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