Andrew's git - gitweb.git/blob

   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                $sha1 $sha1_short $_revision
   8                $_q $_authors %users/;
   9$AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
  10$VERSION = '@@GIT_VERSION@@';
  11
  12my $git_dir_user_set = 1 if defined $ENV{GIT_DIR};
  13$ENV{GIT_DIR} ||= '.git';
  14$Git::SVN::default_repo_id = 'svn';
  15$Git::SVN::default_ref_id = $ENV{GIT_SVN_ID} || 'git-svn';
  16$Git::SVN::Ra::_log_window_size = 100;
  17
  18$Git::SVN::Log::TZ = $ENV{TZ};
  19$ENV{TZ} = 'UTC';
  20$| = 1; # unbuffer STDOUT
  21
  22sub fatal (@) { print STDERR @_; exit 1 }
  23require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
  24require SVN::Ra;
  25require SVN::Delta;
  26if ($SVN::Core::VERSION lt '1.1.0') {
  27        fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)\n";
  28}
  29push @Git::SVN::Ra::ISA, 'SVN::Ra';
  30push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
  31push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
  32use Carp qw/croak/;
  33use IO::File qw//;
  34use File::Basename qw/dirname basename/;
  35use File::Path qw/mkpath/;
  36use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev/;
  37use IPC::Open3;
  38use Git;
  39
  40BEGIN {
  41        # import functions from Git into our packages, en masse
  42        no strict 'refs';
  43        foreach (qw/command command_oneline command_noisy command_output_pipe
  44                    command_input_pipe command_close_pipe/) {
  45                for my $package ( qw(SVN::Git::Editor SVN::Git::Fetcher
  46                        Git::SVN::Migration Git::SVN::Log Git::SVN),
  47                        __PACKAGE__) {
  48                        *{"${package}::$_"} = \&{"Git::$_"};
  49                }
  50        }
  51}
  52
  53my ($SVN);
  54
  55$sha1 = qr/[a-f\d]{40}/;
  56$sha1_short = qr/[a-f\d]{4,40}/;
  57my ($_stdin, $_help, $_edit,
  58        $_message, $_file,
  59        $_template, $_shared,
  60        $_version, $_fetch_all, $_no_rebase,
  61        $_merge, $_strategy, $_dry_run, $_local,
  62        $_prefix, $_no_checkout, $_verbose);
  63$Git::SVN::_follow_parent = 1;
  64my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
  65                    'config-dir=s' => \$Git::SVN::Ra::config_dir,
  66                    'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache );
  67my %fc_opts = ( 'follow-parent|follow!' => \$Git::SVN::_follow_parent,
  68                'authors-file|A=s' => \$_authors,
  69                'repack:i' => \$Git::SVN::_repack,
  70                'noMetadata' => \$Git::SVN::_no_metadata,
  71                'useSvmProps' => \$Git::SVN::_use_svm_props,
  72                'useSvnsyncProps' => \$Git::SVN::_use_svnsync_props,
  73                'log-window-size=i' => \$Git::SVN::Ra::_log_window_size,
  74                'no-checkout' => \$_no_checkout,
  75                'quiet|q' => \$_q,
  76                'repack-flags|repack-args|repack-opts=s' =>
  77                   \$Git::SVN::_repack_flags,
  78                %remote_opts );
  79
  80my ($_trunk, $_tags, $_branches);
  81my %icv;
  82my %init_opts = ( 'template=s' => \$_template, 'shared:s' => \$_shared,
  83                  'trunk|T=s' => \$_trunk, 'tags|t=s' => \$_tags,
  84                  'branches|b=s' => \$_branches, 'prefix=s' => \$_prefix,
  85                  'minimize-url|m' => \$Git::SVN::_minimize_url,
  86                  'no-metadata' => sub { $icv{noMetadata} = 1 },
  87                  'use-svm-props' => sub { $icv{useSvmProps} = 1 },
  88                  'use-svnsync-props' => sub { $icv{useSvnsyncProps} = 1 },
  89                  'rewrite-root=s' => sub { $icv{rewriteRoot} = $_[1] },
  90                  %remote_opts );
  91my %cmt_opts = ( 'edit|e' => \$_edit,
  92                'rmdir' => \$SVN::Git::Editor::_rmdir,
  93                'find-copies-harder' => \$SVN::Git::Editor::_find_copies_harder,
  94                'l=i' => \$SVN::Git::Editor::_rename_limit,
  95                'copy-similarity|C=i'=> \$SVN::Git::Editor::_cp_similarity
  96);
  97
  98my %cmd = (
  99        fetch => [ \&cmd_fetch, "Download new revisions from SVN",
 100                        { 'revision|r=s' => \$_revision,
 101                          'fetch-all|all' => \$_fetch_all,
 102                           %fc_opts } ],
 103        clone => [ \&cmd_clone, "Initialize and fetch revisions",
 104                        { 'revision|r=s' => \$_revision,
 105                           %fc_opts, %init_opts } ],
 106        init => [ \&cmd_init, "Initialize a repo for tracking" .
 107                          " (requires URL argument)",
 108                          \%init_opts ],
 109        'multi-init' => [ \&cmd_multi_init,
 110                          "Deprecated alias for ".
 111                          "'$0 init -T<trunk> -b<branches> -t<tags>'",
 112                          \%init_opts ],
 113        dcommit => [ \&cmd_dcommit,
 114                     'Commit several diffs to merge with upstream',
 115                        { 'merge|m|M' => \$_merge,
 116                          'strategy|s=s' => \$_strategy,
 117                          'verbose|v' => \$_verbose,
 118                          'dry-run|n' => \$_dry_run,
 119                          'fetch-all|all' => \$_fetch_all,
 120                          'no-rebase' => \$_no_rebase,
 121                        %cmt_opts, %fc_opts } ],
 122        'set-tree' => [ \&cmd_set_tree,
 123                        "Set an SVN repository to a git tree-ish",
 124                        { 'stdin|' => \$_stdin, %cmt_opts, %fc_opts, } ],
 125        'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
 126                        { 'revision|r=i' => \$_revision } ],
 127        'multi-fetch' => [ \&cmd_multi_fetch,
 128                           "Deprecated alias for $0 fetch --all",
 129                           { 'revision|r=s' => \$_revision, %fc_opts } ],
 130        'migrate' => [ sub { },
 131                       # no-op, we automatically run this anyways,
 132                       'Migrate configuration/metadata/layout from
 133                        previous versions of git-svn',
 134                       { 'minimize' => \$Git::SVN::Migration::_minimize,
 135                         %remote_opts } ],
 136        'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
 137                        { 'limit=i' => \$Git::SVN::Log::limit,
 138                          'revision|r=s' => \$_revision,
 139                          'verbose|v' => \$Git::SVN::Log::verbose,
 140                          'incremental' => \$Git::SVN::Log::incremental,
 141                          'oneline' => \$Git::SVN::Log::oneline,
 142                          'show-commit' => \$Git::SVN::Log::show_commit,
 143                          'non-recursive' => \$Git::SVN::Log::non_recursive,
 144                          'authors-file|A=s' => \$_authors,
 145                          'color' => \$Git::SVN::Log::color,
 146                          'pager=s' => \$Git::SVN::Log::pager,
 147                        } ],
 148        'find-rev' => [ \&cmd_find_rev, "Translate between SVN revision numbers and tree-ish",
 149                        { } ],
 150        'rebase' => [ \&cmd_rebase, "Fetch and rebase your working directory",
 151                        { 'merge|m|M' => \$_merge,
 152                          'verbose|v' => \$_verbose,
 153                          'strategy|s=s' => \$_strategy,
 154                          'local|l' => \$_local,
 155                          'fetch-all|all' => \$_fetch_all,
 156                          %fc_opts } ],
 157        'commit-diff' => [ \&cmd_commit_diff,
 158                           'Commit a diff between two trees',
 159                        { 'message|m=s' => \$_message,
 160                          'file|F=s' => \$_file,
 161                          'revision|r=s' => \$_revision,
 162                        %cmt_opts } ],
 163);
 164
 165my $cmd;
 166for (my $i = 0; $i < @ARGV; $i++) {
 167        if (defined $cmd{$ARGV[$i]}) {
 168                $cmd = $ARGV[$i];
 169                splice @ARGV, $i, 1;
 170                last;
 171        }
 172};
 173
 174my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
 175
 176read_repo_config(\%opts);
 177Getopt::Long::Configure('pass_through') if ($cmd && $cmd eq 'log');
 178my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
 179                    'minimize-connections' => \$Git::SVN::Migration::_minimize,
 180                    'id|i=s' => \$Git::SVN::default_ref_id,
 181                    'svn-remote|remote|R=s' => sub {
 182                       $Git::SVN::no_reuse_existing = 1;
 183                       $Git::SVN::default_repo_id = $_[1] });
 184exit 1 if (!$rv && $cmd && $cmd ne 'log');
 185
 186usage(0) if $_help;
 187version() if $_version;
 188usage(1) unless defined $cmd;
 189load_authors() if $_authors;
 190
 191# make sure we're always running
 192unless ($cmd =~ /(?:clone|init|multi-init)$/) {
 193        unless (-d $ENV{GIT_DIR}) {
 194                if ($git_dir_user_set) {
 195                        die "GIT_DIR=$ENV{GIT_DIR} explicitly set, ",
 196                            "but it is not a directory\n";
 197                }
 198                my $git_dir = delete $ENV{GIT_DIR};
 199                chomp(my $cdup = command_oneline(qw/rev-parse --show-cdup/));
 200                unless (length $cdup) {
 201                        die "Already at toplevel, but $git_dir ",
 202                            "not found '$cdup'\n";
 203                }
 204                chdir $cdup or die "Unable to chdir up to '$cdup'\n";
 205                unless (-d $git_dir) {
 206                        die "$git_dir still not found after going to ",
 207                            "'$cdup'\n";
 208                }
 209                $ENV{GIT_DIR} = $git_dir;
 210        }
 211}
 212unless ($cmd =~ /^(?:clone|init|multi-init|commit-diff)$/) {
 213        Git::SVN::Migration::migration_check();
 214}
 215Git::SVN::init_vars();
 216eval {
 217        Git::SVN::verify_remotes_sanity();
 218        $cmd{$cmd}->[0]->(@ARGV);
 219};
 220fatal $@ if $@;
 221post_fetch_checkout();
 222exit 0;
 223
 224####################### primary functions ######################
 225sub usage {
 226        my $exit = shift || 0;
 227        my $fd = $exit ? \*STDERR : \*STDOUT;
 228        print $fd <<"";
 229git-svn - bidirectional operations between a single Subversion tree and git
 230Usage: $0 <command> [options] [arguments]\n
 231
 232        print $fd "Available commands:\n" unless $cmd;
 233
 234        foreach (sort keys %cmd) {
 235                next if $cmd && $cmd ne $_;
 236                next if /^multi-/; # don't show deprecated commands
 237                print $fd '  ',pack('A17',$_),$cmd{$_}->[1],"\n";
 238                foreach (keys %{$cmd{$_}->[2]}) {
 239                        # mixed-case options are for .git/config only
 240                        next if /[A-Z]/ && /^[a-z]+$/i;
 241                        # prints out arguments as they should be passed:
 242                        my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
 243                        print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
 244                                                        "--$_" : "-$_" }
 245                                                split /\|/,$_)," $x\n";
 246                }
 247        }
 248        print $fd <<"";
 249\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
 250arbitrary identifier if you're tracking multiple SVN branches/repositories in
 251one git repository and want to keep them separate.  See git-svn(1) for more
 252information.
 253
 254        exit $exit;
 255}
 256
 257sub version {
 258        print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
 259        exit 0;
 260}
 261
 262sub do_git_init_db {
 263        unless (-d $ENV{GIT_DIR}) {
 264                my @init_db = ('init');
 265                push @init_db, "--template=$_template" if defined $_template;
 266                if (defined $_shared) {
 267                        if ($_shared =~ /[a-z]/) {
 268                                push @init_db, "--shared=$_shared";
 269                        } else {
 270                                push @init_db, "--shared";
 271                        }
 272                }
 273                command_noisy(@init_db);
 274        }
 275        my $set;
 276        my $pfx = "svn-remote.$Git::SVN::default_repo_id";
 277        foreach my $i (keys %icv) {
 278                die "'$set' and '$i' cannot both be set\n" if $set;
 279                next unless defined $icv{$i};
 280                command_noisy('config', "$pfx.$i", $icv{$i});
 281                $set = $i;
 282        }
 283}
 284
 285sub init_subdir {
 286        my $repo_path = shift or return;
 287        mkpath([$repo_path]) unless -d $repo_path;
 288        chdir $repo_path or die "Couldn't chdir to $repo_path: $!\n";
 289        $ENV{GIT_DIR} = '.git';
 290}
 291
 292sub cmd_clone {
 293        my ($url, $path) = @_;
 294        if (!defined $path &&
 295            (defined $_trunk || defined $_branches || defined $_tags) &&
 296            $url !~ m#^[a-z\+]+://#) {
 297                $path = $url;
 298        }
 299        $path = basename($url) if !defined $path || !length $path;
 300        cmd_init($url, $path);
 301        Git::SVN::fetch_all($Git::SVN::default_repo_id);
 302}
 303
 304sub cmd_init {
 305        if (defined $_trunk || defined $_branches || defined $_tags) {
 306                return cmd_multi_init(@_);
 307        }
 308        my $url = shift or die "SVN repository location required ",
 309                               "as a command-line argument\n";
 310        init_subdir(@_);
 311        do_git_init_db();
 312
 313        Git::SVN->init($url);
 314}
 315
 316sub cmd_fetch {
 317        if (grep /^\d+=./, @_) {
 318                die "'<rev>=<commit>' fetch arguments are ",
 319                    "no longer supported.\n";
 320        }
 321        my ($remote) = @_;
 322        if (@_ > 1) {
 323                die "Usage: $0 fetch [--all] [svn-remote]\n";
 324        }
 325        $remote ||= $Git::SVN::default_repo_id;
 326        if ($_fetch_all) {
 327                cmd_multi_fetch();
 328        } else {
 329                Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
 330        }
 331}
 332
 333sub cmd_set_tree {
 334        my (@commits) = @_;
 335        if ($_stdin || !@commits) {
 336                print "Reading from stdin...\n";
 337                @commits = ();
 338                while (<STDIN>) {
 339                        if (/\b($sha1_short)\b/o) {
 340                                unshift @commits, $1;
 341                        }
 342                }
 343        }
 344        my @revs;
 345        foreach my $c (@commits) {
 346                my @tmp = command('rev-parse',$c);
 347                if (scalar @tmp == 1) {
 348                        push @revs, $tmp[0];
 349                } elsif (scalar @tmp > 1) {
 350                        push @revs, reverse(command('rev-list',@tmp));
 351                } else {
 352                        fatal "Failed to rev-parse $c\n";
 353                }
 354        }
 355        my $gs = Git::SVN->new;
 356        my ($r_last, $cmt_last) = $gs->last_rev_commit;
 357        $gs->fetch;
 358        if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
 359                fatal "There are new revisions that were fetched ",
 360                      "and need to be merged (or acknowledged) ",
 361                      "before committing.\nlast rev: $r_last\n",
 362                      " current: $gs->{last_rev}\n";
 363        }
 364        $gs->set_tree($_) foreach @revs;
 365        print "Done committing ",scalar @revs," revisions to SVN\n";
 366}
 367
 368sub cmd_dcommit {
 369        my $head = shift;
 370        $head ||= 'HEAD';
 371        my @refs;
 372        my ($url, $rev, $uuid, $gs) = working_head_info($head, \@refs);
 373        unless ($gs) {
 374                die "Unable to determine upstream SVN information from ",
 375                    "$head history\n";
 376        }
 377        my $c = $refs[-1];
 378        my $last_rev;
 379        foreach my $d (@refs) {
 380                if (!verify_ref("$d~1")) {
 381                        fatal "Commit $d\n",
 382                              "has no parent commit, and therefore ",
 383                              "nothing to diff against.\n",
 384                              "You should be working from a repository ",
 385                              "originally created by git-svn\n";
 386                }
 387                unless (defined $last_rev) {
 388                        (undef, $last_rev, undef) = cmt_metadata("$d~1");
 389                        unless (defined $last_rev) {
 390                                fatal "Unable to extract revision information ",
 391                                      "from commit $d~1\n";
 392                        }
 393                }
 394                if ($_dry_run) {
 395                        print "diff-tree $d~1 $d\n";
 396                } else {
 397                        my %ed_opts = ( r => $last_rev,
 398                                        log => get_commit_entry($d)->{log},
 399                                        ra => Git::SVN::Ra->new($gs->full_url),
 400                                        tree_a => "$d~1",
 401                                        tree_b => $d,
 402                                        editor_cb => sub {
 403                                               print "Committed r$_[0]\n";
 404                                               $last_rev = $_[0]; },
 405                                        svn_path => '');
 406                        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
 407                                print "No changes\n$d~1 == $d\n";
 408                        }
 409                }
 410        }
 411        return if $_dry_run;
 412        unless ($gs) {
 413                warn "Could not determine fetch information for $url\n",
 414                     "Will not attempt to fetch and rebase commits.\n",
 415                     "This probably means you have useSvmProps and should\n",
 416                     "now resync your SVN::Mirror repository.\n";
 417                return;
 418        }
 419        $_fetch_all ? $gs->fetch_all : $gs->fetch;
 420        unless ($_no_rebase) {
 421                # we always want to rebase against the current HEAD, not any
 422                # head that was passed to us
 423                my @diff = command('diff-tree', 'HEAD', $gs->refname, '--');
 424                my @finish;
 425                if (@diff) {
 426                        @finish = rebase_cmd();
 427                        print STDERR "W: HEAD and ", $gs->refname, " differ, ",
 428                                     "using @finish:\n", "@diff";
 429                } else {
 430                        print "No changes between current HEAD and ",
 431                              $gs->refname, "\nResetting to the latest ",
 432                              $gs->refname, "\n";
 433                        @finish = qw/reset --mixed/;
 434                }
 435                command_noisy(@finish, $gs->refname);
 436        }
 437}
 438
 439sub cmd_find_rev {
 440        my $revision_or_hash = shift;
 441        my $result;
 442        if ($revision_or_hash =~ /^r\d+$/) {
 443                my $head = shift;
 444                $head ||= 'HEAD';
 445                my @refs;
 446                my (undef, undef, undef, $gs) = working_head_info($head, \@refs);
 447                unless ($gs) {
 448                        die "Unable to determine upstream SVN information from ",
 449                            "$head history\n";
 450                }
 451                my $desired_revision = substr($revision_or_hash, 1);
 452                $result = $gs->rev_db_get($desired_revision);
 453        } else {
 454                my (undef, $rev, undef) = cmt_metadata($revision_or_hash);
 455                $result = $rev;
 456        }
 457        print "$result\n" if $result;
 458}
 459
 460sub cmd_rebase {
 461        command_noisy(qw/update-index --refresh/);
 462        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 463        unless ($gs) {
 464                die "Unable to determine upstream SVN information from ",
 465                    "working tree history\n";
 466        }
 467        if (command(qw/diff-index HEAD --/)) {
 468                print STDERR "Cannot rebase with uncommited changes:\n";
 469                command_noisy('status');
 470                exit 1;
 471        }
 472        unless ($_local) {
 473                $_fetch_all ? $gs->fetch_all : $gs->fetch;
 474        }
 475        command_noisy(rebase_cmd(), $gs->refname);
 476}
 477
 478sub cmd_show_ignore {
 479        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 480        $gs ||= Git::SVN->new;
 481        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 482        $gs->traverse_ignore(\*STDOUT, $gs->{path}, $r);
 483}
 484
 485sub cmd_multi_init {
 486        my $url = shift;
 487        unless (defined $_trunk || defined $_branches || defined $_tags) {
 488                usage(1);
 489        }
 490
 491        # there are currently some bugs that prevent multi-init/multi-fetch
 492        # setups from working well without this.
 493        $Git::SVN::_minimize_url = 1;
 494
 495        $_prefix = '' unless defined $_prefix;
 496        if (defined $url) {
 497                $url =~ s#/+$##;
 498                init_subdir(@_);
 499        }
 500        do_git_init_db();
 501        if (defined $_trunk) {
 502                my $trunk_ref = $_prefix . 'trunk';
 503                # try both old-style and new-style lookups:
 504                my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
 505                unless ($gs_trunk) {
 506                        my ($trunk_url, $trunk_path) =
 507                                              complete_svn_url($url, $_trunk);
 508                        $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
 509                                                   undef, $trunk_ref);
 510                }
 511        }
 512        return unless defined $_branches || defined $_tags;
 513        my $ra = $url ? Git::SVN::Ra->new($url) : undef;
 514        complete_url_ls_init($ra, $_branches, '--branches/-b', $_prefix);
 515        complete_url_ls_init($ra, $_tags, '--tags/-t', $_prefix . 'tags/');
 516}
 517
 518sub cmd_multi_fetch {
 519        my $remotes = Git::SVN::read_all_remotes();
 520        foreach my $repo_id (sort keys %$remotes) {
 521                if ($remotes->{$repo_id}->{url}) {
 522                        Git::SVN::fetch_all($repo_id, $remotes);
 523                }
 524        }
 525}
 526
 527# this command is special because it requires no metadata
 528sub cmd_commit_diff {
 529        my ($ta, $tb, $url) = @_;
 530        my $usage = "Usage: $0 commit-diff -r<revision> ".
 531                    "<tree-ish> <tree-ish> [<URL>]\n";
 532        fatal($usage) if (!defined $ta || !defined $tb);
 533        my $svn_path;
 534        if (!defined $url) {
 535                my $gs = eval { Git::SVN->new };
 536                if (!$gs) {
 537                        fatal("Needed URL or usable git-svn --id in ",
 538                              "the command-line\n", $usage);
 539                }
 540                $url = $gs->{url};
 541                $svn_path = $gs->{path};
 542        }
 543        unless (defined $_revision) {
 544                fatal("-r|--revision is a required argument\n", $usage);
 545        }
 546        if (defined $_message && defined $_file) {
 547                fatal("Both --message/-m and --file/-F specified ",
 548                      "for the commit message.\n",
 549                      "I have no idea what you mean\n");
 550        }
 551        if (defined $_file) {
 552                $_message = file_to_s($_file);
 553        } else {
 554                $_message ||= get_commit_entry($tb)->{log};
 555        }
 556        my $ra ||= Git::SVN::Ra->new($url);
 557        $svn_path ||= $ra->{svn_path};
 558        my $r = $_revision;
 559        if ($r eq 'HEAD') {
 560                $r = $ra->get_latest_revnum;
 561        } elsif ($r !~ /^\d+$/) {
 562                die "revision argument: $r not understood by git-svn\n";
 563        }
 564        my %ed_opts = ( r => $r,
 565                        log => $_message,
 566                        ra => $ra,
 567                        tree_a => $ta,
 568                        tree_b => $tb,
 569                        editor_cb => sub { print "Committed r$_[0]\n" },
 570                        svn_path => $svn_path );
 571        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
 572                print "No changes\n$ta == $tb\n";
 573        }
 574}
 575
 576########################### utility functions #########################
 577
 578sub rebase_cmd {
 579        my @cmd = qw/rebase/;
 580        push @cmd, '-v' if $_verbose;
 581        push @cmd, qw/--merge/ if $_merge;
 582        push @cmd, "--strategy=$_strategy" if $_strategy;
 583        @cmd;
 584}
 585
 586sub post_fetch_checkout {
 587        return if $_no_checkout;
 588        my $gs = $Git::SVN::_head or return;
 589        return if verify_ref('refs/heads/master^0');
 590
 591        my $valid_head = verify_ref('HEAD^0');
 592        command_noisy(qw(update-ref refs/heads/master), $gs->refname);
 593        return if ($valid_head || !verify_ref('HEAD^0'));
 594
 595        return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
 596        my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
 597        return if -f $index;
 598
 599        chomp(my $bare = `git config --bool --get core.bare`);
 600        return if $bare eq 'true';
 601        return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
 602        command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
 603        print STDERR "Checked out HEAD:\n  ",
 604                     $gs->full_url, " r", $gs->last_rev, "\n";
 605}
 606
 607sub complete_svn_url {
 608        my ($url, $path) = @_;
 609        $path =~ s#/+$##;
 610        if ($path !~ m#^[a-z\+]+://#) {
 611                if (!defined $url || $url !~ m#^[a-z\+]+://#) {
 612                        fatal("E: '$path' is not a complete URL ",
 613                              "and a separate URL is not specified\n");
 614                }
 615                return ($url, $path);
 616        }
 617        return ($path, '');
 618}
 619
 620sub complete_url_ls_init {
 621        my ($ra, $repo_path, $switch, $pfx) = @_;
 622        unless ($repo_path) {
 623                print STDERR "W: $switch not specified\n";
 624                return;
 625        }
 626        $repo_path =~ s#/+$##;
 627        if ($repo_path =~ m#^[a-z\+]+://#) {
 628                $ra = Git::SVN::Ra->new($repo_path);
 629                $repo_path = '';
 630        } else {
 631                $repo_path =~ s#^/+##;
 632                unless ($ra) {
 633                        fatal("E: '$repo_path' is not a complete URL ",
 634                              "and a separate URL is not specified\n");
 635                }
 636        }
 637        my $url = $ra->{url};
 638        my $gs = Git::SVN->init($url, undef, undef, undef, 1);
 639        my $k = "svn-remote.$gs->{repo_id}.url";
 640        my $orig_url = eval { command_oneline(qw/config --get/, $k) };
 641        if ($orig_url && ($orig_url ne $gs->{url})) {
 642                die "$k already set: $orig_url\n",
 643                    "wanted to set to: $gs->{url}\n";
 644        }
 645        command_oneline('config', $k, $gs->{url}) unless $orig_url;
 646        my $remote_path = "$ra->{svn_path}/$repo_path/*";
 647        $remote_path =~ s#/+#/#g;
 648        $remote_path =~ s#^/##g;
 649        my ($n) = ($switch =~ /^--(\w+)/);
 650        if (length $pfx && $pfx !~ m#/$#) {
 651                die "--prefix='$pfx' must have a trailing slash '/'\n";
 652        }
 653        command_noisy('config', "svn-remote.$gs->{repo_id}.$n",
 654                                "$remote_path:refs/remotes/$pfx*");
 655}
 656
 657sub verify_ref {
 658        my ($ref) = @_;
 659        eval { command_oneline([ 'rev-parse', '--verify', $ref ],
 660                               { STDERR => 0 }); };
 661}
 662
 663sub get_tree_from_treeish {
 664        my ($treeish) = @_;
 665        # $treeish can be a symbolic ref, too:
 666        my $type = command_oneline(qw/cat-file -t/, $treeish);
 667        my $expected;
 668        while ($type eq 'tag') {
 669                ($treeish, $type) = command(qw/cat-file tag/, $treeish);
 670        }
 671        if ($type eq 'commit') {
 672                $expected = (grep /^tree /, command(qw/cat-file commit/,
 673                                                    $treeish))[0];
 674                ($expected) = ($expected =~ /^tree ($sha1)$/o);
 675                die "Unable to get tree from $treeish\n" unless $expected;
 676        } elsif ($type eq 'tree') {
 677                $expected = $treeish;
 678        } else {
 679                die "$treeish is a $type, expected tree, tag or commit\n";
 680        }
 681        return $expected;
 682}
 683
 684sub get_commit_entry {
 685        my ($treeish) = shift;
 686        my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
 687        my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
 688        my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
 689        open my $log_fh, '>', $commit_editmsg or croak $!;
 690
 691        my $type = command_oneline(qw/cat-file -t/, $treeish);
 692        if ($type eq 'commit' || $type eq 'tag') {
 693                my ($msg_fh, $ctx) = command_output_pipe('cat-file',
 694                                                         $type, $treeish);
 695                my $in_msg = 0;
 696                while (<$msg_fh>) {
 697                        if (!$in_msg) {
 698                                $in_msg = 1 if (/^\s*$/);
 699                        } elsif (/^git-svn-id: /) {
 700                                # skip this for now, we regenerate the
 701                                # correct one on re-fetch anyways
 702                                # TODO: set *:merge properties or like...
 703                        } else {
 704                                print $log_fh $_ or croak $!;
 705                        }
 706                }
 707                command_close_pipe($msg_fh, $ctx);
 708        }
 709        close $log_fh or croak $!;
 710
 711        if ($_edit || ($type eq 'tree')) {
 712                my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
 713                # TODO: strip out spaces, comments, like git-commit.sh
 714                system($editor, $commit_editmsg);
 715        }
 716        rename $commit_editmsg, $commit_msg or croak $!;
 717        open $log_fh, '<', $commit_msg or croak $!;
 718        { local $/; chomp($log_entry{log} = <$log_fh>); }
 719        close $log_fh or croak $!;
 720        unlink $commit_msg;
 721        \%log_entry;
 722}
 723
 724sub s_to_file {
 725        my ($str, $file, $mode) = @_;
 726        open my $fd,'>',$file or croak $!;
 727        print $fd $str,"\n" or croak $!;
 728        close $fd or croak $!;
 729        chmod ($mode &~ umask, $file) if (defined $mode);
 730}
 731
 732sub file_to_s {
 733        my $file = shift;
 734        open my $fd,'<',$file or croak "$!: file: $file\n";
 735        local $/;
 736        my $ret = <$fd>;
 737        close $fd or croak $!;
 738        $ret =~ s/\s*$//s;
 739        return $ret;
 740}
 741
 742# '<svn username> = real-name <email address>' mapping based on git-svnimport:
 743sub load_authors {
 744        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
 745        my $log = $cmd eq 'log';
 746        while (<$authors>) {
 747                chomp;
 748                next unless /^(\S+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
 749                my ($user, $name, $email) = ($1, $2, $3);
 750                if ($log) {
 751                        $Git::SVN::Log::rusers{"$name <$email>"} = $user;
 752                } else {
 753                        $users{$user} = [$name, $email];
 754                }
 755        }
 756        close $authors or croak $!;
 757}
 758
 759# convert GetOpt::Long specs for use by git-config
 760sub read_repo_config {
 761        return unless -d $ENV{GIT_DIR};
 762        my $opts = shift;
 763        my @config_only;
 764        foreach my $o (keys %$opts) {
 765                # if we have mixedCase and a long option-only, then
 766                # it's a config-only variable that we don't need for
 767                # the command-line.
 768                push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
 769                my $v = $opts->{$o};
 770                my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
 771                $key =~ s/-//g;
 772                my $arg = 'git-config';
 773                $arg .= ' --int' if ($o =~ /[:=]i$/);
 774                $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
 775                if (ref $v eq 'ARRAY') {
 776                        chomp(my @tmp = `$arg --get-all svn.$key`);
 777                        @$v = @tmp if @tmp;
 778                } else {
 779                        chomp(my $tmp = `$arg --get svn.$key`);
 780                        if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
 781                                $$v = $tmp;
 782                        }
 783                }
 784        }
 785        delete @$opts{@config_only} if @config_only;
 786}
 787
 788sub extract_metadata {
 789        my $id = shift or return (undef, undef, undef);
 790        my ($url, $rev, $uuid) = ($id =~ /^git-svn-id:\s(\S+?)\@(\d+)
 791                                                        \s([a-f\d\-]+)$/x);
 792        if (!defined $rev || !$uuid || !$url) {
 793                # some of the original repositories I made had
 794                # identifiers like this:
 795                ($rev, $uuid) = ($id =~/^git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
 796        }
 797        return ($url, $rev, $uuid);
 798}
 799
 800sub cmt_metadata {
 801        return extract_metadata((grep(/^git-svn-id: /,
 802                command(qw/cat-file commit/, shift)))[-1]);
 803}
 804
 805sub working_head_info {
 806        my ($head, $refs) = @_;
 807        my ($fh, $ctx) = command_output_pipe('rev-list', $head);
 808        while (my $hash = <$fh>) {
 809                chomp($hash);
 810                my ($url, $rev, $uuid) = cmt_metadata($hash);
 811                if (defined $url && defined $rev) {
 812                        if (my $gs = Git::SVN->find_by_url($url)) {
 813                                my $c = $gs->rev_db_get($rev);
 814                                if ($c && $c eq $hash) {
 815                                        close $fh; # break the pipe
 816                                        return ($url, $rev, $uuid, $gs);
 817                                }
 818                        }
 819                }
 820                unshift @$refs, $hash if $refs;
 821        }
 822        command_close_pipe($fh, $ctx);
 823        (undef, undef, undef, undef);
 824}
 825
 826package Git::SVN;
 827use strict;
 828use warnings;
 829use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
 830            $_repack $_repack_flags $_use_svm_props $_head
 831            $_use_svnsync_props $no_reuse_existing $_minimize_url/;
 832use Carp qw/croak/;
 833use File::Path qw/mkpath/;
 834use File::Copy qw/copy/;
 835use IPC::Open3;
 836
 837my $_repack_nr;
 838# properties that we do not log:
 839my %SKIP_PROP;
 840BEGIN {
 841        %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
 842                                        svn:special svn:executable
 843                                        svn:entry:committed-rev
 844                                        svn:entry:last-author
 845                                        svn:entry:uuid
 846                                        svn:entry:committed-date/;
 847
 848        # some options are read globally, but can be overridden locally
 849        # per [svn-remote "..."] section.  Command-line options will *NOT*
 850        # override options set in an [svn-remote "..."] section
 851        no strict 'refs';
 852        for my $option (qw/follow_parent no_metadata use_svm_props
 853                           use_svnsync_props/) {
 854                my $key = $option;
 855                $key =~ tr/_//d;
 856                my $prop = "-$option";
 857                *$option = sub {
 858                        my ($self) = @_;
 859                        return $self->{$prop} if exists $self->{$prop};
 860                        my $k = "svn-remote.$self->{repo_id}.$key";
 861                        eval { command_oneline(qw/config --get/, $k) };
 862                        if ($@) {
 863                                $self->{$prop} = ${"Git::SVN::_$option"};
 864                        } else {
 865                                my $v = command_oneline(qw/config --bool/,$k);
 866                                $self->{$prop} = $v eq 'false' ? 0 : 1;
 867                        }
 868                        return $self->{$prop};
 869                }
 870        }
 871}
 872
 873my %LOCKFILES;
 874END { unlink keys %LOCKFILES if %LOCKFILES }
 875
 876sub resolve_local_globs {
 877        my ($url, $fetch, $glob_spec) = @_;
 878        return unless defined $glob_spec;
 879        my $ref = $glob_spec->{ref};
 880        my $path = $glob_spec->{path};
 881        foreach (command(qw#for-each-ref --format=%(refname) refs/remotes#)) {
 882                next unless m#^refs/remotes/$ref->{regex}$#;
 883                my $p = $1;
 884                my $pathname = $path->full_path($p);
 885                my $refname = $ref->full_path($p);
 886                if (my $existing = $fetch->{$pathname}) {
 887                        if ($existing ne $refname) {
 888                                die "Refspec conflict:\n",
 889                                    "existing: refs/remotes/$existing\n",
 890                                    " globbed: refs/remotes/$refname\n";
 891                        }
 892                        my $u = (::cmt_metadata("refs/remotes/$refname"))[0];
 893                        $u =~ s!^\Q$url\E(/|$)!! or die
 894                          "refs/remotes/$refname: '$url' not found in '$u'\n";
 895                        if ($pathname ne $u) {
 896                                warn "W: Refspec glob conflict ",
 897                                     "(ref: refs/remotes/$refname):\n",
 898                                     "expected path: $pathname\n",
 899                                     "    real path: $u\n",
 900                                     "Continuing ahead with $u\n";
 901                                next;
 902                        }
 903                } else {
 904                        $fetch->{$pathname} = $refname;
 905                }
 906        }
 907}
 908
 909sub parse_revision_argument {
 910        my ($base, $head) = @_;
 911        if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
 912                return ($base, $head);
 913        }
 914        return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
 915        return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
 916        return ($head, $head) if ($::_revision eq 'HEAD');
 917        return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
 918        return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
 919        die "revision argument: $::_revision not understood by git-svn\n";
 920}
 921
 922sub fetch_all {
 923        my ($repo_id, $remotes) = @_;
 924        if (ref $repo_id) {
 925                my $gs = $repo_id;
 926                $repo_id = undef;
 927                $repo_id = $gs->{repo_id};
 928        }
 929        $remotes ||= read_all_remotes();
 930        my $remote = $remotes->{$repo_id} or
 931                     die "[svn-remote \"$repo_id\"] unknown\n";
 932        my $fetch = $remote->{fetch};
 933        my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
 934        my (@gs, @globs);
 935        my $ra = Git::SVN::Ra->new($url);
 936        my $uuid = $ra->get_uuid;
 937        my $head = $ra->get_latest_revnum;
 938        my $base = defined $fetch ? $head : 0;
 939
 940        # read the max revs for wildcard expansion (branches/*, tags/*)
 941        foreach my $t (qw/branches tags/) {
 942                defined $remote->{$t} or next;
 943                push @globs, $remote->{$t};
 944                my $max_rev = eval { tmp_config(qw/--int --get/,
 945                                         "svn-remote.$repo_id.${t}-maxRev") };
 946                if (defined $max_rev && ($max_rev < $base)) {
 947                        $base = $max_rev;
 948                } elsif (!defined $max_rev) {
 949                        $base = 0;
 950                }
 951        }
 952
 953        if ($fetch) {
 954                foreach my $p (sort keys %$fetch) {
 955                        my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
 956                        my $lr = $gs->rev_db_max;
 957                        if (defined $lr) {
 958                                $base = $lr if ($lr < $base);
 959                        }
 960                        push @gs, $gs;
 961                }
 962        }
 963
 964        ($base, $head) = parse_revision_argument($base, $head);
 965        $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
 966}
 967
 968sub read_all_remotes {
 969        my $r = {};
 970        foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
 971                if (m!^(.+)\.fetch=\s*(.*)\s*:\s*refs/remotes/(.+)\s*$!) {
 972                        $r->{$1}->{fetch}->{$2} = $3;
 973                } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
 974                        $r->{$1}->{url} = $2;
 975                } elsif (m!^(.+)\.(branches|tags)=
 976                           (.*):refs/remotes/(.+)\s*$/!x) {
 977                        my ($p, $g) = ($3, $4);
 978                        my $rs = $r->{$1}->{$2} = {
 979                                          t => $2,
 980                                          remote => $1,
 981                                          path => Git::SVN::GlobSpec->new($p),
 982                                          ref => Git::SVN::GlobSpec->new($g) };
 983                        if (length($rs->{ref}->{right}) != 0) {
 984                                die "The '*' glob character must be the last ",
 985                                    "character of '$g'\n";
 986                        }
 987                }
 988        }
 989        $r;
 990}
 991
 992sub init_vars {
 993        if (defined $_repack) {
 994                $_repack = 1000 if ($_repack <= 0);
 995                $_repack_nr = $_repack;
 996                $_repack_flags ||= '-d';
 997        }
 998}
 999
1000sub verify_remotes_sanity {
1001        return unless -d $ENV{GIT_DIR};
1002        my %seen;
1003        foreach (command(qw/config -l/)) {
1004                if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
1005                        if ($seen{$1}) {
1006                                die "Remote ref refs/remote/$1 is tracked by",
1007                                    "\n  \"$_\"\nand\n  \"$seen{$1}\"\n",
1008                                    "Please resolve this ambiguity in ",
1009                                    "your git configuration file before ",
1010                                    "continuing\n";
1011                        }
1012                        $seen{$1} = $_;
1013                }
1014        }
1015}
1016
1017# we allow more chars than remotes2config.sh...
1018sub sanitize_remote_name {
1019        my ($name) = @_;
1020        $name =~ tr{A-Za-z0-9:,/+-}{.}c;
1021        $name;
1022}
1023
1024sub find_existing_remote {
1025        my ($url, $remotes) = @_;
1026        return undef if $no_reuse_existing;
1027        my $existing;
1028        foreach my $repo_id (keys %$remotes) {
1029                my $u = $remotes->{$repo_id}->{url} or next;
1030                next if $u ne $url;
1031                $existing = $repo_id;
1032                last;
1033        }
1034        $existing;
1035}
1036
1037sub init_remote_config {
1038        my ($self, $url, $no_write) = @_;
1039        $url =~ s!/+$!!; # strip trailing slash
1040        my $r = read_all_remotes();
1041        my $existing = find_existing_remote($url, $r);
1042        if ($existing) {
1043                unless ($no_write) {
1044                        print STDERR "Using existing ",
1045                                     "[svn-remote \"$existing\"]\n";
1046                }
1047                $self->{repo_id} = $existing;
1048        } elsif ($_minimize_url) {
1049                my $min_url = Git::SVN::Ra->new($url)->minimize_url;
1050                $existing = find_existing_remote($min_url, $r);
1051                if ($existing) {
1052                        unless ($no_write) {
1053                                print STDERR "Using existing ",
1054                                             "[svn-remote \"$existing\"]\n";
1055                        }
1056                        $self->{repo_id} = $existing;
1057                }
1058                if ($min_url ne $url) {
1059                        unless ($no_write) {
1060                                print STDERR "Using higher level of URL: ",
1061                                             "$url => $min_url\n";
1062                        }
1063                        my $old_path = $self->{path};
1064                        $self->{path} = $url;
1065                        $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
1066                        if (length $old_path) {
1067                                $self->{path} .= "/$old_path";
1068                        }
1069                        $url = $min_url;
1070                }
1071        }
1072        my $orig_url;
1073        if (!$existing) {
1074                # verify that we aren't overwriting anything:
1075                $orig_url = eval {
1076                        command_oneline('config', '--get',
1077                                        "svn-remote.$self->{repo_id}.url")
1078                };
1079                if ($orig_url && ($orig_url ne $url)) {
1080                        die "svn-remote.$self->{repo_id}.url already set: ",
1081                            "$orig_url\nwanted to set to: $url\n";
1082                }
1083        }
1084        my ($xrepo_id, $xpath) = find_ref($self->refname);
1085        if (defined $xpath) {
1086                die "svn-remote.$xrepo_id.fetch already set to track ",
1087                    "$xpath:refs/remotes/", $self->refname, "\n";
1088        }
1089        unless ($no_write) {
1090                command_noisy('config',
1091                              "svn-remote.$self->{repo_id}.url", $url);
1092                command_noisy('config', '--add',
1093                              "svn-remote.$self->{repo_id}.fetch",
1094                              "$self->{path}:".$self->refname);
1095        }
1096        $self->{url} = $url;
1097}
1098
1099sub find_by_url { # repos_root and, path are optional
1100        my ($class, $full_url, $repos_root, $path) = @_;
1101
1102        return undef unless defined $full_url;
1103        remove_username($full_url);
1104        remove_username($repos_root) if defined $repos_root;
1105        my $remotes = read_all_remotes();
1106        if (defined $full_url && defined $repos_root && !defined $path) {
1107                $path = $full_url;
1108                $path =~ s#^\Q$repos_root\E(?:/|$)##;
1109        }
1110        foreach my $repo_id (keys %$remotes) {
1111                my $u = $remotes->{$repo_id}->{url} or next;
1112                remove_username($u);
1113                next if defined $repos_root && $repos_root ne $u;
1114
1115                my $fetch = $remotes->{$repo_id}->{fetch} || {};
1116                foreach (qw/branches tags/) {
1117                        resolve_local_globs($u, $fetch,
1118                                            $remotes->{$repo_id}->{$_});
1119                }
1120                my $p = $path;
1121                unless (defined $p) {
1122                        $p = $full_url;
1123                        $p =~ s#^\Q$u\E(?:/|$)## or next;
1124                }
1125                foreach my $f (keys %$fetch) {
1126                        next if $f ne $p;
1127                        return Git::SVN->new($fetch->{$f}, $repo_id, $f);
1128                }
1129        }
1130        undef;
1131}
1132
1133sub init {
1134        my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
1135        my $self = _new($class, $repo_id, $ref_id, $path);
1136        if (defined $url) {
1137                $self->init_remote_config($url, $no_write);
1138        }
1139        $self;
1140}
1141
1142sub find_ref {
1143        my ($ref_id) = @_;
1144        foreach (command(qw/config -l/)) {
1145                next unless m!^svn-remote\.(.+)\.fetch=
1146                              \s*(.*)\s*:\s*refs/remotes/(.+)\s*$!x;
1147                my ($repo_id, $path, $ref) = ($1, $2, $3);
1148                if ($ref eq $ref_id) {
1149                        $path = '' if ($path =~ m#^\./?#);
1150                        return ($repo_id, $path);
1151                }
1152        }
1153        (undef, undef, undef);
1154}
1155
1156sub new {
1157        my ($class, $ref_id, $repo_id, $path) = @_;
1158        if (defined $ref_id && !defined $repo_id && !defined $path) {
1159                ($repo_id, $path) = find_ref($ref_id);
1160                if (!defined $repo_id) {
1161                        die "Could not find a \"svn-remote.*.fetch\" key ",
1162                            "in the repository configuration matching: ",
1163                            "refs/remotes/$ref_id\n";
1164                }
1165        }
1166        my $self = _new($class, $repo_id, $ref_id, $path);
1167        if (!defined $self->{path} || !length $self->{path}) {
1168                my $fetch = command_oneline('config', '--get',
1169                                            "svn-remote.$repo_id.fetch",
1170                                            ":refs/remotes/$ref_id\$") or
1171                     die "Failed to read \"svn-remote.$repo_id.fetch\" ",
1172                         "\":refs/remotes/$ref_id\$\" in config\n";
1173                ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
1174        }
1175        $self->{url} = command_oneline('config', '--get',
1176                                       "svn-remote.$repo_id.url") or
1177                  die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
1178        $self->rebuild;
1179        $self;
1180}
1181
1182sub refname { "refs/remotes/$_[0]->{ref_id}" }
1183
1184sub svm_uuid {
1185        my ($self) = @_;
1186        return $self->{svm}->{uuid} if $self->svm;
1187        $self->ra;
1188        unless ($self->{svm}) {
1189                die "SVM UUID not cached, and reading remotely failed\n";
1190        }
1191        $self->{svm}->{uuid};
1192}
1193
1194sub svm {
1195        my ($self) = @_;
1196        return $self->{svm} if $self->{svm};
1197        my $svm;
1198        # see if we have it in our config, first:
1199        eval {
1200                my $section = "svn-remote.$self->{repo_id}";
1201                $svm = {
1202                  source => tmp_config('--get', "$section.svm-source"),
1203                  uuid => tmp_config('--get', "$section.svm-uuid"),
1204                  replace => tmp_config('--get', "$section.svm-replace"),
1205                }
1206        };
1207        if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
1208                $self->{svm} = $svm;
1209        }
1210        $self->{svm};
1211}
1212
1213sub _set_svm_vars {
1214        my ($self, $ra) = @_;
1215        return $ra if $self->svm;
1216
1217        my @err = ( "useSvmProps set, but failed to read SVM properties\n",
1218                    "(svm:source, svm:uuid) ",
1219                    "from the following URLs:\n" );
1220        sub read_svm_props {
1221                my ($self, $ra, $path, $r) = @_;
1222                my $props = ($ra->get_dir($path, $r))[2];
1223                my $src = $props->{'svm:source'};
1224                my $uuid = $props->{'svm:uuid'};
1225                return undef if (!$src || !$uuid);
1226
1227                chomp($src, $uuid);
1228
1229                $uuid =~ m{^[0-9a-f\-]{30,}$}
1230                    or die "doesn't look right - svm:uuid is '$uuid'\n";
1231
1232                # the '!' is used to mark the repos_root!/relative/path
1233                $src =~ s{/?!/?}{/};
1234                $src =~ s{/+$}{}; # no trailing slashes please
1235                # username is of no interest
1236                $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
1237
1238                my $replace = $ra->{url};
1239                $replace .= "/$path" if length $path;
1240
1241                my $section = "svn-remote.$self->{repo_id}";
1242                tmp_config("$section.svm-source", $src);
1243                tmp_config("$section.svm-replace", $replace);
1244                tmp_config("$section.svm-uuid", $uuid);
1245                $self->{svm} = {
1246                        source => $src,
1247                        uuid => $uuid,
1248                        replace => $replace
1249                };
1250        }
1251
1252        my $r = $ra->get_latest_revnum;
1253        my $path = $self->{path};
1254        my %tried;
1255        while (length $path) {
1256                unless ($tried{"$self->{url}/$path"}) {
1257                        return $ra if $self->read_svm_props($ra, $path, $r);
1258                        $tried{"$self->{url}/$path"} = 1;
1259                }
1260                $path =~ s#/?[^/]+$##;
1261        }
1262        die "Path: '$path' should be ''\n" if $path ne '';
1263        return $ra if $self->read_svm_props($ra, $path, $r);
1264        $tried{"$self->{url}/$path"} = 1;
1265
1266        if ($ra->{repos_root} eq $self->{url}) {
1267                die @err, (map { "  $_\n" } keys %tried), "\n";
1268        }
1269
1270        # nope, make sure we're connected to the repository root:
1271        my $ok;
1272        my @tried_b;
1273        $path = $ra->{svn_path};
1274        $ra = Git::SVN::Ra->new($ra->{repos_root});
1275        while (length $path) {
1276                unless ($tried{"$ra->{url}/$path"}) {
1277                        $ok = $self->read_svm_props($ra, $path, $r);
1278                        last if $ok;
1279                        $tried{"$ra->{url}/$path"} = 1;
1280                }
1281                $path =~ s#/?[^/]+$##;
1282        }
1283        die "Path: '$path' should be ''\n" if $path ne '';
1284        $ok ||= $self->read_svm_props($ra, $path, $r);
1285        $tried{"$ra->{url}/$path"} = 1;
1286        if (!$ok) {
1287                die @err, (map { "  $_\n" } keys %tried), "\n";
1288        }
1289        Git::SVN::Ra->new($self->{url});
1290}
1291
1292sub svnsync {
1293        my ($self) = @_;
1294        return $self->{svnsync} if $self->{svnsync};
1295
1296        if ($self->no_metadata) {
1297                die "Can't have both 'noMetadata' and ",
1298                    "'useSvnsyncProps' options set!\n";
1299        }
1300        if ($self->rewrite_root) {
1301                die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
1302                    "options set!\n";
1303        }
1304
1305        my $svnsync;
1306        # see if we have it in our config, first:
1307        eval {
1308                my $section = "svn-remote.$self->{repo_id}";
1309                $svnsync = {
1310                  url => tmp_config('--get', "$section.svnsync-url"),
1311                  uuid => tmp_config('--get', "$section.svnsync-uuid"),
1312                }
1313        };
1314        if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
1315                return $self->{svnsync} = $svnsync;
1316        }
1317
1318        my $err = "useSvnsyncProps set, but failed to read " .
1319                  "svnsync property: svn:sync-from-";
1320        my $rp = $self->ra->rev_proplist(0);
1321
1322        my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
1323        $url =~ m{^[a-z\+]+://} or
1324                   die "doesn't look right - svn:sync-from-url is '$url'\n";
1325
1326        my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
1327        $uuid =~ m{^[0-9a-f\-]{30,}$} or
1328                   die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
1329
1330        my $section = "svn-remote.$self->{repo_id}";
1331        tmp_config('--add', "$section.svnsync-uuid", $uuid);
1332        tmp_config('--add', "$section.svnsync-url", $url);
1333        return $self->{svnsync} = { url => $url, uuid => $uuid };
1334}
1335
1336# this allows us to memoize our SVN::Ra UUID locally and avoid a
1337# remote lookup (useful for 'git svn log').
1338sub ra_uuid {
1339        my ($self) = @_;
1340        unless ($self->{ra_uuid}) {
1341                my $key = "svn-remote.$self->{repo_id}.uuid";
1342                my $uuid = eval { tmp_config('--get', $key) };
1343                if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/) {
1344                        $self->{ra_uuid} = $uuid;
1345                } else {
1346                        die "ra_uuid called without URL\n" unless $self->{url};
1347                        $self->{ra_uuid} = $self->ra->get_uuid;
1348                        tmp_config('--add', $key, $self->{ra_uuid});
1349                }
1350        }
1351        $self->{ra_uuid};
1352}
1353
1354sub ra {
1355        my ($self) = shift;
1356        my $ra = Git::SVN::Ra->new($self->{url});
1357        if ($self->use_svm_props && !$self->{svm}) {
1358                if ($self->no_metadata) {
1359                        die "Can't have both 'noMetadata' and ",
1360                            "'useSvmProps' options set!\n";
1361                } elsif ($self->use_svnsync_props) {
1362                        die "Can't have both 'useSvnsyncProps' and ",
1363                            "'useSvmProps' options set!\n";
1364                }
1365                $ra = $self->_set_svm_vars($ra);
1366                $self->{-want_revprops} = 1;
1367        }
1368        $ra;
1369}
1370
1371sub rel_path {
1372        my ($self) = @_;
1373        my $repos_root = $self->ra->{repos_root};
1374        return $self->{path} if ($self->{url} eq $repos_root);
1375        my $url = $self->{url} .
1376                  (length $self->{path} ? "/$self->{path}" : $self->{path});
1377        $url =~ s!^\Q$repos_root\E(?:/+|$)!!g;
1378        $url;
1379}
1380
1381sub traverse_ignore {
1382        my ($self, $fh, $path, $r) = @_;
1383        $path =~ s#^/+##g;
1384        my $ra = $self->ra;
1385        my ($dirent, undef, $props) = $ra->get_dir($path, $r);
1386        my $p = $path;
1387        $p =~ s#^\Q$self->{path}\E(/|$)##;
1388        print $fh length $p ? "\n# $p\n" : "\n# /\n";
1389        if (my $s = $props->{'svn:ignore'}) {
1390                $s =~ s/[\r\n]+/\n/g;
1391                chomp $s;
1392                if (length $p == 0) {
1393                        $s =~ s#\n#\n/$p#g;
1394                        print $fh "/$s\n";
1395                } else {
1396                        $s =~ s#\n#\n/$p/#g;
1397                        print $fh "/$p/$s\n";
1398                }
1399        }
1400        foreach (sort keys %$dirent) {
1401                next if $dirent->{$_}->{kind} != $SVN::Node::dir;
1402                $self->traverse_ignore($fh, "$path/$_", $r);
1403        }
1404}
1405
1406sub last_rev { ($_[0]->last_rev_commit)[0] }
1407sub last_commit { ($_[0]->last_rev_commit)[1] }
1408
1409# returns the newest SVN revision number and newest commit SHA1
1410sub last_rev_commit {
1411        my ($self) = @_;
1412        if (defined $self->{last_rev} && defined $self->{last_commit}) {
1413                return ($self->{last_rev}, $self->{last_commit});
1414        }
1415        my $c = ::verify_ref($self->refname.'^0');
1416        if ($c && !$self->use_svm_props && !$self->no_metadata) {
1417                my $rev = (::cmt_metadata($c))[1];
1418                if (defined $rev) {
1419                        ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
1420                        return ($rev, $c);
1421                }
1422        }
1423        my $db_path = $self->db_path;
1424        unless (-e $db_path) {
1425                ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
1426                return (undef, undef);
1427        }
1428        my $offset = -41; # from tail
1429        my $rl;
1430        open my $fh, '<', $db_path or croak "$db_path not readable: $!\n";
1431        sysseek($fh, $offset, 2); # don't care for errors
1432        sysread($fh, $rl, 41) == 41 or return (undef, undef);
1433        chomp $rl;
1434        while (('0' x40) eq $rl && sysseek($fh, 0, 1) != 0) {
1435                $offset -= 41;
1436                sysseek($fh, $offset, 2); # don't care for errors
1437                sysread($fh, $rl, 41) == 41 or return (undef, undef);
1438                chomp $rl;
1439        }
1440        if ($c && $c ne $rl) {
1441                die "$db_path and ", $self->refname,
1442                    " inconsistent!:\n$c != $rl\n";
1443        }
1444        my $rev = sysseek($fh, 0, 1) or croak $!;
1445        $rev =  ($rev - 41) / 41;
1446        close $fh or croak $!;
1447        ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
1448        return ($rev, $c);
1449}
1450
1451sub get_fetch_range {
1452        my ($self, $min, $max) = @_;
1453        $max ||= $self->ra->get_latest_revnum;
1454        $min ||= $self->rev_db_max;
1455        (++$min, $max);
1456}
1457
1458sub tmp_config {
1459        my (@args) = @_;
1460        my $old_def_config = "$ENV{GIT_DIR}/svn/config";
1461        my $config = "$ENV{GIT_DIR}/svn/.metadata";
1462        if (! -f $config && -f $old_def_config) {
1463                rename $old_def_config, $config or
1464                       die "Failed rename $old_def_config => $config: $!\n";
1465        }
1466        my $old_config = $ENV{GIT_CONFIG};
1467        $ENV{GIT_CONFIG} = $config;
1468        $@ = undef;
1469        my @ret = eval {
1470                unless (-f $config) {
1471                        mkfile($config);
1472                        open my $fh, '>', $config or
1473                            die "Can't open $config: $!\n";
1474                        print $fh "; This file is used internally by ",
1475                                  "git-svn\n" or die
1476                                  "Couldn't write to $config: $!\n";
1477                        print $fh "; You should not have to edit it\n" or
1478                              die "Couldn't write to $config: $!\n";
1479                        close $fh or die "Couldn't close $config: $!\n";
1480                }
1481                command('config', @args);
1482        };
1483        my $err = $@;
1484        if (defined $old_config) {
1485                $ENV{GIT_CONFIG} = $old_config;
1486        } else {
1487                delete $ENV{GIT_CONFIG};
1488        }
1489        die $err if $err;
1490        wantarray ? @ret : $ret[0];
1491}
1492
1493sub tmp_index_do {
1494        my ($self, $sub) = @_;
1495        my $old_index = $ENV{GIT_INDEX_FILE};
1496        $ENV{GIT_INDEX_FILE} = $self->{index};
1497        $@ = undef;
1498        my @ret = eval {
1499                my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
1500                mkpath([$dir]) unless -d $dir;
1501                &$sub;
1502        };
1503        my $err = $@;
1504        if (defined $old_index) {
1505                $ENV{GIT_INDEX_FILE} = $old_index;
1506        } else {
1507                delete $ENV{GIT_INDEX_FILE};
1508        }
1509        die $err if $err;
1510        wantarray ? @ret : $ret[0];
1511}
1512
1513sub assert_index_clean {
1514        my ($self, $treeish) = @_;
1515
1516        $self->tmp_index_do(sub {
1517                command_noisy('read-tree', $treeish) unless -e $self->{index};
1518                my $x = command_oneline('write-tree');
1519                my ($y) = (command(qw/cat-file commit/, $treeish) =~
1520                           /^tree ($::sha1)/mo);
1521                return if $y eq $x;
1522
1523                warn "Index mismatch: $y != $x\nrereading $treeish\n";
1524                unlink $self->{index} or die "unlink $self->{index}: $!\n";
1525                command_noisy('read-tree', $treeish);
1526                $x = command_oneline('write-tree');
1527                if ($y ne $x) {
1528                        ::fatal "trees ($treeish) $y != $x\n",
1529                                "Something is seriously wrong...\n";
1530                }
1531        });
1532}
1533
1534sub get_commit_parents {
1535        my ($self, $log_entry) = @_;
1536        my (%seen, @ret, @tmp);
1537        # legacy support for 'set-tree'; this is only used by set_tree_cb:
1538        if (my $ip = $self->{inject_parents}) {
1539                if (my $commit = delete $ip->{$log_entry->{revision}}) {
1540                        push @tmp, $commit;
1541                }
1542        }
1543        if (my $cur = ::verify_ref($self->refname.'^0')) {
1544                push @tmp, $cur;
1545        }
1546        push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
1547        while (my $p = shift @tmp) {
1548                next if $seen{$p};
1549                $seen{$p} = 1;
1550                push @ret, $p;
1551                # MAXPARENT is defined to 16 in commit-tree.c:
1552                last if @ret >= 16;
1553        }
1554        if (@tmp) {
1555                die "r$log_entry->{revision}: No room for parents:\n\t",
1556                    join("\n\t", @tmp), "\n";
1557        }
1558        @ret;
1559}
1560
1561sub rewrite_root {
1562        my ($self) = @_;
1563        return $self->{-rewrite_root} if exists $self->{-rewrite_root};
1564        my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
1565        my $rwr = eval { command_oneline(qw/config --get/, $k) };
1566        if ($rwr) {
1567                $rwr =~ s#/+$##;
1568                if ($rwr !~ m#^[a-z\+]+://#) {
1569                        die "$rwr is not a valid URL (key: $k)\n";
1570                }
1571        }
1572        $self->{-rewrite_root} = $rwr;
1573}
1574
1575sub metadata_url {
1576        my ($self) = @_;
1577        ($self->rewrite_root || $self->{url}) .
1578           (length $self->{path} ? '/' . $self->{path} : '');
1579}
1580
1581sub full_url {
1582        my ($self) = @_;
1583        $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
1584}
1585
1586sub do_git_commit {
1587        my ($self, $log_entry) = @_;
1588        my $lr = $self->last_rev;
1589        if (defined $lr && $lr >= $log_entry->{revision}) {
1590                die "Last fetched revision of ", $self->refname,
1591                    " was r$lr, but we are about to fetch: ",
1592                    "r$log_entry->{revision}!\n";
1593        }
1594        if (my $c = $self->rev_db_get($log_entry->{revision})) {
1595                croak "$log_entry->{revision} = $c already exists! ",
1596                      "Why are we refetching it?\n";
1597        }
1598        $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} = $log_entry->{name};
1599        $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} =
1600                                                          $log_entry->{email};
1601        $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
1602
1603        my $tree = $log_entry->{tree};
1604        if (!defined $tree) {
1605                $tree = $self->tmp_index_do(sub {
1606                                            command_oneline('write-tree') });
1607        }
1608        die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
1609
1610        my @exec = ('git-commit-tree', $tree);
1611        foreach ($self->get_commit_parents($log_entry)) {
1612                push @exec, '-p', $_;
1613        }
1614        defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
1615                                                                   or croak $!;
1616        print $msg_fh $log_entry->{log} or croak $!;
1617        unless ($self->no_metadata) {
1618                print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
1619                              or croak $!;
1620        }
1621        $msg_fh->flush == 0 or croak $!;
1622        close $msg_fh or croak $!;
1623        chomp(my $commit = do { local $/; <$out_fh> });
1624        close $out_fh or croak $!;
1625        waitpid $pid, 0;
1626        croak $? if $?;
1627        if ($commit !~ /^$::sha1$/o) {
1628                die "Failed to commit, invalid sha1: $commit\n";
1629        }
1630
1631        $self->rev_db_set($log_entry->{revision}, $commit, 1);
1632
1633        $self->{last_rev} = $log_entry->{revision};
1634        $self->{last_commit} = $commit;
1635        print "r$log_entry->{revision}";
1636        if (defined $log_entry->{svm_revision}) {
1637                 print " (\@$log_entry->{svm_revision})";
1638                 $self->rev_db_set($log_entry->{svm_revision}, $commit,
1639                                   0, $self->svm_uuid);
1640        }
1641        print " = $commit ($self->{ref_id})\n";
1642        if (defined $_repack && (--$_repack_nr == 0)) {
1643                $_repack_nr = $_repack;
1644                # repack doesn't use any arguments with spaces in them, does it?
1645                print "Running git repack $_repack_flags ...\n";
1646                command_noisy('repack', split(/\s+/, $_repack_flags));
1647                print "Done repacking\n";
1648        }
1649        return $commit;
1650}
1651
1652sub match_paths {
1653        my ($self, $paths, $r) = @_;
1654        return 1 if $self->{path} eq '';
1655        if (my $path = $paths->{"/$self->{path}"}) {
1656                return ($path->{action} eq 'D') ? 0 : 1;
1657        }
1658        $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
1659        if (grep /$self->{path_regex}/, keys %$paths) {
1660                return 1;
1661        }
1662        my $c = '';
1663        foreach (split m#/#, $self->{path}) {
1664                $c .= "/$_";
1665                next unless ($paths->{$c} &&
1666                             ($paths->{$c}->{action} =~ /^[AR]$/));
1667                if ($self->ra->check_path($self->{path}, $r) ==
1668                    $SVN::Node::dir) {
1669                        return 1;
1670                }
1671        }
1672        return 0;
1673}
1674
1675sub find_parent_branch {
1676        my ($self, $paths, $rev) = @_;
1677        return undef unless $self->follow_parent;
1678        unless (defined $paths) {
1679                my $err_handler = $SVN::Error::handler;
1680                $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
1681                $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1, sub {
1682                                   $paths =
1683                                      Git::SVN::Ra::dup_changed_paths($_[0]) });
1684                $SVN::Error::handler = $err_handler;
1685        }
1686        return undef unless defined $paths;
1687
1688        # look for a parent from another branch:
1689        my @b_path_components = split m#/#, $self->rel_path;
1690        my @a_path_components;
1691        my $i;
1692        while (@b_path_components) {
1693                $i = $paths->{'/'.join('/', @b_path_components)};
1694                last if $i && defined $i->{copyfrom_path};
1695                unshift(@a_path_components, pop(@b_path_components));
1696        }
1697        return undef unless defined $i && defined $i->{copyfrom_path};
1698        my $branch_from = $i->{copyfrom_path};
1699        if (@a_path_components) {
1700                print STDERR "branch_from: $branch_from => ";
1701                $branch_from .= '/'.join('/', @a_path_components);
1702                print STDERR $branch_from, "\n";
1703        }
1704        my $r = $i->{copyfrom_rev};
1705        my $repos_root = $self->ra->{repos_root};
1706        my $url = $self->ra->{url};
1707        my $new_url = $repos_root . $branch_from;
1708        print STDERR  "Found possible branch point: ",
1709                      "$new_url => ", $self->full_url, ", $r\n";
1710        $branch_from =~ s#^/##;
1711        my $gs = Git::SVN->find_by_url($new_url, $repos_root, $branch_from);
1712        unless ($gs) {
1713                my $ref_id = $self->{ref_id};
1714                $ref_id =~ s/\@\d+$//;
1715                $ref_id .= "\@$r";
1716                # just grow a tail if we're not unique enough :x
1717                $ref_id .= '-' while find_ref($ref_id);
1718                print STDERR "Initializing parent: $ref_id\n";
1719                $gs = Git::SVN->init($new_url, '', $ref_id, $ref_id, 1);
1720        }
1721        my ($r0, $parent) = $gs->find_rev_before($r, 1);
1722        if (!defined $r0 || !defined $parent) {
1723                my ($base, $head) = parse_revision_argument(0, $r);
1724                if ($base <= $r) {
1725                        $gs->fetch($base, $r);
1726                }
1727                ($r0, $parent) = $gs->last_rev_commit;
1728        }
1729        if (defined $r0 && defined $parent) {
1730                print STDERR "Found branch parent: ($self->{ref_id}) $parent\n";
1731                my $ed;
1732                if ($self->ra->can_do_switch) {
1733                        $self->assert_index_clean($parent);
1734                        print STDERR "Following parent with do_switch\n";
1735                        # do_switch works with svn/trunk >= r22312, but that
1736                        # is not included with SVN 1.4.3 (the latest version
1737                        # at the moment), so we can't rely on it
1738                        $self->{last_commit} = $parent;
1739                        $ed = SVN::Git::Fetcher->new($self);
1740                        $gs->ra->gs_do_switch($r0, $rev, $gs,
1741                                              $self->full_url, $ed)
1742                          or die "SVN connection failed somewhere...\n";
1743                } else {
1744                        print STDERR "Following parent with do_update\n";
1745                        $ed = SVN::Git::Fetcher->new($self);
1746                        $self->ra->gs_do_update($rev, $rev, $self, $ed)
1747                          or die "SVN connection failed somewhere...\n";
1748                }
1749                print STDERR "Successfully followed parent\n";
1750                return $self->make_log_entry($rev, [$parent], $ed);
1751        }
1752        return undef;
1753}
1754
1755sub do_fetch {
1756        my ($self, $paths, $rev) = @_;
1757        my $ed;
1758        my ($last_rev, @parents);
1759        if (my $lc = $self->last_commit) {
1760                # we can have a branch that was deleted, then re-added
1761                # under the same name but copied from another path, in
1762                # which case we'll have multiple parents (we don't
1763                # want to break the original ref, nor lose copypath info):
1764                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
1765                        push @{$log_entry->{parents}}, $lc;
1766                        return $log_entry;
1767                }
1768                $ed = SVN::Git::Fetcher->new($self);
1769                $last_rev = $self->{last_rev};
1770                $ed->{c} = $lc;
1771                @parents = ($lc);
1772        } else {
1773                $last_rev = $rev;
1774                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
1775                        return $log_entry;
1776                }
1777                $ed = SVN::Git::Fetcher->new($self);
1778        }
1779        unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
1780                die "SVN connection failed somewhere...\n";
1781        }
1782        $self->make_log_entry($rev, \@parents, $ed);
1783}
1784
1785sub get_untracked {
1786        my ($self, $ed) = @_;
1787        my @out;
1788        my $h = $ed->{empty};
1789        foreach (sort keys %$h) {
1790                my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
1791                push @out, "  $act: " . uri_encode($_);
1792                warn "W: $act: $_\n";
1793        }
1794        foreach my $t (qw/dir_prop file_prop/) {
1795                $h = $ed->{$t} or next;
1796                foreach my $path (sort keys %$h) {
1797                        my $ppath = $path eq '' ? '.' : $path;
1798                        foreach my $prop (sort keys %{$h->{$path}}) {
1799                                next if $SKIP_PROP{$prop};
1800                                my $v = $h->{$path}->{$prop};
1801                                my $t_ppath_prop = "$t: " .
1802                                                    uri_encode($ppath) . ' ' .
1803                                                    uri_encode($prop);
1804                                if (defined $v) {
1805                                        push @out, "  +$t_ppath_prop " .
1806                                                   uri_encode($v);
1807                                } else {
1808                                        push @out, "  -$t_ppath_prop";
1809                                }
1810                        }
1811                }
1812        }
1813        foreach my $t (qw/absent_file absent_directory/) {
1814                $h = $ed->{$t} or next;
1815                foreach my $parent (sort keys %$h) {
1816                        foreach my $path (sort @{$h->{$parent}}) {
1817                                push @out, "  $t: " .
1818                                           uri_encode("$parent/$path");
1819                                warn "W: $t: $parent/$path ",
1820                                     "Insufficient permissions?\n";
1821                        }
1822                }
1823        }
1824        \@out;
1825}
1826
1827sub parse_svn_date {
1828        my $date = shift || return '+0000 1970-01-01 00:00:00';
1829        my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
1830                                            (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x) or
1831                                         croak "Unable to parse date: $date\n";
1832        "+0000 $Y-$m-$d $H:$M:$S";
1833}
1834
1835sub check_author {
1836        my ($author) = @_;
1837        if (!defined $author || length $author == 0) {
1838                $author = '(no author)';
1839        }
1840        if (defined $::_authors && ! defined $::users{$author}) {
1841                die "Author: $author not defined in $::_authors file\n";
1842        }
1843        $author;
1844}
1845
1846sub make_log_entry {
1847        my ($self, $rev, $parents, $ed) = @_;
1848        my $untracked = $self->get_untracked($ed);
1849
1850        open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
1851        print $un "r$rev\n" or croak $!;
1852        print $un $_, "\n" foreach @$untracked;
1853        my %log_entry = ( parents => $parents || [], revision => $rev,
1854                          log => '');
1855
1856        my $headrev;
1857        my $logged = delete $self->{logged_rev_props};
1858        if (!$logged || $self->{-want_revprops}) {
1859                my $rp = $self->ra->rev_proplist($rev);
1860                foreach (sort keys %$rp) {
1861                        my $v = $rp->{$_};
1862                        if (/^svn:(author|date|log)$/) {
1863                                $log_entry{$1} = $v;
1864                        } elsif ($_ eq 'svm:headrev') {
1865                                $headrev = $v;
1866                        } else {
1867                                print $un "  rev_prop: ", uri_encode($_), ' ',
1868                                          uri_encode($v), "\n";
1869                        }
1870                }
1871        } else {
1872                map { $log_entry{$_} = $logged->{$_} } keys %$logged;
1873        }
1874        close $un or croak $!;
1875
1876        $log_entry{date} = parse_svn_date($log_entry{date});
1877        $log_entry{log} .= "\n";
1878        my $author = $log_entry{author} = check_author($log_entry{author});
1879        my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
1880                                                       : ($author, undef);
1881        if (defined $headrev && $self->use_svm_props) {
1882                if ($self->rewrite_root) {
1883                        die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
1884                            "options set!\n";
1885                }
1886                my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$};
1887                # we don't want "SVM: initializing mirror for junk" ...
1888                return undef if $r == 0;
1889                my $svm = $self->svm;
1890                if ($uuid ne $svm->{uuid}) {
1891                        die "UUID mismatch on SVM path:\n",
1892                            "expected: $svm->{uuid}\n",
1893                            "     got: $uuid\n";
1894                }
1895                my $full_url = $self->full_url;
1896                $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
1897                             die "Failed to replace '$svm->{replace}' with ",
1898                                 "'$svm->{source}' in $full_url\n";
1899                # throw away username for storing in records
1900                remove_username($full_url);
1901                $log_entry{metadata} = "$full_url\@$r $uuid";
1902                $log_entry{svm_revision} = $r;
1903                $email ||= "$author\@$uuid"
1904        } elsif ($self->use_svnsync_props) {
1905                my $full_url = $self->svnsync->{url};
1906                $full_url .= "/$self->{path}" if length $self->{path};
1907                remove_username($full_url);
1908                my $uuid = $self->svnsync->{uuid};
1909                $log_entry{metadata} = "$full_url\@$rev $uuid";
1910                $email ||= "$author\@$uuid"
1911        } else {
1912                my $url = $self->metadata_url;
1913                remove_username($url);
1914                $log_entry{metadata} = "$url\@$rev " .
1915                                       $self->ra->get_uuid;
1916                $email ||= "$author\@" . $self->ra->get_uuid;
1917        }
1918        $log_entry{name} = $name;
1919        $log_entry{email} = $email;
1920        \%log_entry;
1921}
1922
1923sub fetch {
1924        my ($self, $min_rev, $max_rev, @parents) = @_;
1925        my ($last_rev, $last_commit) = $self->last_rev_commit;
1926        my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
1927        $self->ra->gs_fetch_loop_common($base, $head, [$self]);
1928}
1929
1930sub set_tree_cb {
1931        my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
1932        $self->{inject_parents} = { $rev => $tree };
1933        $self->fetch(undef, undef);
1934}
1935
1936sub set_tree {
1937        my ($self, $tree) = (shift, shift);
1938        my $log_entry = ::get_commit_entry($tree);
1939        unless ($self->{last_rev}) {
1940                fatal("Must have an existing revision to commit\n");
1941        }
1942        my %ed_opts = ( r => $self->{last_rev},
1943                        log => $log_entry->{log},
1944                        ra => $self->ra,
1945                        tree_a => $self->{last_commit},
1946                        tree_b => $tree,
1947                        editor_cb => sub {
1948                               $self->set_tree_cb($log_entry, $tree, @_) },
1949                        svn_path => $self->{path} );
1950        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
1951                print "No changes\nr$self->{last_rev} = $tree\n";
1952        }
1953}
1954
1955sub rebuild {
1956        my ($self) = @_;
1957        my $db_path = $self->db_path;
1958        return if (-e $db_path && ! -z $db_path);
1959        return unless ::verify_ref($self->refname.'^0');
1960        if (-f $self->{db_root}) {
1961                rename $self->{db_root}, $db_path or die
1962                     "rename $self->{db_root} => $db_path failed: $!\n";
1963                my ($dir, $base) = ($db_path =~ m#^(.*?)/?([^/]+)$#);
1964                symlink $base, $self->{db_root} or die
1965                     "symlink $base => $self->{db_root} failed: $!\n";
1966                return;
1967        }
1968        print "Rebuilding $db_path ...\n";
1969        my ($rev_list, $ctx) = command_output_pipe("rev-list", $self->refname);
1970        my $latest;
1971        my $full_url = $self->full_url;
1972        remove_username($full_url);
1973        my $svn_uuid;
1974        while (<$rev_list>) {
1975                chomp;
1976                my $c = $_;
1977                die "Non-SHA1: $c\n" unless $c =~ /^$::sha1$/o;
1978                my ($url, $rev, $uuid) = ::cmt_metadata($c);
1979                remove_username($url);
1980
1981                # ignore merges (from set-tree)
1982                next if (!defined $rev || !$uuid);
1983
1984                # if we merged or otherwise started elsewhere, this is
1985                # how we break out of it
1986                if ((defined $svn_uuid && ($uuid ne $svn_uuid)) ||
1987                    ($full_url && $url && ($url ne $full_url))) {
1988                        next;
1989                }
1990                $latest ||= $rev;
1991                $svn_uuid ||= $uuid;
1992
1993                $self->rev_db_set($rev, $c);
1994                print "r$rev = $c\n";
1995        }
1996        command_close_pipe($rev_list, $ctx);
1997        print "Done rebuilding $db_path\n";
1998}
1999
2000# rev_db:
2001# Tie::File seems to be prone to offset errors if revisions get sparse,
2002# it's not that fast, either.  Tie::File is also not in Perl 5.6.  So
2003# one of my favorite modules is out :<  Next up would be one of the DBM
2004# modules, but I'm not sure which is most portable...  So I'll just
2005# go with something that's plain-text, but still capable of
2006# being randomly accessed.  So here's my ultra-simple fixed-width
2007# database.  All records are 40 characters + "\n", so it's easy to seek
2008# to a revision: (41 * rev) is the byte offset.
2009# A record of 40 0s denotes an empty revision.
2010# And yes, it's still pretty fast (faster than Tie::File).
2011# These files are disposable unless noMetadata or useSvmProps is set
2012
2013sub _rev_db_set {
2014        my ($fh, $rev, $commit) = @_;
2015        my $offset = $rev * 41;
2016        # assume that append is the common case:
2017        seek $fh, 0, 2 or croak $!;
2018        my $pos = tell $fh;
2019        if ($pos < $offset) {
2020                for (1 .. (($offset - $pos) / 41)) {
2021                        print $fh (('0' x 40),"\n") or croak $!;
2022                }
2023        }
2024        seek $fh, $offset, 0 or croak $!;
2025        print $fh $commit,"\n" or croak $!;
2026}
2027
2028sub mkfile {
2029        my ($path) = @_;
2030        unless (-e $path) {
2031                my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
2032                mkpath([$dir]) unless -d $dir;
2033                open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
2034                close $fh or die "Couldn't close (create) $path: $!\n";
2035        }
2036}
2037
2038sub rev_db_set {
2039        my ($self, $rev, $commit, $update_ref, $uuid) = @_;
2040        length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
2041        my $db = $self->db_path($uuid);
2042        my $db_lock = "$db.lock";
2043        my $sig;
2044        if ($update_ref) {
2045                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
2046                            $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
2047        }
2048        mkfile($db);
2049
2050        $LOCKFILES{$db_lock} = 1;
2051        my $sync;
2052        # both of these options make our .rev_db file very, very important
2053        # and we can't afford to lose it because rebuild() won't work
2054        if ($self->use_svm_props || $self->no_metadata) {
2055                $sync = 1;
2056                copy($db, $db_lock) or die "rev_db_set(@_): ",
2057                                           "Failed to copy: ",
2058                                           "$db => $db_lock ($!)\n";
2059        } else {
2060                rename $db, $db_lock or die "rev_db_set(@_): ",
2061                                            "Failed to rename: ",
2062                                            "$db => $db_lock ($!)\n";
2063        }
2064        open my $fh, '+<', $db_lock or die "Couldn't open $db_lock: $!\n";
2065        _rev_db_set($fh, $rev, $commit);
2066        if ($sync) {
2067                $fh->flush or die "Couldn't flush $db_lock: $!\n";
2068                $fh->sync or die "Couldn't sync $db_lock: $!\n";
2069        }
2070        close $fh or croak $!;
2071        if ($update_ref) {
2072                $_head = $self;
2073                command_noisy('update-ref', '-m', "r$rev",
2074                              $self->refname, $commit);
2075        }
2076        rename $db_lock, $db or die "rev_db_set(@_): ", "Failed to rename: ",
2077                                    "$db_lock => $db ($!)\n";
2078        delete $LOCKFILES{$db_lock};
2079        if ($update_ref) {
2080                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
2081                            $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
2082                kill $sig, $$ if defined $sig;
2083        }
2084}
2085
2086sub rev_db_max {
2087        my ($self) = @_;
2088        $self->rebuild;
2089        my $db_path = $self->db_path;
2090        my @stat = stat $db_path or return 0;
2091        ($stat[7] % 41) == 0 or die "$db_path inconsistent size: $stat[7]\n";
2092        my $max = $stat[7] / 41;
2093        (($max > 0) ? $max - 1 : 0);
2094}
2095
2096sub rev_db_get {
2097        my ($self, $rev, $uuid) = @_;
2098        my $ret;
2099        my $offset = $rev * 41;
2100        my $db_path = $self->db_path($uuid);
2101        return undef unless -e $db_path;
2102        open my $fh, '<', $db_path or croak $!;
2103        if (sysseek($fh, $offset, 0) == $offset) {
2104                my $read = sysread($fh, $ret, 40);
2105                $ret = undef if ($read != 40 || $ret eq ('0'x40));
2106        }
2107        close $fh or croak $!;
2108        $ret;
2109}
2110
2111sub find_rev_before {
2112        my ($self, $rev, $eq_ok) = @_;
2113        --$rev unless $eq_ok;
2114        while ($rev > 0) {
2115                if (my $c = $self->rev_db_get($rev)) {
2116                        return ($rev, $c);
2117                }
2118                --$rev;
2119        }
2120        return (undef, undef);
2121}
2122
2123sub _new {
2124        my ($class, $repo_id, $ref_id, $path) = @_;
2125        unless (defined $repo_id && length $repo_id) {
2126                $repo_id = $Git::SVN::default_repo_id;
2127        }
2128        unless (defined $ref_id && length $ref_id) {
2129                $_[2] = $ref_id = $Git::SVN::default_ref_id;
2130        }
2131        $_[1] = $repo_id = sanitize_remote_name($repo_id);
2132        my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
2133        $_[3] = $path = '' unless (defined $path);
2134        mkpath(["$ENV{GIT_DIR}/svn"]);
2135        bless {
2136                ref_id => $ref_id, dir => $dir, index => "$dir/index",
2137                path => $path, config => "$ENV{GIT_DIR}/svn/config",
2138                db_root => "$dir/.rev_db", repo_id => $repo_id }, $class;
2139}
2140
2141sub db_path {
2142        my ($self, $uuid) = @_;
2143        $uuid ||= $self->ra_uuid;
2144        "$self->{db_root}.$uuid";
2145}
2146
2147sub uri_encode {
2148        my ($f) = @_;
2149        $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
2150        $f
2151}
2152
2153sub remove_username {
2154        $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
2155}
2156
2157package Git::SVN::Prompt;
2158use strict;
2159use warnings;
2160require SVN::Core;
2161use vars qw/$_no_auth_cache $_username/;
2162
2163sub simple {
2164        my ($cred, $realm, $default_username, $may_save, $pool) = @_;
2165        $may_save = undef if $_no_auth_cache;
2166        $default_username = $_username if defined $_username;
2167        if (defined $default_username && length $default_username) {
2168                if (defined $realm && length $realm) {
2169                        print STDERR "Authentication realm: $realm\n";
2170                        STDERR->flush;
2171                }
2172                $cred->username($default_username);
2173        } else {
2174                username($cred, $realm, $may_save, $pool);
2175        }
2176        $cred->password(_read_password("Password for '" .
2177                                       $cred->username . "': ", $realm));
2178        $cred->may_save($may_save);
2179        $SVN::_Core::SVN_NO_ERROR;
2180}
2181
2182sub ssl_server_trust {
2183        my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
2184        $may_save = undef if $_no_auth_cache;
2185        print STDERR "Error validating server certificate for '$realm':\n";
2186        if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
2187                print STDERR " - The certificate is not issued by a trusted ",
2188                      "authority. Use the\n",
2189                      "   fingerprint to validate the certificate manually!\n";
2190        }
2191        if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
2192                print STDERR " - The certificate hostname does not match.\n";
2193        }
2194        if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
2195                print STDERR " - The certificate is not yet valid.\n";
2196        }
2197        if ($failures & $SVN::Auth::SSL::EXPIRED) {
2198                print STDERR " - The certificate has expired.\n";
2199        }
2200        if ($failures & $SVN::Auth::SSL::OTHER) {
2201                print STDERR " - The certificate has an unknown error.\n";
2202        }
2203        printf STDERR
2204                "Certificate information:\n".
2205                " - Hostname: %s\n".
2206                " - Valid: from %s until %s\n".
2207                " - Issuer: %s\n".
2208                " - Fingerprint: %s\n",
2209                map $cert_info->$_, qw(hostname valid_from valid_until
2210                                       issuer_dname fingerprint);
2211        my $choice;
2212prompt:
2213        print STDERR $may_save ?
2214              "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
2215              "(R)eject or accept (t)emporarily? ";
2216        STDERR->flush;
2217        $choice = lc(substr(<STDIN> || 'R', 0, 1));
2218        if ($choice =~ /^t$/i) {
2219                $cred->may_save(undef);
2220        } elsif ($choice =~ /^r$/i) {
2221                return -1;
2222        } elsif ($may_save && $choice =~ /^p$/i) {
2223                $cred->may_save($may_save);
2224        } else {
2225                goto prompt;
2226        }
2227        $cred->accepted_failures($failures);
2228        $SVN::_Core::SVN_NO_ERROR;
2229}
2230
2231sub ssl_client_cert {
2232        my ($cred, $realm, $may_save, $pool) = @_;
2233        $may_save = undef if $_no_auth_cache;
2234        print STDERR "Client certificate filename: ";
2235        STDERR->flush;
2236        chomp(my $filename = <STDIN>);
2237        $cred->cert_file($filename);
2238        $cred->may_save($may_save);
2239        $SVN::_Core::SVN_NO_ERROR;
2240}
2241
2242sub ssl_client_cert_pw {
2243        my ($cred, $realm, $may_save, $pool) = @_;
2244        $may_save = undef if $_no_auth_cache;
2245        $cred->password(_read_password("Password: ", $realm));
2246        $cred->may_save($may_save);
2247        $SVN::_Core::SVN_NO_ERROR;
2248}
2249
2250sub username {
2251        my ($cred, $realm, $may_save, $pool) = @_;
2252        $may_save = undef if $_no_auth_cache;
2253        if (defined $realm && length $realm) {
2254                print STDERR "Authentication realm: $realm\n";
2255        }
2256        my $username;
2257        if (defined $_username) {
2258                $username = $_username;
2259        } else {
2260                print STDERR "Username: ";
2261                STDERR->flush;
2262                chomp($username = <STDIN>);
2263        }
2264        $cred->username($username);
2265        $cred->may_save($may_save);
2266        $SVN::_Core::SVN_NO_ERROR;
2267}
2268
2269sub _read_password {
2270        my ($prompt, $realm) = @_;
2271        print STDERR $prompt;
2272        STDERR->flush;
2273        require Term::ReadKey;
2274        Term::ReadKey::ReadMode('noecho');
2275        my $password = '';
2276        while (defined(my $key = Term::ReadKey::ReadKey(0))) {
2277                last if $key =~ /[\012\015]/; # \n\r
2278                $password .= $key;
2279        }
2280        Term::ReadKey::ReadMode('restore');
2281        print STDERR "\n";
2282        STDERR->flush;
2283        $password;
2284}
2285
2286package main;
2287
2288{
2289        my $kill_stupid_warnings = $SVN::Node::none.$SVN::Node::file.
2290                                $SVN::Node::dir.$SVN::Node::unknown.
2291                                $SVN::Node::none.$SVN::Node::file.
2292                                $SVN::Node::dir.$SVN::Node::unknown.
2293                                $SVN::Auth::SSL::CNMISMATCH.
2294                                $SVN::Auth::SSL::NOTYETVALID.
2295                                $SVN::Auth::SSL::EXPIRED.
2296                                $SVN::Auth::SSL::UNKNOWNCA.
2297                                $SVN::Auth::SSL::OTHER;
2298}
2299
2300package SVN::Git::Fetcher;
2301use vars qw/@ISA/;
2302use strict;
2303use warnings;
2304use Carp qw/croak/;
2305use IO::File qw//;
2306use Digest::MD5;
2307
2308# file baton members: path, mode_a, mode_b, pool, fh, blob, base
2309sub new {
2310        my ($class, $git_svn) = @_;
2311        my $self = SVN::Delta::Editor->new;
2312        bless $self, $class;
2313        $self->{c} = $git_svn->{last_commit} if exists $git_svn->{last_commit};
2314        $self->{empty} = {};
2315        $self->{dir_prop} = {};
2316        $self->{file_prop} = {};
2317        $self->{absent_dir} = {};
2318        $self->{absent_file} = {};
2319        $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
2320        $self;
2321}
2322
2323sub set_path_strip {
2324        my ($self, $path) = @_;
2325        $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
2326}
2327
2328sub open_root {
2329        { path => '' };
2330}
2331
2332sub open_directory {
2333        my ($self, $path, $pb, $rev) = @_;
2334        { path => $path };
2335}
2336
2337sub git_path {
2338        my ($self, $path) = @_;
2339        if ($self->{path_strip}) {
2340                $path =~ s!$self->{path_strip}!! or
2341                  die "Failed to strip path '$path' ($self->{path_strip})\n";
2342        }
2343        $path;
2344}
2345
2346sub delete_entry {
2347        my ($self, $path, $rev, $pb) = @_;
2348
2349        my $gpath = $self->git_path($path);
2350        return undef if ($gpath eq '');
2351
2352        # remove entire directories.
2353        if (command('ls-tree', $self->{c}, '--', $gpath) =~ /^040000 tree/) {
2354                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
2355                                                     -r --name-only -z/,
2356                                                     $self->{c}, '--', $gpath);
2357                local $/ = "\0";
2358                while (<$ls>) {
2359                        chomp;
2360                        $self->{gii}->remove($_);
2361                        print "\tD\t$_\n" unless $::_q;
2362                }
2363                print "\tD\t$gpath/\n" unless $::_q;
2364                command_close_pipe($ls, $ctx);
2365                $self->{empty}->{$path} = 0
2366        } else {
2367                $self->{gii}->remove($gpath);
2368                print "\tD\t$gpath\n" unless $::_q;
2369        }
2370        undef;
2371}
2372
2373sub open_file {
2374        my ($self, $path, $pb, $rev) = @_;
2375        my $gpath = $self->git_path($path);
2376        my ($mode, $blob) = (command('ls-tree', $self->{c}, '--', $gpath)
2377                             =~ /^(\d{6}) blob ([a-f\d]{40})\t/);
2378        unless (defined $mode && defined $blob) {
2379                die "$path was not found in commit $self->{c} (r$rev)\n";
2380        }
2381        { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
2382          pool => SVN::Pool->new, action => 'M' };
2383}
2384
2385sub add_file {
2386        my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
2387        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
2388        delete $self->{empty}->{$dir};
2389        { path => $path, mode_a => 100644, mode_b => 100644,
2390          pool => SVN::Pool->new, action => 'A' };
2391}
2392
2393sub add_directory {
2394        my ($self, $path, $cp_path, $cp_rev) = @_;
2395        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
2396        delete $self->{empty}->{$dir};
2397        $self->{empty}->{$path} = 1;
2398        { path => $path };
2399}
2400
2401sub change_dir_prop {
2402        my ($self, $db, $prop, $value) = @_;
2403        $self->{dir_prop}->{$db->{path}} ||= {};
2404        $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
2405        undef;
2406}
2407
2408sub absent_directory {
2409        my ($self, $path, $pb) = @_;
2410        $self->{absent_dir}->{$pb->{path}} ||= [];
2411        push @{$self->{absent_dir}->{$pb->{path}}}, $path;
2412        undef;
2413}
2414
2415sub absent_file {
2416        my ($self, $path, $pb) = @_;
2417        $self->{absent_file}->{$pb->{path}} ||= [];
2418        push @{$self->{absent_file}->{$pb->{path}}}, $path;
2419        undef;
2420}
2421
2422sub change_file_prop {
2423        my ($self, $fb, $prop, $value) = @_;
2424        if ($prop eq 'svn:executable') {
2425                if ($fb->{mode_b} != 120000) {
2426                        $fb->{mode_b} = defined $value ? 100755 : 100644;
2427                }
2428        } elsif ($prop eq 'svn:special') {
2429                $fb->{mode_b} = defined $value ? 120000 : 100644;
2430        } else {
2431                $self->{file_prop}->{$fb->{path}} ||= {};
2432                $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
2433        }
2434        undef;
2435}
2436
2437sub apply_textdelta {
2438        my ($self, $fb, $exp) = @_;
2439        my $fh = IO::File->new_tmpfile;
2440        $fh->autoflush(1);
2441        # $fh gets auto-closed() by SVN::TxDelta::apply(),
2442        # (but $base does not,) so dup() it for reading in close_file
2443        open my $dup, '<&', $fh or croak $!;
2444        my $base = IO::File->new_tmpfile;
2445        $base->autoflush(1);
2446        if ($fb->{blob}) {
2447                defined (my $pid = fork) or croak $!;
2448                if (!$pid) {
2449                        open STDOUT, '>&', $base or croak $!;
2450                        print STDOUT 'link ' if ($fb->{mode_a} == 120000);
2451                        exec qw/git-cat-file blob/, $fb->{blob} or croak $!;
2452                }
2453                waitpid $pid, 0;
2454                croak $? if $?;
2455
2456                if (defined $exp) {
2457                        seek $base, 0, 0 or croak $!;
2458                        my $md5 = Digest::MD5->new;
2459                        $md5->addfile($base);
2460                        my $got = $md5->hexdigest;
2461                        die "Checksum mismatch: $fb->{path} $fb->{blob}\n",
2462                            "expected: $exp\n",
2463                            "     got: $got\n" if ($got ne $exp);
2464                }
2465        }
2466        seek $base, 0, 0 or croak $!;
2467        $fb->{fh} = $dup;
2468        $fb->{base} = $base;
2469        [ SVN::TxDelta::apply($base, $fh, undef, $fb->{path}, $fb->{pool}) ];
2470}
2471
2472sub close_file {
2473        my ($self, $fb, $exp) = @_;
2474        my $hash;
2475        my $path = $self->git_path($fb->{path});
2476        if (my $fh = $fb->{fh}) {
2477                if (defined $exp) {
2478                        seek($fh, 0, 0) or croak $!;
2479                        my $md5 = Digest::MD5->new;
2480                        $md5->addfile($fh);
2481                        my $got = $md5->hexdigest;
2482                        if ($got ne $exp) {
2483                                die "Checksum mismatch: $path\n",
2484                                    "expected: $exp\n    got: $got\n";
2485                        }
2486                }
2487                sysseek($fh, 0, 0) or croak $!;
2488                if ($fb->{mode_b} == 120000) {
2489                        sysread($fh, my $buf, 5) == 5 or croak $!;
2490                        $buf eq 'link ' or die "$path has mode 120000",
2491                                               "but is not a link\n";
2492                }
2493                defined(my $pid = open my $out,'-|') or die "Can't fork: $!\n";
2494                if (!$pid) {
2495                        open STDIN, '<&', $fh or croak $!;
2496                        exec qw/git-hash-object -w --stdin/ or croak $!;
2497                }
2498                chomp($hash = do { local $/; <$out> });
2499                close $out or croak $!;
2500                close $fh or croak $!;
2501                $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
2502                close $fb->{base} or croak $!;
2503        } else {
2504                $hash = $fb->{blob} or die "no blob information\n";
2505        }
2506        $fb->{pool}->clear;
2507        $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
2508        print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
2509        undef;
2510}
2511
2512sub abort_edit {
2513        my $self = shift;
2514        $self->{nr} = $self->{gii}->{nr};
2515        delete $self->{gii};
2516        $self->SUPER::abort_edit(@_);
2517}
2518
2519sub close_edit {
2520        my $self = shift;
2521        $self->{git_commit_ok} = 1;
2522        $self->{nr} = $self->{gii}->{nr};
2523        delete $self->{gii};
2524        $self->SUPER::close_edit(@_);
2525}
2526
2527package SVN::Git::Editor;
2528use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
2529use strict;
2530use warnings;
2531use Carp qw/croak/;
2532use IO::File;
2533use Digest::MD5;
2534
2535sub new {
2536        my ($class, $opts) = @_;
2537        foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
2538                die "$_ required!\n" unless (defined $opts->{$_});
2539        }
2540
2541        my $pool = SVN::Pool->new;
2542        my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
2543        my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
2544                                     $opts->{r}, $mods);
2545
2546        # $opts->{ra} functions should not be used after this:
2547        my @ce  = $opts->{ra}->get_commit_editor($opts->{log},
2548                                                $opts->{editor_cb}, $pool);
2549        my $self = SVN::Delta::Editor->new(@ce, $pool);
2550        bless $self, $class;
2551        foreach (qw/svn_path r tree_a tree_b/) {
2552                $self->{$_} = $opts->{$_};
2553        }
2554        $self->{url} = $opts->{ra}->{url};
2555        $self->{mods} = $mods;
2556        $self->{types} = $types;
2557        $self->{pool} = $pool;
2558        $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
2559        $self->{rm} = { };
2560        $self->{path_prefix} = length $self->{svn_path} ?
2561                               "$self->{svn_path}/" : '';
2562        return $self;
2563}
2564
2565sub generate_diff {
2566        my ($tree_a, $tree_b) = @_;
2567        my @diff_tree = qw(diff-tree -z -r);
2568        if ($_cp_similarity) {
2569                push @diff_tree, "-C$_cp_similarity";
2570        } else {
2571                push @diff_tree, '-C';
2572        }
2573        push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
2574        push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
2575        push @diff_tree, $tree_a, $tree_b;
2576        my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
2577        local $/ = "\0";
2578        my $state = 'meta';
2579        my @mods;
2580        while (<$diff_fh>) {
2581                chomp $_; # this gets rid of the trailing "\0"
2582                if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
2583                                        $::sha1\s($::sha1)\s
2584                                        ([MTCRAD])\d*$/xo) {
2585                        push @mods, {   mode_a => $1, mode_b => $2,
2586                                        sha1_b => $3, chg => $4 };
2587                        if ($4 =~ /^(?:C|R)$/) {
2588                                $state = 'file_a';
2589                        } else {
2590                                $state = 'file_b';
2591                        }
2592                } elsif ($state eq 'file_a') {
2593                        my $x = $mods[$#mods] or croak "Empty array\n";
2594                        if ($x->{chg} !~ /^(?:C|R)$/) {
2595                                croak "Error parsing $_, $x->{chg}\n";
2596                        }
2597                        $x->{file_a} = $_;
2598                        $state = 'file_b';
2599                } elsif ($state eq 'file_b') {
2600                        my $x = $mods[$#mods] or croak "Empty array\n";
2601                        if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
2602                                croak "Error parsing $_, $x->{chg}\n";
2603                        }
2604                        if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
2605                                croak "Error parsing $_, $x->{chg}\n";
2606                        }
2607                        $x->{file_b} = $_;
2608                        $state = 'meta';
2609                } else {
2610                        croak "Error parsing $_\n";
2611                }
2612        }
2613        command_close_pipe($diff_fh, $ctx);
2614        \@mods;
2615}
2616
2617sub check_diff_paths {
2618        my ($ra, $pfx, $rev, $mods) = @_;
2619        my %types;
2620        $pfx .= '/' if length $pfx;
2621
2622        sub type_diff_paths {
2623                my ($ra, $types, $path, $rev) = @_;
2624                my @p = split m#/+#, $path;
2625                my $c = shift @p;
2626                unless (defined $types->{$c}) {
2627                        $types->{$c} = $ra->check_path($c, $rev);
2628                }
2629                while (@p) {
2630                        $c .= '/' . shift @p;
2631                        next if defined $types->{$c};
2632                        $types->{$c} = $ra->check_path($c, $rev);
2633                }
2634        }
2635
2636        foreach my $m (@$mods) {
2637                foreach my $f (qw/file_a file_b/) {
2638                        next unless defined $m->{$f};
2639                        my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
2640                        if (length $pfx.$dir && ! defined $types{$dir}) {
2641                                type_diff_paths($ra, \%types, $pfx.$dir, $rev);
2642                        }
2643                }
2644        }
2645        \%types;
2646}
2647
2648sub split_path {
2649        return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
2650}
2651
2652sub repo_path {
2653        my ($self, $path) = @_;
2654        $self->{path_prefix}.(defined $path ? $path : '');
2655}
2656
2657sub url_path {
2658        my ($self, $path) = @_;
2659        $self->{url} . '/' . $self->repo_path($path);
2660}
2661
2662sub rmdirs {
2663        my ($self) = @_;
2664        my $rm = $self->{rm};
2665        delete $rm->{''}; # we never delete the url we're tracking
2666        return unless %$rm;
2667
2668        foreach (keys %$rm) {
2669                my @d = split m#/#, $_;
2670                my $c = shift @d;
2671                $rm->{$c} = 1;
2672                while (@d) {
2673                        $c .= '/' . shift @d;
2674                        $rm->{$c} = 1;
2675                }
2676        }
2677        delete $rm->{$self->{svn_path}};
2678        delete $rm->{''}; # we never delete the url we're tracking
2679        return unless %$rm;
2680
2681        my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
2682                                             $self->{tree_b});
2683        local $/ = "\0";
2684        while (<$fh>) {
2685                chomp;
2686                my @dn = split m#/#, $_;
2687                while (pop @dn) {
2688                        delete $rm->{join '/', @dn};
2689                }
2690                unless (%$rm) {
2691                        close $fh;
2692                        return;
2693                }
2694        }
2695        command_close_pipe($fh, $ctx);
2696
2697        my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
2698        foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
2699                $self->close_directory($bat->{$d}, $p);
2700                my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
2701                print "\tD+\t$d/\n" unless $::_q;
2702                $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
2703                delete $bat->{$d};
2704        }
2705}
2706
2707sub open_or_add_dir {
2708        my ($self, $full_path, $baton) = @_;
2709        my $t = $self->{types}->{$full_path};
2710        if (!defined $t) {
2711                die "$full_path not known in r$self->{r} or we have a bug!\n";
2712        }
2713        if ($t == $SVN::Node::none) {
2714                return $self->add_directory($full_path, $baton,
2715                                                undef, -1, $self->{pool});
2716        } elsif ($t == $SVN::Node::dir) {
2717                return $self->open_directory($full_path, $baton,
2718                                                $self->{r}, $self->{pool});
2719        }
2720        print STDERR "$full_path already exists in repository at ",
2721                "r$self->{r} and it is not a directory (",
2722                ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
2723        exit 1;
2724}
2725
2726sub ensure_path {
2727        my ($self, $path) = @_;
2728        my $bat = $self->{bat};
2729        my $repo_path = $self->repo_path($path);
2730        return $bat->{''} unless (length $repo_path);
2731        my @p = split m#/+#, $repo_path;
2732        my $c = shift @p;
2733        $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
2734        while (@p) {
2735                my $c0 = $c;
2736                $c .= '/' . shift @p;
2737                $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
2738        }
2739        return $bat->{$c};
2740}
2741
2742sub A {
2743        my ($self, $m) = @_;
2744        my ($dir, $file) = split_path($m->{file_b});
2745        my $pbat = $self->ensure_path($dir);
2746        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2747                                        undef, -1);
2748        print "\tA\t$m->{file_b}\n" unless $::_q;
2749        $self->chg_file($fbat, $m);
2750        $self->close_file($fbat,undef,$self->{pool});
2751}
2752
2753sub C {
2754        my ($self, $m) = @_;
2755        my ($dir, $file) = split_path($m->{file_b});
2756        my $pbat = $self->ensure_path($dir);
2757        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2758                                $self->url_path($m->{file_a}), $self->{r});
2759        print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
2760        $self->chg_file($fbat, $m);
2761        $self->close_file($fbat,undef,$self->{pool});
2762}
2763
2764sub delete_entry {
2765        my ($self, $path, $pbat) = @_;
2766        my $rpath = $self->repo_path($path);
2767        my ($dir, $file) = split_path($rpath);
2768        $self->{rm}->{$dir} = 1;
2769        $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
2770}
2771
2772sub R {
2773        my ($self, $m) = @_;
2774        my ($dir, $file) = split_path($m->{file_b});
2775        my $pbat = $self->ensure_path($dir);
2776        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2777                                $self->url_path($m->{file_a}), $self->{r});
2778        print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
2779        $self->chg_file($fbat, $m);
2780        $self->close_file($fbat,undef,$self->{pool});
2781
2782        ($dir, $file) = split_path($m->{file_a});
2783        $pbat = $self->ensure_path($dir);
2784        $self->delete_entry($m->{file_a}, $pbat);
2785}
2786
2787sub M {
2788        my ($self, $m) = @_;
2789        my ($dir, $file) = split_path($m->{file_b});
2790        my $pbat = $self->ensure_path($dir);
2791        my $fbat = $self->open_file($self->repo_path($m->{file_b}),
2792                                $pbat,$self->{r},$self->{pool});
2793        print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
2794        $self->chg_file($fbat, $m);
2795        $self->close_file($fbat,undef,$self->{pool});
2796}
2797
2798sub T { shift->M(@_) }
2799
2800sub change_file_prop {
2801        my ($self, $fbat, $pname, $pval) = @_;
2802        $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
2803}
2804
2805sub chg_file {
2806        my ($self, $fbat, $m) = @_;
2807        if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
2808                $self->change_file_prop($fbat,'svn:executable','*');
2809        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
2810                $self->change_file_prop($fbat,'svn:executable',undef);
2811        }
2812        my $fh = IO::File->new_tmpfile or croak $!;
2813        if ($m->{mode_b} =~ /^120/) {
2814                print $fh 'link ' or croak $!;
2815                $self->change_file_prop($fbat,'svn:special','*');
2816        } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
2817                $self->change_file_prop($fbat,'svn:special',undef);
2818        }
2819        defined(my $pid = fork) or croak $!;
2820        if (!$pid) {
2821                open STDOUT, '>&', $fh or croak $!;
2822                exec qw/git-cat-file blob/, $m->{sha1_b} or croak $!;
2823        }
2824        waitpid $pid, 0;
2825        croak $? if $?;
2826        $fh->flush == 0 or croak $!;
2827        seek $fh, 0, 0 or croak $!;
2828
2829        my $md5 = Digest::MD5->new;
2830        $md5->addfile($fh) or croak $!;
2831        seek $fh, 0, 0 or croak $!;
2832
2833        my $exp = $md5->hexdigest;
2834        my $pool = SVN::Pool->new;
2835        my $atd = $self->apply_textdelta($fbat, undef, $pool);
2836        my $got = SVN::TxDelta::send_stream($fh, @$atd, $pool);
2837        die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
2838        $pool->clear;
2839
2840        close $fh or croak $!;
2841}
2842
2843sub D {
2844        my ($self, $m) = @_;
2845        my ($dir, $file) = split_path($m->{file_b});
2846        my $pbat = $self->ensure_path($dir);
2847        print "\tD\t$m->{file_b}\n" unless $::_q;
2848        $self->delete_entry($m->{file_b}, $pbat);
2849}
2850
2851sub close_edit {
2852        my ($self) = @_;
2853        my ($p,$bat) = ($self->{pool}, $self->{bat});
2854        foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
2855                next if $_ eq '';
2856                $self->close_directory($bat->{$_}, $p);
2857        }
2858        $self->close_directory($bat->{''}, $p);
2859        $self->SUPER::close_edit($p);
2860        $p->clear;
2861}
2862
2863sub abort_edit {
2864        my ($self) = @_;
2865        $self->SUPER::abort_edit($self->{pool});
2866}
2867
2868sub DESTROY {
2869        my $self = shift;
2870        $self->SUPER::DESTROY(@_);
2871        $self->{pool}->clear;
2872}
2873
2874# this drives the editor
2875sub apply_diff {
2876        my ($self) = @_;
2877        my $mods = $self->{mods};
2878        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
2879        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
2880                my $f = $m->{chg};
2881                if (defined $o{$f}) {
2882                        $self->$f($m);
2883                } else {
2884                        fatal("Invalid change type: $f\n");
2885                }
2886        }
2887        $self->rmdirs if $_rmdir;
2888        if (@$mods == 0) {
2889                $self->abort_edit;
2890        } else {
2891                $self->close_edit;
2892        }
2893        return scalar @$mods;
2894}
2895
2896package Git::SVN::Ra;
2897use vars qw/@ISA $config_dir $_log_window_size/;
2898use strict;
2899use warnings;
2900my ($can_do_switch, %ignored_err, $RA);
2901
2902BEGIN {
2903        # enforce temporary pool usage for some simple functions
2904        no strict 'refs';
2905        for my $f (qw/rev_proplist get_latest_revnum get_uuid get_repos_root/) {
2906                my $SUPER = "SUPER::$f";
2907                *$f = sub {
2908                        my $self = shift;
2909                        my $pool = SVN::Pool->new;
2910                        my @ret = $self->$SUPER(@_,$pool);
2911                        $pool->clear;
2912                        wantarray ? @ret : $ret[0];
2913                };
2914        }
2915}
2916
2917sub new {
2918        my ($class, $url) = @_;
2919        $url =~ s!/+$!!;
2920        return $RA if ($RA && $RA->{url} eq $url);
2921
2922        SVN::_Core::svn_config_ensure($config_dir, undef);
2923        my ($baton, $callbacks) = SVN::Core::auth_open_helper([
2924            SVN::Client::get_simple_provider(),
2925            SVN::Client::get_ssl_server_trust_file_provider(),
2926            SVN::Client::get_simple_prompt_provider(
2927              \&Git::SVN::Prompt::simple, 2),
2928            SVN::Client::get_ssl_client_cert_prompt_provider(
2929              \&Git::SVN::Prompt::ssl_client_cert, 2),
2930            SVN::Client::get_ssl_client_cert_pw_prompt_provider(
2931              \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
2932            SVN::Client::get_username_provider(),
2933            SVN::Client::get_ssl_server_trust_prompt_provider(
2934              \&Git::SVN::Prompt::ssl_server_trust),
2935            SVN::Client::get_username_prompt_provider(
2936              \&Git::SVN::Prompt::username, 2),
2937          ]);
2938        my $config = SVN::Core::config_get_config($config_dir);
2939        my $self = SVN::Ra->new(url => $url, auth => $baton,
2940                              config => $config,
2941                              pool => SVN::Pool->new,
2942                              auth_provider_callbacks => $callbacks);
2943        $self->{svn_path} = $url;
2944        $self->{repos_root} = $self->get_repos_root;
2945        $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
2946        $self->{cache} = { check_path => { r => 0, data => {} },
2947                           get_dir => { r => 0, data => {} } };
2948        $RA = bless $self, $class;
2949}
2950
2951sub check_path {
2952        my ($self, $path, $r) = @_;
2953        my $cache = $self->{cache}->{check_path};
2954        if ($r == $cache->{r} && exists $cache->{data}->{$path}) {
2955                return $cache->{data}->{$path};
2956        }
2957        my $pool = SVN::Pool->new;
2958        my $t = $self->SUPER::check_path($path, $r, $pool);
2959        $pool->clear;
2960        if ($r != $cache->{r}) {
2961                %{$cache->{data}} = ();
2962                $cache->{r} = $r;
2963        }
2964        $cache->{data}->{$path} = $t;
2965}
2966
2967sub get_dir {
2968        my ($self, $dir, $r) = @_;
2969        my $cache = $self->{cache}->{get_dir};
2970        if ($r == $cache->{r}) {
2971                if (my $x = $cache->{data}->{$dir}) {
2972                        return wantarray ? @$x : $x->[0];
2973                }
2974        }
2975        my $pool = SVN::Pool->new;
2976        my ($d, undef, $props) = $self->SUPER::get_dir($dir, $r, $pool);
2977        my %dirents = map { $_ => { kind => $d->{$_}->kind } } keys %$d;
2978        $pool->clear;
2979        if ($r != $cache->{r}) {
2980                %{$cache->{data}} = ();
2981                $cache->{r} = $r;
2982        }
2983        $cache->{data}->{$dir} = [ \%dirents, $r, $props ];
2984        wantarray ? (\%dirents, $r, $props) : \%dirents;
2985}
2986
2987sub DESTROY {
2988        # do not call the real DESTROY since we store ourselves in $RA
2989}
2990
2991sub get_log {
2992        my ($self, @args) = @_;
2993        my $pool = SVN::Pool->new;
2994        splice(@args, 3, 1) if ($SVN::Core::VERSION le '1.2.0');
2995        my $ret = $self->SUPER::get_log(@args, $pool);
2996        $pool->clear;
2997        $ret;
2998}
2999
3000sub get_commit_editor {
3001        my ($self, $log, $cb, $pool) = @_;
3002        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
3003        $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
3004}
3005
3006sub gs_do_update {
3007        my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
3008        my $new = ($rev_a == $rev_b);
3009        my $path = $gs->{path};
3010
3011        if ($new && -e $gs->{index}) {
3012                unlink $gs->{index} or die
3013                  "Couldn't unlink index: $gs->{index}: $!\n";
3014        }
3015        my $pool = SVN::Pool->new;
3016        $editor->set_path_strip($path);
3017        my (@pc) = split m#/#, $path;
3018        my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
3019                                        1, $editor, $pool);
3020        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
3021
3022        # Since we can't rely on svn_ra_reparent being available, we'll
3023        # just have to do some magic with set_path to make it so
3024        # we only want a partial path.
3025        my $sp = '';
3026        my $final = join('/', @pc);
3027        while (@pc) {
3028                $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
3029                $sp .= '/' if length $sp;
3030                $sp .= shift @pc;
3031        }
3032        die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
3033
3034        $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
3035
3036        $reporter->finish_report($pool);
3037        $pool->clear;
3038        $editor->{git_commit_ok};
3039}
3040
3041# this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
3042# svn_ra_reparent didn't work before 1.4)
3043sub gs_do_switch {
3044        my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
3045        my $path = $gs->{path};
3046        my $pool = SVN::Pool->new;
3047
3048        my $full_url = $self->{url};
3049        my $old_url = $full_url;
3050        $full_url .= "/$path" if length $path;
3051        my ($ra, $reparented);
3052        if ($old_url ne $full_url) {
3053                if ($old_url !~ m#^svn(\+ssh)?://#) {
3054                        SVN::_Ra::svn_ra_reparent($self->{session}, $full_url,
3055                                                  $pool);
3056                        $self->{url} = $full_url;
3057                        $reparented = 1;
3058                } else {
3059                        $ra = Git::SVN::Ra->new($full_url);
3060                }
3061        }
3062        $ra ||= $self;
3063        my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
3064        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
3065        $reporter->set_path('', $rev_a, 0, @lock, $pool);
3066        $reporter->finish_report($pool);
3067
3068        if ($reparented) {
3069                SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
3070                $self->{url} = $old_url;
3071        }
3072
3073        $pool->clear;
3074        $editor->{git_commit_ok};
3075}
3076
3077sub longest_common_path {
3078        my ($gsv, $globs) = @_;
3079        my %common;
3080        my $common_max = scalar @$gsv;
3081
3082        foreach my $gs (@$gsv) {
3083                my @tmp = split m#/#, $gs->{path};
3084                my $p = '';
3085                foreach (@tmp) {
3086                        $p .= length($p) ? "/$_" : $_;
3087                        $common{$p} ||= 0;
3088                        $common{$p}++;
3089                }
3090        }
3091        $globs ||= [];
3092        $common_max += scalar @$globs;
3093        foreach my $glob (@$globs) {
3094                my @tmp = split m#/#, $glob->{path}->{left};
3095                my $p = '';
3096                foreach (@tmp) {
3097                        $p .= length($p) ? "/$_" : $_;
3098                        $common{$p} ||= 0;
3099                        $common{$p}++;
3100                }
3101        }
3102
3103        my $longest_path = '';
3104        foreach (sort {length $b <=> length $a} keys %common) {
3105                if ($common{$_} == $common_max) {
3106                        $longest_path = $_;
3107                        last;
3108                }
3109        }
3110        $longest_path;
3111}
3112
3113sub gs_fetch_loop_common {
3114        my ($self, $base, $head, $gsv, $globs) = @_;
3115        return if ($base > $head);
3116        my $inc = $_log_window_size;
3117        my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
3118        my $longest_path = longest_common_path($gsv, $globs);
3119        while (1) {
3120                my %revs;
3121                my $err;
3122                my $err_handler = $SVN::Error::handler;
3123                $SVN::Error::handler = sub {
3124                        ($err) = @_;
3125                        skip_unknown_revs($err);
3126                };
3127                sub _cb {
3128                        my ($paths, $r, $author, $date, $log) = @_;
3129                        [ dup_changed_paths($paths),
3130                          { author => $author, date => $date, log => $log } ];
3131                }
3132                $self->get_log([$longest_path], $min, $max, 0, 1, 1,
3133                               sub { $revs{$_[1]} = _cb(@_) });
3134                if ($err && $max >= $head) {
3135                        print STDERR "Path '$longest_path' ",
3136                                     "was probably deleted:\n",
3137                                     $err->expanded_message,
3138                                     "\nWill attempt to follow ",
3139                                     "revisions r$min .. r$max ",
3140                                     "committed before the deletion\n";
3141                        my $hi = $max;
3142                        while (--$hi >= $min) {
3143                                my $ok;
3144                                $self->get_log([$longest_path], $min, $hi,
3145                                               0, 1, 1, sub {
3146                                               $ok ||= $_[1];
3147                                               $revs{$_[1]} = _cb(@_) });
3148                                if ($ok) {
3149                                        print STDERR "r$min .. r$ok OK\n";
3150                                        last;
3151                                }
3152                        }
3153                }
3154                $SVN::Error::handler = $err_handler;
3155
3156                my %exists = map { $_->{path} => $_ } @$gsv;
3157                foreach my $r (sort {$a <=> $b} keys %revs) {
3158                        my ($paths, $logged) = @{$revs{$r}};
3159
3160                        foreach my $gs ($self->match_globs(\%exists, $paths,
3161                                                           $globs, $r)) {
3162                                if ($gs->rev_db_max >= $r) {
3163                                        next;
3164                                }
3165                                next unless $gs->match_paths($paths, $r);
3166                                $gs->{logged_rev_props} = $logged;
3167                                if (my $last_commit = $gs->last_commit) {
3168                                        $gs->assert_index_clean($last_commit);
3169                                }
3170                                my $log_entry = $gs->do_fetch($paths, $r);
3171                                if ($log_entry) {
3172                                        $gs->do_git_commit($log_entry);
3173                                }
3174                        }
3175                        foreach my $g (@$globs) {
3176                                my $k = "svn-remote.$g->{remote}." .
3177                                        "$g->{t}-maxRev";
3178                                Git::SVN::tmp_config($k, $r);
3179                        }
3180                }
3181                # pre-fill the .rev_db since it'll eventually get filled in
3182                # with '0' x40 if something new gets committed
3183                foreach my $gs (@$gsv) {
3184                        next if defined $gs->rev_db_get($max);
3185                        $gs->rev_db_set($max, 0 x40);
3186                }
3187                foreach my $g (@$globs) {
3188                        my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
3189                        Git::SVN::tmp_config($k, $max);
3190                }
3191                last if $max >= $head;
3192                $min = $max + 1;
3193                $max += $inc;
3194                $max = $head if ($max > $head);
3195        }
3196}
3197
3198sub match_globs {
3199        my ($self, $exists, $paths, $globs, $r) = @_;
3200
3201        sub get_dir_check {
3202                my ($self, $exists, $g, $r) = @_;
3203                my @x = eval { $self->get_dir($g->{path}->{left}, $r) };
3204                return unless scalar @x == 3;
3205                my $dirents = $x[0];
3206                foreach my $de (keys %$dirents) {
3207                        next if $dirents->{$de}->{kind} != $SVN::Node::dir;
3208                        my $p = $g->{path}->full_path($de);
3209                        next if $exists->{$p};
3210                        next if (length $g->{path}->{right} &&
3211                                 ($self->check_path($p, $r) !=
3212                                  $SVN::Node::dir));
3213                        $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
3214                                         $g->{ref}->full_path($de), 1);
3215                }
3216        }
3217        foreach my $g (@$globs) {
3218                if (my $path = $paths->{"/$g->{path}->{left}"}) {
3219                        if ($path->{action} =~ /^[AR]$/) {
3220                                get_dir_check($self, $exists, $g, $r);
3221                        }
3222                }
3223                foreach (keys %$paths) {
3224                        if (/$g->{path}->{left_regex}/ &&
3225                            !/$g->{path}->{regex}/) {
3226                                next if $paths->{$_}->{action} !~ /^[AR]$/;
3227                                get_dir_check($self, $exists, $g, $r);
3228                        }
3229                        next unless /$g->{path}->{regex}/;
3230                        my $p = $1;
3231                        my $pathname = $g->{path}->full_path($p);
3232                        next if $exists->{$pathname};
3233                        next if ($self->check_path($pathname, $r) !=
3234                                 $SVN::Node::dir);
3235                        $exists->{$pathname} = Git::SVN->init(
3236                                              $self->{url}, $pathname, undef,
3237                                              $g->{ref}->full_path($p), 1);
3238                }
3239                my $c = '';
3240                foreach (split m#/#, $g->{path}->{left}) {
3241                        $c .= "/$_";
3242                        next unless ($paths->{$c} &&
3243                                     ($paths->{$c}->{action} =~ /^[AR]$/));
3244                        get_dir_check($self, $exists, $g, $r);
3245                }
3246        }
3247        values %$exists;
3248}
3249
3250sub minimize_url {
3251        my ($self) = @_;
3252        return $self->{url} if ($self->{url} eq $self->{repos_root});
3253        my $url = $self->{repos_root};
3254        my @components = split(m!/!, $self->{svn_path});
3255        my $c = '';
3256        do {
3257                $url .= "/$c" if length $c;
3258                eval { (ref $self)->new($url)->get_latest_revnum };
3259        } while ($@ && ($c = shift @components));
3260        $url;
3261}
3262
3263sub can_do_switch {
3264        my $self = shift;
3265        unless (defined $can_do_switch) {
3266                my $pool = SVN::Pool->new;
3267                my $rep = eval {
3268                        $self->do_switch(1, '', 0, $self->{url},
3269                                         SVN::Delta::Editor->new, $pool);
3270                };
3271                if ($@) {
3272                        $can_do_switch = 0;
3273                } else {
3274                        $rep->abort_report($pool);
3275                        $can_do_switch = 1;
3276                }
3277                $pool->clear;
3278        }
3279        $can_do_switch;
3280}
3281
3282sub skip_unknown_revs {
3283        my ($err) = @_;
3284        my $errno = $err->apr_err();
3285        # Maybe the branch we're tracking didn't
3286        # exist when the repo started, so it's
3287        # not an error if it doesn't, just continue
3288        #
3289        # Wonderfully consistent library, eh?
3290        # 160013 - svn:// and file://
3291        # 175002 - http(s)://
3292        # 175007 - http(s):// (this repo required authorization, too...)
3293        #   More codes may be discovered later...
3294        if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
3295                my $err_key = $err->expanded_message;
3296                # revision numbers change every time, filter them out
3297                $err_key =~ s/\d+/\0/g;
3298                $err_key = "$errno\0$err_key";
3299                unless ($ignored_err{$err_key}) {
3300                        warn "W: Ignoring error from SVN, path probably ",
3301                             "does not exist: ($errno): ",
3302                             $err->expanded_message,"\n";
3303                        $ignored_err{$err_key} = 1;
3304                }
3305                return;
3306        }
3307        die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
3308}
3309
3310# svn_log_changed_path_t objects passed to get_log are likely to be
3311# overwritten even if only the refs are copied to an external variable,
3312# so we should dup the structures in their entirety.  Using an externally
3313# passed pool (instead of our temporary and quickly cleared pool in
3314# Git::SVN::Ra) does not help matters at all...
3315sub dup_changed_paths {
3316        my ($paths) = @_;
3317        return undef unless $paths;
3318        my %ret;
3319        foreach my $p (keys %$paths) {
3320                my $i = $paths->{$p};
3321                my %s = map { $_ => $i->$_ }
3322                              qw/copyfrom_path copyfrom_rev action/;
3323                $ret{$p} = \%s;
3324        }
3325        \%ret;
3326}
3327
3328package Git::SVN::Log;
3329use strict;
3330use warnings;
3331use POSIX qw/strftime/;
3332use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
3333            %rusers $show_commit $incremental/;
3334my $l_fmt;
3335
3336sub cmt_showable {
3337        my ($c) = @_;
3338        return 1 if defined $c->{r};
3339
3340        # big commit message got truncated by the 16k pretty buffer in rev-list
3341        if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
3342                                $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
3343                @{$c->{l}} = ();
3344                my @log = command(qw/cat-file commit/, $c->{c});
3345
3346                # shift off the headers
3347                shift @log while ($log[0] ne '');
3348                shift @log;
3349
3350                # TODO: make $c->{l} not have a trailing newline in the future
3351                @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
3352
3353                (undef, $c->{r}, undef) = ::extract_metadata(
3354                                (grep(/^git-svn-id: /, @log))[-1]);
3355        }
3356        return defined $c->{r};
3357}
3358
3359sub log_use_color {
3360        return 1 if $color;
3361        my ($dc, $dcvar);
3362        $dcvar = 'color.diff';
3363        $dc = `git-config --get $dcvar`;
3364        if ($dc eq '') {
3365                # nothing at all; fallback to "diff.color"
3366                $dcvar = 'diff.color';
3367                $dc = `git-config --get $dcvar`;
3368        }
3369        chomp($dc);
3370        if ($dc eq 'auto') {
3371                my $pc;
3372                $pc = `git-config --get color.pager`;
3373                if ($pc eq '') {
3374                        # does not have it -- fallback to pager.color
3375                        $pc = `git-config --bool --get pager.color`;
3376                }
3377                else {
3378                        $pc = `git-config --bool --get color.pager`;
3379                        if ($?) {
3380                                $pc = 'false';
3381                        }
3382                }
3383                chomp($pc);
3384                if (-t *STDOUT || (defined $pager && $pc eq 'true')) {
3385                        return ($ENV{TERM} && $ENV{TERM} ne 'dumb');
3386                }
3387                return 0;
3388        }
3389        return 0 if $dc eq 'never';
3390        return 1 if $dc eq 'always';
3391        chomp($dc = `git-config --bool --get $dcvar`);
3392        return ($dc eq 'true');
3393}
3394
3395sub git_svn_log_cmd {
3396        my ($r_min, $r_max, @args) = @_;
3397        my $head = 'HEAD';
3398        foreach my $x (@args) {
3399                last if $x eq '--';
3400                next unless ::verify_ref("$x^0");
3401                $head = $x;
3402                last;
3403        }
3404
3405        my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
3406        $gs ||= Git::SVN->_new;
3407        my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
3408                   $gs->refname);
3409        push @cmd, '-r' unless $non_recursive;
3410        push @cmd, qw/--raw --name-status/ if $verbose;
3411        push @cmd, '--color' if log_use_color();
3412        return @cmd unless defined $r_max;
3413        if ($r_max == $r_min) {
3414                push @cmd, '--max-count=1';
3415                if (my $c = $gs->rev_db_get($r_max)) {
3416                        push @cmd, $c;
3417                }
3418        } else {
3419                my ($c_min, $c_max);
3420                $c_max = $gs->rev_db_get($r_max);
3421                $c_min = $gs->rev_db_get($r_min);
3422                if (defined $c_min && defined $c_max) {
3423                        if ($r_max > $r_max) {
3424                                push @cmd, "$c_min..$c_max";
3425                        } else {
3426                                push @cmd, "$c_max..$c_min";
3427                        }
3428                } elsif ($r_max > $r_min) {
3429                        push @cmd, $c_max;
3430                } else {
3431                        push @cmd, $c_min;
3432                }
3433        }
3434        return @cmd;
3435}
3436
3437# adapted from pager.c
3438sub config_pager {
3439        $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
3440        if (!defined $pager) {
3441                $pager = 'less';
3442        } elsif (length $pager == 0 || $pager eq 'cat') {
3443                $pager = undef;
3444        }
3445}
3446
3447sub run_pager {
3448        return unless -t *STDOUT;
3449        pipe my $rfd, my $wfd or return;
3450        defined(my $pid = fork) or ::fatal "Can't fork: $!\n";
3451        if (!$pid) {
3452                open STDOUT, '>&', $wfd or
3453                                     ::fatal "Can't redirect to stdout: $!\n";
3454                return;
3455        }
3456        open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!\n";
3457        $ENV{LESS} ||= 'FRSX';
3458        exec $pager or ::fatal "Can't run pager: $! ($pager)\n";
3459}
3460
3461sub tz_to_s_offset {
3462        my ($tz) = @_;
3463        $tz =~ s/(\d\d)$//;
3464        return ($1 * 60) + ($tz * 3600);
3465}
3466
3467sub get_author_info {
3468        my ($dest, $author, $t, $tz) = @_;
3469        $author =~ s/(?:^\s*|\s*$)//g;
3470        $dest->{a_raw} = $author;
3471        my $au;
3472        if ($::_authors) {
3473                $au = $rusers{$author} || undef;
3474        }
3475        if (!$au) {
3476                ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
3477        }
3478        $dest->{t} = $t;
3479        $dest->{tz} = $tz;
3480        $dest->{a} = $au;
3481        # Date::Parse isn't in the standard Perl distro :(
3482        if ($tz =~ s/^\+//) {
3483                $t += tz_to_s_offset($tz);
3484        } elsif ($tz =~ s/^\-//) {
3485                $t -= tz_to_s_offset($tz);
3486        }
3487        $dest->{t_utc} = $t;
3488}
3489
3490sub process_commit {
3491        my ($c, $r_min, $r_max, $defer) = @_;
3492        if (defined $r_min && defined $r_max) {
3493                if ($r_min == $c->{r} && $r_min == $r_max) {
3494                        show_commit($c);
3495                        return 0;
3496                }
3497                return 1 if $r_min == $r_max;
3498                if ($r_min < $r_max) {
3499                        # we need to reverse the print order
3500                        return 0 if (defined $limit && --$limit < 0);
3501                        push @$defer, $c;
3502                        return 1;
3503                }
3504                if ($r_min != $r_max) {
3505                        return 1 if ($r_min < $c->{r});
3506                        return 1 if ($r_max > $c->{r});
3507                }
3508        }
3509        return 0 if (defined $limit && --$limit < 0);
3510        show_commit($c);
3511        return 1;
3512}
3513
3514sub show_commit {
3515        my $c = shift;
3516        if ($oneline) {
3517                my $x = "\n";
3518                if (my $l = $c->{l}) {
3519                        while ($l->[0] =~ /^\s*$/) { shift @$l }
3520                        $x = $l->[0];
3521                }
3522                $l_fmt ||= 'A' . length($c->{r});
3523                print 'r',pack($l_fmt, $c->{r}),' | ';
3524                print "$c->{c} | " if $show_commit;
3525                print $x;
3526        } else {
3527                show_commit_normal($c);
3528        }
3529}
3530
3531sub show_commit_changed_paths {
3532        my ($c) = @_;
3533        return unless $c->{changed};
3534        print "Changed paths:\n", @{$c->{changed}};
3535}
3536
3537sub show_commit_normal {
3538        my ($c) = @_;
3539        print '-' x72, "\nr$c->{r} | ";
3540        print "$c->{c} | " if $show_commit;
3541        print "$c->{a} | ", strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)",
3542                                 localtime($c->{t_utc})), ' | ';
3543        my $nr_line = 0;
3544
3545        if (my $l = $c->{l}) {
3546                while ($l->[$#$l] eq "\n" && $#$l > 0
3547                                          && $l->[($#$l - 1)] eq "\n") {
3548                        pop @$l;
3549                }
3550                $nr_line = scalar @$l;
3551                if (!$nr_line) {
3552                        print "1 line\n\n\n";
3553                } else {
3554                        if ($nr_line == 1) {
3555                                $nr_line = '1 line';
3556                        } else {
3557                                $nr_line .= ' lines';
3558                        }
3559                        print $nr_line, "\n";
3560                        show_commit_changed_paths($c);
3561                        print "\n";
3562                        print $_ foreach @$l;
3563                }
3564        } else {
3565                print "1 line\n";
3566                show_commit_changed_paths($c);
3567                print "\n";
3568
3569        }
3570        foreach my $x (qw/raw stat diff/) {
3571                if ($c->{$x}) {
3572                        print "\n";
3573                        print $_ foreach @{$c->{$x}}
3574                }
3575        }
3576}
3577
3578sub cmd_show_log {
3579        my (@args) = @_;
3580        my ($r_min, $r_max);
3581        my $r_last = -1; # prevent dupes
3582        if (defined $TZ) {
3583                $ENV{TZ} = $TZ;
3584        } else {
3585                delete $ENV{TZ};
3586        }
3587        if (defined $::_revision) {
3588                if ($::_revision =~ /^(\d+):(\d+)$/) {
3589                        ($r_min, $r_max) = ($1, $2);
3590                } elsif ($::_revision =~ /^\d+$/) {
3591                        $r_min = $r_max = $::_revision;
3592                } else {
3593                        ::fatal "-r$::_revision is not supported, use ",
3594                                "standard \'git log\' arguments instead\n";
3595                }
3596        }
3597
3598        config_pager();
3599        @args = (git_svn_log_cmd($r_min, $r_max, @args), @args);
3600        my $log = command_output_pipe(@args);
3601        run_pager();
3602        my (@k, $c, $d, $stat);
3603        my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
3604        while (<$log>) {
3605                if (/^${esc_color}commit ($::sha1_short)/o) {
3606                        my $cmt = $1;
3607                        if ($c && cmt_showable($c) && $c->{r} != $r_last) {
3608                                $r_last = $c->{r};
3609                                process_commit($c, $r_min, $r_max, \@k) or
3610                                                                goto out;
3611                        }
3612                        $d = undef;
3613                        $c = { c => $cmt };
3614                } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
3615                        get_author_info($c, $1, $2, $3);
3616                } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
3617                        # ignore
3618                } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
3619                        push @{$c->{raw}}, $_;
3620                } elsif (/^${esc_color}[ACRMDT]\t/) {
3621                        # we could add $SVN->{svn_path} here, but that requires
3622                        # remote access at the moment (repo_path_split)...
3623                        s#^(${esc_color})([ACRMDT])\t#$1   $2 #o;
3624                        push @{$c->{changed}}, $_;
3625                } elsif (/^${esc_color}diff /o) {
3626                        $d = 1;
3627                        push @{$c->{diff}}, $_;
3628                } elsif ($d) {
3629                        push @{$c->{diff}}, $_;
3630                } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
3631                          $esc_color*[\+\-]*$esc_color$/x) {
3632                        $stat = 1;
3633                        push @{$c->{stat}}, $_;
3634                } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
3635                        push @{$c->{stat}}, $_;
3636                        $stat = undef;
3637                } elsif (/^${esc_color}    (git-svn-id:.+)$/o) {
3638                        ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
3639                } elsif (s/^${esc_color}    //o) {
3640                        push @{$c->{l}}, $_;
3641                }
3642        }
3643        if ($c && defined $c->{r} && $c->{r} != $r_last) {
3644                $r_last = $c->{r};
3645                process_commit($c, $r_min, $r_max, \@k);
3646        }
3647        if (@k) {
3648                my $swap = $r_max;
3649                $r_max = $r_min;
3650                $r_min = $swap;
3651                process_commit($_, $r_min, $r_max) foreach reverse @k;
3652        }
3653out:
3654        close $log;
3655        print '-' x72,"\n" unless $incremental || $oneline;
3656}
3657
3658package Git::SVN::Migration;
3659# these version numbers do NOT correspond to actual version numbers
3660# of git nor git-svn.  They are just relative.
3661#
3662# v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
3663#
3664# v1 layout: .git/$id/info/url, refs/remotes/$id
3665#
3666# v2 layout: .git/svn/$id/info/url, refs/remotes/$id
3667#
3668# v3 layout: .git/svn/$id, refs/remotes/$id
3669#            - info/url may remain for backwards compatibility
3670#            - this is what we migrate up to this layout automatically,
3671#            - this will be used by git svn init on single branches
3672# v3.1 layout (auto migrated):
3673#            - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
3674#              for backwards compatibility
3675#
3676# v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
3677#            - this is only created for newly multi-init-ed
3678#              repositories.  Similar in spirit to the
3679#              --use-separate-remotes option in git-clone (now default)
3680#            - we do not automatically migrate to this (following
3681#              the example set by core git)
3682use strict;
3683use warnings;
3684use Carp qw/croak/;
3685use File::Path qw/mkpath/;
3686use File::Basename qw/dirname basename/;
3687use vars qw/$_minimize/;
3688
3689sub migrate_from_v0 {
3690        my $git_dir = $ENV{GIT_DIR};
3691        return undef unless -d $git_dir;
3692        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
3693        my $migrated = 0;
3694        while (<$fh>) {
3695                chomp;
3696                my ($id, $orig_ref) = ($_, $_);
3697                next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
3698                next unless -f "$git_dir/$id/info/url";
3699                my $new_ref = "refs/remotes/$id";
3700                if (::verify_ref("$new_ref^0")) {
3701                        print STDERR "W: $orig_ref is probably an old ",
3702                                     "branch used by an ancient version of ",
3703                                     "git-svn.\n",
3704                                     "However, $new_ref also exists.\n",
3705                                     "We will not be able ",
3706                                     "to use this branch until this ",
3707                                     "ambiguity is resolved.\n";
3708                        next;
3709                }
3710                print STDERR "Migrating from v0 layout...\n" if !$migrated;
3711                print STDERR "Renaming ref: $orig_ref => $new_ref\n";
3712                command_noisy('update-ref', $new_ref, $orig_ref);
3713                command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
3714                $migrated++;
3715        }
3716        command_close_pipe($fh, $ctx);
3717        print STDERR "Done migrating from v0 layout...\n" if $migrated;
3718        $migrated;
3719}
3720
3721sub migrate_from_v1 {
3722        my $git_dir = $ENV{GIT_DIR};
3723        my $migrated = 0;
3724        return $migrated unless -d $git_dir;
3725        my $svn_dir = "$git_dir/svn";
3726
3727        # just in case somebody used 'svn' as their $id at some point...
3728        return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
3729
3730        print STDERR "Migrating from a git-svn v1 layout...\n";
3731        mkpath([$svn_dir]);
3732        print STDERR "Data from a previous version of git-svn exists, but\n\t",
3733                     "$svn_dir\n\t(required for this version ",
3734                     "($::VERSION) of git-svn) does not. exist\n";
3735        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
3736        while (<$fh>) {
3737                my $x = $_;
3738                next unless $x =~ s#^refs/remotes/##;
3739                chomp $x;
3740                next unless -f "$git_dir/$x/info/url";
3741                my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
3742                next unless $u;
3743                my $dn = dirname("$git_dir/svn/$x");
3744                mkpath([$dn]) unless -d $dn;
3745                if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
3746                        mkpath(["$git_dir/svn/svn"]);
3747                        print STDERR " - $git_dir/$x/info => ",
3748                                        "$git_dir/svn/$x/info\n";
3749                        rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
3750                               croak "$!: $x";
3751                        # don't worry too much about these, they probably
3752                        # don't exist with repos this old (save for index,
3753                        # and we can easily regenerate that)
3754                        foreach my $f (qw/unhandled.log index .rev_db/) {
3755                                rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
3756                        }
3757                } else {
3758                        print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
3759                        rename "$git_dir/$x", "$git_dir/svn/$x" or
3760                               croak "$!: $x";
3761                }
3762                $migrated++;
3763        }
3764        command_close_pipe($fh, $ctx);
3765        print STDERR "Done migrating from a git-svn v1 layout\n";
3766        $migrated;
3767}
3768
3769sub read_old_urls {
3770        my ($l_map, $pfx, $path) = @_;
3771        my @dir;
3772        foreach (<$path/*>) {
3773                if (-r "$_/info/url") {
3774                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
3775                        my $ref_id = $pfx . basename $_;
3776                        my $url = ::file_to_s("$_/info/url");
3777                        $l_map->{$ref_id} = $url;
3778                } elsif (-d $_) {
3779                        push @dir, $_;
3780                }
3781        }
3782        foreach (@dir) {
3783                my $x = $_;
3784                $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
3785                read_old_urls($l_map, $x, $_);
3786        }
3787}
3788
3789sub migrate_from_v2 {
3790        my @cfg = command(qw/config -l/);
3791        return if grep /^svn-remote\..+\.url=/, @cfg;
3792        my %l_map;
3793        read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
3794        my $migrated = 0;
3795
3796        foreach my $ref_id (sort keys %l_map) {
3797                eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
3798                if ($@) {
3799                        Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
3800                }
3801                $migrated++;
3802        }
3803        $migrated;
3804}
3805
3806sub minimize_connections {
3807        my $r = Git::SVN::read_all_remotes();
3808        my $new_urls = {};
3809        my $root_repos = {};
3810        foreach my $repo_id (keys %$r) {
3811                my $url = $r->{$repo_id}->{url} or next;
3812                my $fetch = $r->{$repo_id}->{fetch} or next;
3813                my $ra = Git::SVN::Ra->new($url);
3814
3815                # skip existing cases where we already connect to the root
3816                if (($ra->{url} eq $ra->{repos_root}) ||
3817                    (Git::SVN::sanitize_remote_name($ra->{repos_root}) eq
3818                     $repo_id)) {
3819                        $root_repos->{$ra->{url}} = $repo_id;
3820                        next;
3821                }
3822
3823                my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
3824                my $root_path = $ra->{url};
3825                $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
3826                foreach my $path (keys %$fetch) {
3827                        my $ref_id = $fetch->{$path};
3828                        my $gs = Git::SVN->new($ref_id, $repo_id, $path);
3829
3830                        # make sure we can read when connecting to
3831                        # a higher level of a repository
3832                        my ($last_rev, undef) = $gs->last_rev_commit;
3833                        if (!defined $last_rev) {
3834                                $last_rev = eval {
3835                                        $root_ra->get_latest_revnum;
3836                                };
3837                                next if $@;
3838                        }
3839                        my $new = $root_path;
3840                        $new .= length $path ? "/$path" : '';
3841                        eval {
3842                                $root_ra->get_log([$new], $last_rev, $last_rev,
3843                                                  0, 0, 1, sub { });
3844                        };
3845                        next if $@;
3846                        $new_urls->{$ra->{repos_root}}->{$new} =
3847                                { ref_id => $ref_id,
3848                                  old_repo_id => $repo_id,
3849                                  old_path => $path };
3850                }
3851        }
3852
3853        my @emptied;
3854        foreach my $url (keys %$new_urls) {
3855                # see if we can re-use an existing [svn-remote "repo_id"]
3856                # instead of creating a(n ugly) new section:
3857                my $repo_id = $root_repos->{$url} ||
3858                              Git::SVN::sanitize_remote_name($url);
3859
3860                my $fetch = $new_urls->{$url};
3861                foreach my $path (keys %$fetch) {
3862                        my $x = $fetch->{$path};
3863                        Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
3864                        my $pfx = "svn-remote.$x->{old_repo_id}";
3865
3866                        my $old_fetch = quotemeta("$x->{old_path}:".
3867                                                  "refs/remotes/$x->{ref_id}");
3868                        command_noisy(qw/config --unset/,
3869                                      "$pfx.fetch", '^'. $old_fetch . '$');
3870                        delete $r->{$x->{old_repo_id}}->
3871                               {fetch}->{$x->{old_path}};
3872                        if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
3873                                command_noisy(qw/config --unset/,
3874                                              "$pfx.url");
3875                                push @emptied, $x->{old_repo_id}
3876                        }
3877                }
3878        }
3879        if (@emptied) {
3880                my $file = $ENV{GIT_CONFIG} || $ENV{GIT_CONFIG_LOCAL} ||
3881                           "$ENV{GIT_DIR}/config";
3882                print STDERR <<EOF;
3883The following [svn-remote] sections in your config file ($file) are empty
3884and can be safely removed:
3885EOF
3886                print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
3887        }
3888}
3889
3890sub migration_check {
3891        migrate_from_v0();
3892        migrate_from_v1();
3893        migrate_from_v2();
3894        minimize_connections() if $_minimize;
3895}
3896
3897package Git::IndexInfo;
3898use strict;
3899use warnings;
3900use Git qw/command_input_pipe command_close_pipe/;
3901
3902sub new {
3903        my ($class) = @_;
3904        my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
3905        bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
3906}
3907
3908sub remove {
3909        my ($self, $path) = @_;
3910        if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
3911                return ++$self->{nr};
3912        }
3913        undef;
3914}
3915
3916sub update {
3917        my ($self, $mode, $hash, $path) = @_;
3918        if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
3919                return ++$self->{nr};
3920        }
3921        undef;
3922}
3923
3924sub DESTROY {
3925        my ($self) = @_;
3926        command_close_pipe($self->{gui}, $self->{ctx});
3927}
3928
3929package Git::SVN::GlobSpec;
3930use strict;
3931use warnings;
3932
3933sub new {
3934        my ($class, $glob) = @_;
3935        my $re = $glob;
3936        $re =~ s!/+$!!g; # no need for trailing slashes
3937        my $nr = ($re =~ s!^(.*)\*(.*)$!\(\[^/\]+\)!g);
3938        my ($left, $right) = ($1, $2);
3939        if ($nr > 1) {
3940                die "Only one '*' wildcard expansion ",
3941                    "is supported (got $nr): '$glob'\n";
3942        } elsif ($nr == 0) {
3943                die "One '*' is needed for glob: '$glob'\n";
3944        }
3945        $re = quotemeta($left) . $re . quotemeta($right);
3946        if (length $left && !($left =~ s!/+$!!g)) {
3947                die "Missing trailing '/' on left side of: '$glob' ($left)\n";
3948        }
3949        if (length $right && !($right =~ s!^/+!!g)) {
3950                die "Missing leading '/' on right side of: '$glob' ($right)\n";
3951        }
3952        my $left_re = qr/^\/\Q$left\E(\/|$)/;
3953        bless { left => $left, right => $right, left_regex => $left_re,
3954                regex => qr/$re/, glob => $glob }, $class;
3955}
3956
3957sub full_path {
3958        my ($self, $path) = @_;
3959        return (length $self->{left} ? "$self->{left}/" : '') .
3960               $path . (length $self->{right} ? "/$self->{right}" : '');
3961}
3962
3963__END__
3964
3965Data structures:
3966
3967
3968$remotes = { # returned by read_all_remotes()
3969        'svn' => {
3970                # svn-remote.svn.url=https://svn.musicpd.org
3971                url => 'https://svn.musicpd.org',
3972                # svn-remote.svn.fetch=mpd/trunk:trunk
3973                fetch => {
3974                        'mpd/trunk' => 'trunk',
3975                },
3976                # svn-remote.svn.tags=mpd/tags/*:tags/*
3977                tags => {
3978                        path => {
3979                                left => 'mpd/tags',
3980                                right => '',
3981                                regex => qr!mpd/tags/([^/]+)$!,
3982                                glob => 'tags/*',
3983                        },
3984                        ref => {
3985                                left => 'tags',
3986                                right => '',
3987                                regex => qr!tags/([^/]+)$!,
3988                                glob => 'tags/*',
3989                        },
3990                }
3991        }
3992};
3993
3994$log_entry hashref as returned by libsvn_log_entry()
3995{
3996        log => 'whitespace-formatted log entry
3997',                                              # trailing newline is preserved
3998        revision => '8',                        # integer
3999        date => '2004-02-24T17:01:44.108345Z',  # commit date
4000        author => 'committer name'
4001};
4002
4003
4004# this is generated by generate_diff();
4005@mods = array of diff-index line hashes, each element represents one line
4006        of diff-index output
4007
4008diff-index line ($m hash)
4009{
4010        mode_a => first column of diff-index output, no leading ':',
4011        mode_b => second column of diff-index output,
4012        sha1_b => sha1sum of the final blob,
4013        chg => change type [MCRADT],
4014        file_a => original file name of a file (iff chg is 'C' or 'R')
4015        file_b => new/current file name of a file (any chg)
4016}
4017;
4018
4019# retval of read_url_paths{,_all}();
4020$l_map = {
4021        # repository root url
4022        'https://svn.musicpd.org' => {
4023                # repository path               # GIT_SVN_ID
4024                'mpd/trunk'             =>      'trunk',
4025                'mpd/tags/0.11.5'       =>      'tags/0.11.5',
4026        },
4027}
4028
4029Notes:
4030        I don't trust the each() function on unless I created %hash myself
4031        because the internal iterator may not have started at base.