[0] =~ /^\s*$/) { shift @$l }
2510                        $x = $l->[0];
2511                }
2512                $l_fmt ||= 'A' . length($c->{r});
2513                print 'r',pack($l_fmt, $c->{r}),' | ';
2514                print "$c->{c} | " if $show_commit;
2515                print $x;
2516        } else {
2517                show_commit_normal($c);
2518        }
2519}
2520
2521sub show_commit_changed_paths {
2522        my ($c) = @_;
2523        return unless $c->{changed};
2524        print "Changed paths:\n", @{$c->{changed}};
2525}
2526
2527sub show_commit_normal {
2528        my ($c) = @_;
2529        print '-' x72, "\nr$c->{r} | ";
2530        print "$c->{c} | " if $show_commit;
2531        print "$c->{a} | ", strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)",
2532                                 localtime($c->{t_utc})), ' | ';
2533        my $nr_line = 0;
2534
2535        if (my $l = $c->{l}) {
2536                while ($l->[$#$l] eq "\n" && $#$l > 0
2537                                          && $l->[($#$l - 1)] eq "\n") {
2538                        pop @$l;
2539                }
2540                $nr_line = scalar @$l;
2541                if (!$nr_line) {
2542                        print "1 line\n\n\n";
2543                } else {
2544                        if ($nr_line == 1) {
2545                                $nr_line = '1 line';
2546                        } else {
2547                                $nr_line .= ' lines';
2548                        }
2549                        print $nr_line, "\n";
2550                        show_commit_changed_paths($c);
2551                        print "\n";
2552                        print $_ foreach @$l;
2553                }
2554        } else {
2555                print "1 line\n";
2556                show_commit_changed_paths($c);
2557                print "\n";
2558
2559        }
2560        foreach my $x (qw/raw diff/) {
2561                if ($c->{$x}) {
2562                        print "\n";
2563                        print $_ foreach @{$c->{$x}}
2564                }
2565        }
2566}
2567
2568sub cmd_show_log {
2569        my (@args) = @_;
2570        my ($r_min, $r_max);
2571        my $r_last = -1; # prevent dupes
2572        if (defined $TZ) {
2573                $ENV{TZ} = $TZ;
2574        } else {
2575                delete $ENV{TZ};
2576        }
2577        if (defined $::_revision) {
2578                if ($::_revision =~ /^(\d+):(\d+)$/) {
2579                        ($r_min, $r_max) = ($1, $2);
2580                } elsif ($::_revision =~ /^\d+$/) {
2581                        $r_min = $r_max = $::_revision;
2582                } else {
2583                        ::fatal "-r$::_revision is not supported, use ",
2584                                "standard \'git log\' arguments instead\n";
2585                }
2586        }
2587
2588        config_pager();
2589        @args = (git_svn_log_cmd($r_min, $r_max), @args);
2590        my $log = command_output_pipe(@args);
2591        run_pager();
2592        my (@k, $c, $d);
2593        my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
2594        while (<$log>) {
2595                if (/^${esc_color}commit ($::sha1_short)/o) {
2596                        my $cmt = $1;
2597                        if ($c && cmt_showable($c) && $c->{r} != $r_last) {
2598                                $r_last = $c->{r};
2599                                process_commit($c, $r_min, $r_max, \@k) or
2600                                                                goto out;
2601                        }
2602                        $d = undef;
2603                        $c = { c => $cmt };
2604                } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
2605                        get_author_info($c, $1, $2, $3);
2606                } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
2607                        # ignore
2608                } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
2609                        push @{$c->{raw}}, $_;
2610                } elsif (/^${esc_color}[ACRMDT]\t/) {
2611                        # we could add $SVN->{svn_path} here, but that requires
2612                        # remote access at the moment (repo_path_split)...
2613                        s#^(${esc_color})([ACRMDT])\t#$1   $2 #o;
2614                        push @{$c->{changed}}, $_;
2615                } elsif (/^${esc_color}diff /o) {
2616                        $d = 1;
2617                        push @{$c->{diff}}, $_;
2618                } elsif ($d) {
2619                        push @{$c->{diff}}, $_;
2620                } elsif (/^${esc_color}    (git-svn-id:.+)$/o) {
2621                        ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
2622                } elsif (s/^${esc_color}    //o) {
2623                        push @{$c->{l}}, $_;
2624                }
2625        }
2626        if ($c && defined $c->{r} && $c->{r} != $r_last) {
2627                $r_last = $c->{r};
2628                process_commit($c, $r_min, $r_max, \@k);
2629        }
2630        if (@k) {
2631                my $swap = $r_max;
2632                $r_max = $r_min;
2633                $r_min = $swap;
2634                process_commit($_, $r_min, $r_max) foreach reverse @k;
2635        }
2636out:
2637        close $log;
2638        print '-' x72,"\n" unless $incremental || $oneline;
2639}
2640
2641package Git::SVN::Migration;
2642# these version numbers do NOT correspond to actual version numbers
2643# of git nor git-svn.  They are just relative.
2644#
2645# v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
2646#
2647# v1 layout: .git/$id/info/url, refs/remotes/$id
2648#
2649# v2 layout: .git/svn/$id/info/url, refs/remotes/$id
2650#
2651# v3 layout: .git/svn/$id, refs/remotes/$id
2652#            - info/url may remain for backwards compatibility
2653#            - this is what we migrate up to this layout automatically,
2654#            - this will be used by git svn init on single branches
2655#
2656# v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
2657#            - this is only created for newly multi-init-ed
2658#              repositories.  Similar in spirit to the
2659#              --use-separate-remotes option in git-clone (now default)
2660#            - we do not automatically migrate to this (following
2661#              the example set by core git)
2662use strict;
2663use warnings;
2664use Carp qw/croak/;
2665use File::Path qw/mkpath/;
2666use File::Basename qw/dirname basename/;
2667use vars qw/$_minimize/;
2668
2669sub migrate_from_v0 {
2670        my $git_dir = $ENV{GIT_DIR};
2671        return undef unless -d $git_dir;
2672        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
2673        my $migrated = 0;
2674        while (<$fh>) {
2675                chomp;
2676                my ($id, $orig_ref) = ($_, $_);
2677                next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
2678                next unless -f "$git_dir/$id/info/url";
2679                my $new_ref = "refs/remotes/$id";
2680                if (::verify_ref("$new_ref^0")) {
2681                        print STDERR "W: $orig_ref is probably an old ",
2682                                     "branch used by an ancient version of ",
2683                                     "git-svn.\n",
2684                                     "However, $new_ref also exists.\n",
2685                                     "We will not be able ",
2686                                     "to use this branch until this ",
2687                                     "ambiguity is resolved.\n";
2688                        next;
2689                }
2690                print STDERR "Migrating from v0 layout...\n" if !$migrated;
2691                print STDERR "Renaming ref: $orig_ref => $new_ref\n";
2692                command_noisy('update-ref', $new_ref, $orig_ref);
2693                command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
2694                $migrated++;
2695        }
2696        command_close_pipe($fh, $ctx);
2697        print STDERR "Done migrating from v0 layout...\n" if $migrated;
2698        $migrated;
2699}
2700
2701sub migrate_from_v1 {
2702        my $git_dir = $ENV{GIT_DIR};
2703        my $migrated = 0;
2704        return $migrated unless -d $git_dir;
2705        my $svn_dir = "$git_dir/svn";
2706
2707        # just in case somebody used 'svn' as their $id at some point...
2708        return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
2709
2710        print STDERR "Migrating from a git-svn v1 layout...\n";
2711        mkpath([$svn_dir]);
2712        print STDERR "Data from a previous version of git-svn exists, but\n\t",
2713                     "$svn_dir\n\t(required for this version ",
2714                     "($::VERSION) of git-svn) does not. exist\n";
2715        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
2716        while (<$fh>) {
2717                my $x = $_;
2718                next unless $x =~ s#^refs/remotes/##;
2719                chomp $x;
2720                next unless -f "$git_dir/$x/info/url";
2721                my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
2722                next unless $u;
2723                my $dn = dirname("$git_dir/svn/$x");
2724                mkpath([$dn]) unless -d $dn;
2725                if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
2726                        mkpath(["$git_dir/svn/svn"]);
2727                        print STDERR " - $git_dir/$x/info => ",
2728                                        "$git_dir/svn/$x/info\n";
2729                        rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
2730                               croak "$!: $x";
2731                        # don't worry too much about these, they probably
2732                        # don't exist with repos this old (save for index,
2733                        # and we can easily regenerate that)
2734                        foreach my $f (qw/unhandled.log index .rev_db/) {
2735                                rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
2736                        }
2737                } else {
2738                        print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
2739                        rename "$git_dir/$x", "$git_dir/svn/$x" or
2740                               croak "$!: $x";
2741                }
2742                $migrated++;
2743        }
2744        command_close_pipe($fh, $ctx);
2745        print STDERR "Done migrating from a git-svn v1 layout\n";
2746        $migrated;
2747}
2748
2749sub read_old_urls {
2750        my ($l_map, $pfx, $path) = @_;
2751        my @dir;
2752        foreach (<$path/*>) {
2753                if (-r "$_/info/url") {
2754                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
2755                        my $ref_id = $pfx . basename $_;
2756                        my $url = ::file_to_s("$_/info/url");
2757                        $l_map->{$ref_id} = $url;
2758                } elsif (-d $_) {
2759                        push @dir, $_;
2760                }
2761        }
2762        foreach (@dir) {
2763                my $x = $_;
2764                $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
2765                read_old_urls($l_map, $x, $_);
2766        }
2767}
2768
2769sub migrate_from_v2 {
2770        my @cfg = command(qw/config -l/);
2771        return if grep /^svn-remote\..+\.url=/, @cfg;
2772        my %l_map;
2773        read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
2774        my $migrated = 0;
2775
2776        foreach my $ref_id (sort keys %l_map) {
2777                Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
2778                $migrated++;
2779        }
2780        $migrated;
2781}
2782
2783sub minimize_connections {
2784        my $r = Git::SVN::read_all_remotes();
2785        my $new_urls = {};
2786        my $root_repos = {};
2787        foreach my $repo_id (keys %$r) {
2788                my $url = $r->{$repo_id}->{url} or next;
2789                my $fetch = $r->{$repo_id}->{fetch} or next;
2790                my $ra = Git::SVN::Ra->new($url);
2791
2792                # skip existing cases where we already connect to the root
2793                if (($ra->{url} eq $ra->{repos_root}) ||
2794                    (Git::SVN::sanitize_remote_name($ra->{repos_root}) eq
2795                     $repo_id)) {
2796                        $root_repos->{$ra->{url}} = $repo_id;
2797                        next;
2798                }
2799
2800                my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
2801                my $root_path = $ra->{url};
2802                $root_path =~ s#^\Q$ra->{repos_root}\E/*##;
2803                foreach my $path (keys %$fetch) {
2804                        my $ref_id = $fetch->{$path};
2805                        my $gs = Git::SVN->new($ref_id, $repo_id, $path);
2806
2807                        # make sure we can read when connecting to
2808                        # a higher level of a repository
2809                        my ($last_rev, undef) = $gs->last_rev_commit;
2810                        if (!defined $last_rev) {
2811                                $last_rev = eval {
2812                                        $root_ra->get_latest_revnum;
2813                                };
2814                                next if $@;
2815                        }
2816                        my $new = $root_path;
2817                        $new .= length $path ? "/$path" : '';
2818                        eval {
2819                                $root_ra->get_log([$new], $last_rev, $last_rev,
2820                                                  0, 0, 1, sub { });
2821                        };
2822                        next if $@;
2823                        $new_urls->{$ra->{repos_root}}->{$new} =
2824                                { ref_id => $ref_id,
2825                                  old_repo_id => $repo_id,
2826                                  old_path => $path };
2827                }
2828        }
2829
2830        my @emptied;
2831        foreach my $url (keys %$new_urls) {
2832                # see if we can re-use an existing [svn-remote "repo_id"]
2833                # instead of creating a(n ugly) new section:
2834                my $repo_id = $root_repos->{$url} ||
2835                              Git::SVN::sanitize_remote_name($url);
2836
2837                my $fetch = $new_urls->{$url};
2838                foreach my $path (keys %$fetch) {
2839                        my $x = $fetch->{$path};
2840                        Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
2841                        my $pfx = "svn-remote.$x->{old_repo_id}";
2842
2843                        my $old_fetch = quotemeta("$x->{old_path}:".
2844                                                  "refs/remotes/$x->{ref_id}");
2845                        command_noisy(qw/config --unset/,
2846                                      "$pfx.fetch", '^'. $old_fetch . '$');
2847                        delete $r->{$x->{old_repo_id}}->
2848                               {fetch}->{$x->{old_path}};
2849                        if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
2850                                command_noisy(qw/config --unset/,
2851                                              "$pfx.url");
2852                                push @emptied, $x->{old_repo_id}
2853                        }
2854                }
2855        }
2856        if (@emptied) {
2857                my $file = $ENV{GIT_CONFIG} || $ENV{GIT_CONFIG_LOCAL} ||
2858                           "$ENV{GIT_DIR}/config";
2859                print STDERR <<EOF;
2860The following [svn-remote] sections in your config file ($file) are empty
2861and can be safely removed:
2862EOF
2863                print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
2864        }
2865}
2866
2867sub migration_check {
2868        migrate_from_v0();
2869        migrate_from_v1();
2870        migrate_from_v2();
2871        minimize_connections() if $_minimize;
2872}
2873
2874package Git::IndexInfo;
2875use strict;
2876use warnings;
2877use Git qw/command_input_pipe command_close_pipe/;
2878
2879sub new {
2880        my ($class) = @_;
2881        my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
2882        bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
2883}
2884
2885sub remove {
2886        my ($self, $path) = @_;
2887        if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
2888                return ++$self->{nr};
2889        }
2890        undef;
2891}
2892
2893sub update {
2894        my ($self, $mode, $hash, $path) = @_;
2895        if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
2896                return ++$self->{nr};
2897        }
2898        undef;
2899}
2900
2901sub DESTROY {
2902        my ($self) = @_;
2903        command_close_pipe($self->{gui}, $self->{ctx});
2904}
2905
2906__END__
2907
2908Data structures:
2909
2910$log_entry hashref as returned by libsvn_log_entry()
2911{
2912        log => 'whitespace-formatted log entry
2913',                                              # trailing newline is preserved
2914        revision => '8',                        # integer
2915        date => '2004-02-24T17:01:44.108345Z',  # commit date
2916        author => 'committer name'
2917};
2918
2919@mods = array of diff-index line hashes, each element represents one line
2920        of diff-index output
2921
2922diff-index line ($m hash)
2923{
2924        mode_a => first column of diff-index output, no leading ':',
2925        mode_b => second column of diff-index output,
2926        sha1_b => sha1sum of the final blob,
2927        chg => change type [MCRADT],
2928        file_a => original file name of a file (iff chg is 'C' or 'R')
2929        file_b => new/current file name of a file (any chg)
2930}
2931;
2932
2933# retval of read_url_paths{,_all}();
2934$l_map = {
2935        # repository root url
2936        'https://svn.musicpd.org' => {
2937                # repository path               # GIT_SVN_ID
2938                'mpd/trunk'             =>      'trunk',
2939                'mpd/tags/0.11.5'       =>      'tags/0.11.5',
2940        },
2941}
2942
2943Notes:
2944        I don't trust the each() function on unless I created %hash myself
2945        because the internal iterator may not have started at base.