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