5e61dd9c12794043a44044ac09c9fcc1c75bc606
   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 $_repository
   8                $_q $_authors %users/;
   9$AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
  10$VERSION = '@@GIT_VERSION@@';
  11
  12# From which subdir have we been invoked?
  13my $cmd_dir_prefix = eval {
  14        command_oneline([qw/rev-parse --show-prefix/], STDERR => 0)
  15} || '';
  16
  17my $git_dir_user_set = 1 if defined $ENV{GIT_DIR};
  18$ENV{GIT_DIR} ||= '.git';
  19$Git::SVN::default_repo_id = 'svn';
  20$Git::SVN::default_ref_id = $ENV{GIT_SVN_ID} || 'git-svn';
  21$Git::SVN::Ra::_log_window_size = 100;
  22
  23$Git::SVN::Log::TZ = $ENV{TZ};
  24$ENV{TZ} = 'UTC';
  25$| = 1; # unbuffer STDOUT
  26
  27sub fatal (@) { print STDERR "@_\n"; exit 1 }
  28require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
  29require SVN::Ra;
  30require SVN::Delta;
  31if ($SVN::Core::VERSION lt '1.1.0') {
  32        fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)";
  33}
  34push @Git::SVN::Ra::ISA, 'SVN::Ra';
  35push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
  36push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
  37use Carp qw/croak/;
  38use Digest::MD5;
  39use IO::File qw//;
  40use File::Basename qw/dirname basename/;
  41use File::Path qw/mkpath/;
  42use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev/;
  43use IPC::Open3;
  44use Git;
  45
  46BEGIN {
  47        # import functions from Git into our packages, en masse
  48        no strict 'refs';
  49        foreach (qw/command command_oneline command_noisy command_output_pipe
  50                    command_input_pipe command_close_pipe/) {
  51                for my $package ( qw(SVN::Git::Editor SVN::Git::Fetcher
  52                        Git::SVN::Migration Git::SVN::Log Git::SVN),
  53                        __PACKAGE__) {
  54                        *{"${package}::$_"} = \&{"Git::$_"};
  55                }
  56        }
  57}
  58
  59my ($SVN);
  60
  61$sha1 = qr/[a-f\d]{40}/;
  62$sha1_short = qr/[a-f\d]{4,40}/;
  63my ($_stdin, $_help, $_edit,
  64        $_message, $_file,
  65        $_template, $_shared,
  66        $_version, $_fetch_all, $_no_rebase,
  67        $_merge, $_strategy, $_dry_run, $_local,
  68        $_prefix, $_no_checkout, $_url, $_verbose,
  69        $_git_format, $_commit_url);
  70$Git::SVN::_follow_parent = 1;
  71my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
  72                    'config-dir=s' => \$Git::SVN::Ra::config_dir,
  73                    'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache );
  74my %fc_opts = ( 'follow-parent|follow!' => \$Git::SVN::_follow_parent,
  75                'authors-file|A=s' => \$_authors,
  76                'repack:i' => \$Git::SVN::_repack,
  77                'noMetadata' => \$Git::SVN::_no_metadata,
  78                'useSvmProps' => \$Git::SVN::_use_svm_props,
  79                'useSvnsyncProps' => \$Git::SVN::_use_svnsync_props,
  80                'log-window-size=i' => \$Git::SVN::Ra::_log_window_size,
  81                'no-checkout' => \$_no_checkout,
  82                'quiet|q' => \$_q,
  83                'repack-flags|repack-args|repack-opts=s' =>
  84                   \$Git::SVN::_repack_flags,
  85                'use-log-author' => \$Git::SVN::_use_log_author,
  86                'add-author-from' => \$Git::SVN::_add_author_from,
  87                %remote_opts );
  88
  89my ($_trunk, $_tags, $_branches, $_stdlayout);
  90my %icv;
  91my %init_opts = ( 'template=s' => \$_template, 'shared:s' => \$_shared,
  92                  'trunk|T=s' => \$_trunk, 'tags|t=s' => \$_tags,
  93                  'branches|b=s' => \$_branches, 'prefix=s' => \$_prefix,
  94                  'stdlayout|s' => \$_stdlayout,
  95                  'minimize-url|m' => \$Git::SVN::_minimize_url,
  96                  'no-metadata' => sub { $icv{noMetadata} = 1 },
  97                  'use-svm-props' => sub { $icv{useSvmProps} = 1 },
  98                  'use-svnsync-props' => sub { $icv{useSvnsyncProps} = 1 },
  99                  'rewrite-root=s' => sub { $icv{rewriteRoot} = $_[1] },
 100                  %remote_opts );
 101my %cmt_opts = ( 'edit|e' => \$_edit,
 102                'rmdir' => \$SVN::Git::Editor::_rmdir,
 103                'find-copies-harder' => \$SVN::Git::Editor::_find_copies_harder,
 104                'l=i' => \$SVN::Git::Editor::_rename_limit,
 105                'copy-similarity|C=i'=> \$SVN::Git::Editor::_cp_similarity
 106);
 107
 108my %cmd = (
 109        fetch => [ \&cmd_fetch, "Download new revisions from SVN",
 110                        { 'revision|r=s' => \$_revision,
 111                          'fetch-all|all' => \$_fetch_all,
 112                           %fc_opts } ],
 113        clone => [ \&cmd_clone, "Initialize and fetch revisions",
 114                        { 'revision|r=s' => \$_revision,
 115                           %fc_opts, %init_opts } ],
 116        init => [ \&cmd_init, "Initialize a repo for tracking" .
 117                          " (requires URL argument)",
 118                          \%init_opts ],
 119        'multi-init' => [ \&cmd_multi_init,
 120                          "Deprecated alias for ".
 121                          "'$0 init -T<trunk> -b<branches> -t<tags>'",
 122                          \%init_opts ],
 123        dcommit => [ \&cmd_dcommit,
 124                     'Commit several diffs to merge with upstream',
 125                        { 'merge|m|M' => \$_merge,
 126                          'strategy|s=s' => \$_strategy,
 127                          'verbose|v' => \$_verbose,
 128                          'dry-run|n' => \$_dry_run,
 129                          'fetch-all|all' => \$_fetch_all,
 130                          'commit-url=s' => \$_commit_url,
 131                          'revision|r=i' => \$_revision,
 132                          'no-rebase' => \$_no_rebase,
 133                        %cmt_opts, %fc_opts } ],
 134        'set-tree' => [ \&cmd_set_tree,
 135                        "Set an SVN repository to a git tree-ish",
 136                        { 'stdin|' => \$_stdin, %cmt_opts, %fc_opts, } ],
 137        'create-ignore' => [ \&cmd_create_ignore,
 138                             'Create a .gitignore per svn:ignore',
 139                             { 'revision|r=i' => \$_revision
 140                             } ],
 141        'propget' => [ \&cmd_propget,
 142                       'Print the value of a property on a file or directory',
 143                       { 'revision|r=i' => \$_revision } ],
 144        'proplist' => [ \&cmd_proplist,
 145                       'List all properties of a file or directory',
 146                       { 'revision|r=i' => \$_revision } ],
 147        'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
 148                        { 'revision|r=i' => \$_revision
 149                        } ],
 150        'show-externals' => [ \&cmd_show_externals, "Show svn:externals listings",
 151                        { 'revision|r=i' => \$_revision
 152                        } ],
 153        'multi-fetch' => [ \&cmd_multi_fetch,
 154                           "Deprecated alias for $0 fetch --all",
 155                           { 'revision|r=s' => \$_revision, %fc_opts } ],
 156        'migrate' => [ sub { },
 157                       # no-op, we automatically run this anyways,
 158                       'Migrate configuration/metadata/layout from
 159                        previous versions of git-svn',
 160                       { 'minimize' => \$Git::SVN::Migration::_minimize,
 161                         %remote_opts } ],
 162        'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
 163                        { 'limit=i' => \$Git::SVN::Log::limit,
 164                          'revision|r=s' => \$_revision,
 165                          'verbose|v' => \$Git::SVN::Log::verbose,
 166                          'incremental' => \$Git::SVN::Log::incremental,
 167                          'oneline' => \$Git::SVN::Log::oneline,
 168                          'show-commit' => \$Git::SVN::Log::show_commit,
 169                          'non-recursive' => \$Git::SVN::Log::non_recursive,
 170                          'authors-file|A=s' => \$_authors,
 171                          'color' => \$Git::SVN::Log::color,
 172                          'pager=s' => \$Git::SVN::Log::pager
 173                        } ],
 174        'find-rev' => [ \&cmd_find_rev,
 175                        "Translate between SVN revision numbers and tree-ish",
 176                        {} ],
 177        'rebase' => [ \&cmd_rebase, "Fetch and rebase your working directory",
 178                        { 'merge|m|M' => \$_merge,
 179                          'verbose|v' => \$_verbose,
 180                          'strategy|s=s' => \$_strategy,
 181                          'local|l' => \$_local,
 182                          'fetch-all|all' => \$_fetch_all,
 183                          'dry-run|n' => \$_dry_run,
 184                          %fc_opts } ],
 185        'commit-diff' => [ \&cmd_commit_diff,
 186                           'Commit a diff between two trees',
 187                        { 'message|m=s' => \$_message,
 188                          'file|F=s' => \$_file,
 189                          'revision|r=s' => \$_revision,
 190                        %cmt_opts } ],
 191        'info' => [ \&cmd_info,
 192                    "Show info about the latest SVN revision
 193                     on the current branch",
 194                    { 'url' => \$_url, } ],
 195        'blame' => [ \&Git::SVN::Log::cmd_blame,
 196                    "Show what revision and author last modified each line of a file",
 197                    { 'git-format' => \$_git_format } ],
 198);
 199
 200my $cmd;
 201for (my $i = 0; $i < @ARGV; $i++) {
 202        if (defined $cmd{$ARGV[$i]}) {
 203                $cmd = $ARGV[$i];
 204                splice @ARGV, $i, 1;
 205                last;
 206        }
 207};
 208
 209# make sure we're always running at the top-level working directory
 210unless ($cmd && $cmd =~ /(?:clone|init|multi-init)$/) {
 211        unless (-d $ENV{GIT_DIR}) {
 212                if ($git_dir_user_set) {
 213                        die "GIT_DIR=$ENV{GIT_DIR} explicitly set, ",
 214                            "but it is not a directory\n";
 215                }
 216                my $git_dir = delete $ENV{GIT_DIR};
 217                chomp(my $cdup = command_oneline(qw/rev-parse --show-cdup/));
 218                unless (length $cdup) {
 219                        die "Already at toplevel, but $git_dir ",
 220                            "not found '$cdup'\n";
 221                }
 222                chdir $cdup or die "Unable to chdir up to '$cdup'\n";
 223                unless (-d $git_dir) {
 224                        die "$git_dir still not found after going to ",
 225                            "'$cdup'\n";
 226                }
 227                $ENV{GIT_DIR} = $git_dir;
 228        }
 229        $_repository = Git->repository(Repository => $ENV{GIT_DIR});
 230}
 231
 232my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
 233
 234read_repo_config(\%opts);
 235if ($cmd && ($cmd eq 'log' || $cmd eq 'blame')) {
 236        Getopt::Long::Configure('pass_through');
 237}
 238my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
 239                    'minimize-connections' => \$Git::SVN::Migration::_minimize,
 240                    'id|i=s' => \$Git::SVN::default_ref_id,
 241                    'svn-remote|remote|R=s' => sub {
 242                       $Git::SVN::no_reuse_existing = 1;
 243                       $Git::SVN::default_repo_id = $_[1] });
 244exit 1 if (!$rv && $cmd && $cmd ne 'log');
 245
 246usage(0) if $_help;
 247version() if $_version;
 248usage(1) unless defined $cmd;
 249load_authors() if $_authors;
 250
 251unless ($cmd =~ /^(?:clone|init|multi-init|commit-diff)$/) {
 252        Git::SVN::Migration::migration_check();
 253}
 254Git::SVN::init_vars();
 255eval {
 256        Git::SVN::verify_remotes_sanity();
 257        $cmd{$cmd}->[0]->(@ARGV);
 258};
 259fatal $@ if $@;
 260post_fetch_checkout();
 261exit 0;
 262
 263####################### primary functions ######################
 264sub usage {
 265        my $exit = shift || 0;
 266        my $fd = $exit ? \*STDERR : \*STDOUT;
 267        print $fd <<"";
 268git-svn - bidirectional operations between a single Subversion tree and git
 269Usage: git svn <command> [options] [arguments]\n
 270
 271        print $fd "Available commands:\n" unless $cmd;
 272
 273        foreach (sort keys %cmd) {
 274                next if $cmd && $cmd ne $_;
 275                next if /^multi-/; # don't show deprecated commands
 276                print $fd '  ',pack('A17',$_),$cmd{$_}->[1],"\n";
 277                foreach (sort keys %{$cmd{$_}->[2]}) {
 278                        # mixed-case options are for .git/config only
 279                        next if /[A-Z]/ && /^[a-z]+$/i;
 280                        # prints out arguments as they should be passed:
 281                        my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
 282                        print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
 283                                                        "--$_" : "-$_" }
 284                                                split /\|/,$_)," $x\n";
 285                }
 286        }
 287        print $fd <<"";
 288\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
 289arbitrary identifier if you're tracking multiple SVN branches/repositories in
 290one git repository and want to keep them separate.  See git-svn(1) for more
 291information.
 292
 293        exit $exit;
 294}
 295
 296sub version {
 297        print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
 298        exit 0;
 299}
 300
 301sub do_git_init_db {
 302        unless (-d $ENV{GIT_DIR}) {
 303                my @init_db = ('init');
 304                push @init_db, "--template=$_template" if defined $_template;
 305                if (defined $_shared) {
 306                        if ($_shared =~ /[a-z]/) {
 307                                push @init_db, "--shared=$_shared";
 308                        } else {
 309                                push @init_db, "--shared";
 310                        }
 311                }
 312                command_noisy(@init_db);
 313                $_repository = Git->repository(Repository => ".git");
 314        }
 315        my $set;
 316        my $pfx = "svn-remote.$Git::SVN::default_repo_id";
 317        foreach my $i (keys %icv) {
 318                die "'$set' and '$i' cannot both be set\n" if $set;
 319                next unless defined $icv{$i};
 320                command_noisy('config', "$pfx.$i", $icv{$i});
 321                $set = $i;
 322        }
 323}
 324
 325sub init_subdir {
 326        my $repo_path = shift or return;
 327        mkpath([$repo_path]) unless -d $repo_path;
 328        chdir $repo_path or die "Couldn't chdir to $repo_path: $!\n";
 329        $ENV{GIT_DIR} = '.git';
 330        $_repository = Git->repository(Repository => $ENV{GIT_DIR});
 331}
 332
 333sub cmd_clone {
 334        my ($url, $path) = @_;
 335        if (!defined $path &&
 336            (defined $_trunk || defined $_branches || defined $_tags ||
 337             defined $_stdlayout) &&
 338            $url !~ m#^[a-z\+]+://#) {
 339                $path = $url;
 340        }
 341        $path = basename($url) if !defined $path || !length $path;
 342        cmd_init($url, $path);
 343        Git::SVN::fetch_all($Git::SVN::default_repo_id);
 344}
 345
 346sub cmd_init {
 347        if (defined $_stdlayout) {
 348                $_trunk = 'trunk' if (!defined $_trunk);
 349                $_tags = 'tags' if (!defined $_tags);
 350                $_branches = 'branches' if (!defined $_branches);
 351        }
 352        if (defined $_trunk || defined $_branches || defined $_tags) {
 353                return cmd_multi_init(@_);
 354        }
 355        my $url = shift or die "SVN repository location required ",
 356                               "as a command-line argument\n";
 357        init_subdir(@_);
 358        do_git_init_db();
 359
 360        Git::SVN->init($url);
 361}
 362
 363sub cmd_fetch {
 364        if (grep /^\d+=./, @_) {
 365                die "'<rev>=<commit>' fetch arguments are ",
 366                    "no longer supported.\n";
 367        }
 368        my ($remote) = @_;
 369        if (@_ > 1) {
 370                die "Usage: $0 fetch [--all] [svn-remote]\n";
 371        }
 372        $remote ||= $Git::SVN::default_repo_id;
 373        if ($_fetch_all) {
 374                cmd_multi_fetch();
 375        } else {
 376                Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
 377        }
 378}
 379
 380sub cmd_set_tree {
 381        my (@commits) = @_;
 382        if ($_stdin || !@commits) {
 383                print "Reading from stdin...\n";
 384                @commits = ();
 385                while (<STDIN>) {
 386                        if (/\b($sha1_short)\b/o) {
 387                                unshift @commits, $1;
 388                        }
 389                }
 390        }
 391        my @revs;
 392        foreach my $c (@commits) {
 393                my @tmp = command('rev-parse',$c);
 394                if (scalar @tmp == 1) {
 395                        push @revs, $tmp[0];
 396                } elsif (scalar @tmp > 1) {
 397                        push @revs, reverse(command('rev-list',@tmp));
 398                } else {
 399                        fatal "Failed to rev-parse $c";
 400                }
 401        }
 402        my $gs = Git::SVN->new;
 403        my ($r_last, $cmt_last) = $gs->last_rev_commit;
 404        $gs->fetch;
 405        if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
 406                fatal "There are new revisions that were fetched ",
 407                      "and need to be merged (or acknowledged) ",
 408                      "before committing.\nlast rev: $r_last\n",
 409                      " current: $gs->{last_rev}";
 410        }
 411        $gs->set_tree($_) foreach @revs;
 412        print "Done committing ",scalar @revs," revisions to SVN\n";
 413        unlink $gs->{index};
 414}
 415
 416sub cmd_dcommit {
 417        my $head = shift;
 418        git_cmd_try { command_oneline(qw/diff-index --quiet HEAD/) }
 419                'Cannot dcommit with a dirty index.  Commit your changes first, '
 420                . "or stash them with `git stash'.\n";
 421        $head ||= 'HEAD';
 422        my @refs;
 423        my ($url, $rev, $uuid, $gs) = working_head_info($head, \@refs);
 424        unless ($gs) {
 425                die "Unable to determine upstream SVN information from ",
 426                    "$head history.\nPerhaps the repository is empty.";
 427        }
 428        $url = defined $_commit_url ? $_commit_url : $gs->full_url;
 429        my $last_rev = $_revision if defined $_revision;
 430        if ($url) {
 431                print "Committing to $url ...\n";
 432        }
 433        my ($linear_refs, $parents) = linearize_history($gs, \@refs);
 434        if ($_no_rebase && scalar(@$linear_refs) > 1) {
 435                warn "Attempting to commit more than one change while ",
 436                     "--no-rebase is enabled.\n",
 437                     "If these changes depend on each other, re-running ",
 438                     "without --no-rebase may be required."
 439        }
 440        my $expect_url = $url;
 441        Git::SVN::remove_username($expect_url);
 442        while (1) {
 443                my $d = shift @$linear_refs or last;
 444                unless (defined $last_rev) {
 445                        (undef, $last_rev, undef) = cmt_metadata("$d~1");
 446                        unless (defined $last_rev) {
 447                                fatal "Unable to extract revision information ",
 448                                      "from commit $d~1";
 449                        }
 450                }
 451                if ($_dry_run) {
 452                        print "diff-tree $d~1 $d\n";
 453                } else {
 454                        my $cmt_rev;
 455                        my %ed_opts = ( r => $last_rev,
 456                                        log => get_commit_entry($d)->{log},
 457                                        ra => Git::SVN::Ra->new($url),
 458                                        config => SVN::Core::config_get_config(
 459                                                $Git::SVN::Ra::config_dir
 460                                        ),
 461                                        tree_a => "$d~1",
 462                                        tree_b => $d,
 463                                        editor_cb => sub {
 464                                               print "Committed r$_[0]\n";
 465                                               $cmt_rev = $_[0];
 466                                        },
 467                                        svn_path => '');
 468                        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
 469                                print "No changes\n$d~1 == $d\n";
 470                        } elsif ($parents->{$d} && @{$parents->{$d}}) {
 471                                $gs->{inject_parents_dcommit}->{$cmt_rev} =
 472                                                               $parents->{$d};
 473                        }
 474                        $_fetch_all ? $gs->fetch_all : $gs->fetch;
 475                        $last_rev = $cmt_rev;
 476                        next if $_no_rebase;
 477
 478                        # we always want to rebase against the current HEAD,
 479                        # not any head that was passed to us
 480                        my @diff = command('diff-tree', $d,
 481                                           $gs->refname, '--');
 482                        my @finish;
 483                        if (@diff) {
 484                                @finish = rebase_cmd();
 485                                print STDERR "W: $d and ", $gs->refname,
 486                                             " differ, using @finish:\n",
 487                                             join("\n", @diff), "\n";
 488                        } else {
 489                                print "No changes between current HEAD and ",
 490                                      $gs->refname,
 491                                      "\nResetting to the latest ",
 492                                      $gs->refname, "\n";
 493                                @finish = qw/reset --mixed/;
 494                        }
 495                        command_noisy(@finish, $gs->refname);
 496                        if (@diff) {
 497                                @refs = ();
 498                                my ($url_, $rev_, $uuid_, $gs_) =
 499                                              working_head_info($head, \@refs);
 500                                my ($linear_refs_, $parents_) =
 501                                              linearize_history($gs_, \@refs);
 502                                if (scalar(@$linear_refs) !=
 503                                    scalar(@$linear_refs_)) {
 504                                        fatal "# of revisions changed ",
 505                                          "\nbefore:\n",
 506                                          join("\n", @$linear_refs),
 507                                          "\n\nafter:\n",
 508                                          join("\n", @$linear_refs_), "\n",
 509                                          'If you are attempting to commit ',
 510                                          "merges, try running:\n\t",
 511                                          'git rebase --interactive',
 512                                          '--preserve-merges ',
 513                                          $gs->refname,
 514                                          "\nBefore dcommitting";
 515                                }
 516                                if ($url_ ne $expect_url) {
 517                                        fatal "URL mismatch after rebase: ",
 518                                              "$url_ != $expect_url";
 519                                }
 520                                if ($uuid_ ne $uuid) {
 521                                        fatal "uuid mismatch after rebase: ",
 522                                              "$uuid_ != $uuid";
 523                                }
 524                                # remap parents
 525                                my (%p, @l, $i);
 526                                for ($i = 0; $i < scalar @$linear_refs; $i++) {
 527                                        my $new = $linear_refs_->[$i] or next;
 528                                        $p{$new} =
 529                                                $parents->{$linear_refs->[$i]};
 530                                        push @l, $new;
 531                                }
 532                                $parents = \%p;
 533                                $linear_refs = \@l;
 534                        }
 535                }
 536        }
 537        unlink $gs->{index};
 538}
 539
 540sub cmd_find_rev {
 541        my $revision_or_hash = shift or die "SVN or git revision required ",
 542                                            "as a command-line argument\n";
 543        my $result;
 544        if ($revision_or_hash =~ /^r\d+$/) {
 545                my $head = shift;
 546                $head ||= 'HEAD';
 547                my @refs;
 548                my (undef, undef, $uuid, $gs) = working_head_info($head, \@refs);
 549                unless ($gs) {
 550                        die "Unable to determine upstream SVN information from ",
 551                            "$head history\n";
 552                }
 553                my $desired_revision = substr($revision_or_hash, 1);
 554                $result = $gs->rev_map_get($desired_revision, $uuid);
 555        } else {
 556                my (undef, $rev, undef) = cmt_metadata($revision_or_hash);
 557                $result = $rev;
 558        }
 559        print "$result\n" if $result;
 560}
 561
 562sub cmd_rebase {
 563        command_noisy(qw/update-index --refresh/);
 564        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 565        unless ($gs) {
 566                die "Unable to determine upstream SVN information from ",
 567                    "working tree history\n";
 568        }
 569        if ($_dry_run) {
 570                print "Remote Branch: " . $gs->refname . "\n";
 571                print "SVN URL: " . $url . "\n";
 572                return;
 573        }
 574        if (command(qw/diff-index HEAD --/)) {
 575                print STDERR "Cannot rebase with uncommited changes:\n";
 576                command_noisy('status');
 577                exit 1;
 578        }
 579        unless ($_local) {
 580                # rebase will checkout for us, so no need to do it explicitly
 581                $_no_checkout = 'true';
 582                $_fetch_all ? $gs->fetch_all : $gs->fetch;
 583        }
 584        command_noisy(rebase_cmd(), $gs->refname);
 585}
 586
 587sub cmd_show_ignore {
 588        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 589        $gs ||= Git::SVN->new;
 590        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 591        $gs->prop_walk($gs->{path}, $r, sub {
 592                my ($gs, $path, $props) = @_;
 593                print STDOUT "\n# $path\n";
 594                my $s = $props->{'svn:ignore'} or return;
 595                $s =~ s/[\r\n]+/\n/g;
 596                chomp $s;
 597                $s =~ s#^#$path#gm;
 598                print STDOUT "$s\n";
 599        });
 600}
 601
 602sub cmd_show_externals {
 603        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 604        $gs ||= Git::SVN->new;
 605        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 606        $gs->prop_walk($gs->{path}, $r, sub {
 607                my ($gs, $path, $props) = @_;
 608                print STDOUT "\n# $path\n";
 609                my $s = $props->{'svn:externals'} or return;
 610                $s =~ s/[\r\n]+/\n/g;
 611                chomp $s;
 612                $s =~ s#^#$path#gm;
 613                print STDOUT "$s\n";
 614        });
 615}
 616
 617sub cmd_create_ignore {
 618        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 619        $gs ||= Git::SVN->new;
 620        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 621        $gs->prop_walk($gs->{path}, $r, sub {
 622                my ($gs, $path, $props) = @_;
 623                # $path is of the form /path/to/dir/
 624                my $ignore = '.' . $path . '.gitignore';
 625                my $s = $props->{'svn:ignore'} or return;
 626                open(GITIGNORE, '>', $ignore)
 627                  or fatal("Failed to open `$ignore' for writing: $!");
 628                $s =~ s/[\r\n]+/\n/g;
 629                chomp $s;
 630                # Prefix all patterns so that the ignore doesn't apply
 631                # to sub-directories.
 632                $s =~ s#^#/#gm;
 633                print GITIGNORE "$s\n";
 634                close(GITIGNORE)
 635                  or fatal("Failed to close `$ignore': $!");
 636                command_noisy('add', '-f', $ignore);
 637        });
 638}
 639
 640sub canonicalize_path {
 641        my ($path) = @_;
 642        my $dot_slash_added = 0;
 643        if (substr($path, 0, 1) ne "/") {
 644                $path = "./" . $path;
 645                $dot_slash_added = 1;
 646        }
 647        # File::Spec->canonpath doesn't collapse x/../y into y (for a
 648        # good reason), so let's do this manually.
 649        $path =~ s#/+#/#g;
 650        $path =~ s#/\.(?:/|$)#/#g;
 651        $path =~ s#/[^/]+/\.\.##g;
 652        $path =~ s#/$##g;
 653        $path =~ s#^\./## if $dot_slash_added;
 654        $path =~ s#^/##;
 655        $path =~ s#^\.$##;
 656        return $path;
 657}
 658
 659# get_svnprops(PATH)
 660# ------------------
 661# Helper for cmd_propget and cmd_proplist below.
 662sub get_svnprops {
 663        my $path = shift;
 664        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 665        $gs ||= Git::SVN->new;
 666
 667        # prefix THE PATH by the sub-directory from which the user
 668        # invoked us.
 669        $path = $cmd_dir_prefix . $path;
 670        fatal("No such file or directory: $path") unless -e $path;
 671        my $is_dir = -d $path ? 1 : 0;
 672        $path = $gs->{path} . '/' . $path;
 673
 674        # canonicalize the path (otherwise libsvn will abort or fail to
 675        # find the file)
 676        $path = canonicalize_path($path);
 677
 678        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 679        my $props;
 680        if ($is_dir) {
 681                (undef, undef, $props) = $gs->ra->get_dir($path, $r);
 682        }
 683        else {
 684                (undef, $props) = $gs->ra->get_file($path, $r, undef);
 685        }
 686        return $props;
 687}
 688
 689# cmd_propget (PROP, PATH)
 690# ------------------------
 691# Print the SVN property PROP for PATH.
 692sub cmd_propget {
 693        my ($prop, $path) = @_;
 694        $path = '.' if not defined $path;
 695        usage(1) if not defined $prop;
 696        my $props = get_svnprops($path);
 697        if (not defined $props->{$prop}) {
 698                fatal("`$path' does not have a `$prop' SVN property.");
 699        }
 700        print $props->{$prop} . "\n";
 701}
 702
 703# cmd_proplist (PATH)
 704# -------------------
 705# Print the list of SVN properties for PATH.
 706sub cmd_proplist {
 707        my $path = shift;
 708        $path = '.' if not defined $path;
 709        my $props = get_svnprops($path);
 710        print "Properties on '$path':\n";
 711        foreach (sort keys %{$props}) {
 712                print "  $_\n";
 713        }
 714}
 715
 716sub cmd_multi_init {
 717        my $url = shift;
 718        unless (defined $_trunk || defined $_branches || defined $_tags) {
 719                usage(1);
 720        }
 721
 722        # there are currently some bugs that prevent multi-init/multi-fetch
 723        # setups from working well without this.
 724        $Git::SVN::_minimize_url = 1;
 725
 726        $_prefix = '' unless defined $_prefix;
 727        if (defined $url) {
 728                $url =~ s#/+$##;
 729                init_subdir(@_);
 730        }
 731        do_git_init_db();
 732        if (defined $_trunk) {
 733                my $trunk_ref = $_prefix . 'trunk';
 734                # try both old-style and new-style lookups:
 735                my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
 736                unless ($gs_trunk) {
 737                        my ($trunk_url, $trunk_path) =
 738                                              complete_svn_url($url, $_trunk);
 739                        $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
 740                                                   undef, $trunk_ref);
 741                }
 742        }
 743        return unless defined $_branches || defined $_tags;
 744        my $ra = $url ? Git::SVN::Ra->new($url) : undef;
 745        complete_url_ls_init($ra, $_branches, '--branches/-b', $_prefix);
 746        complete_url_ls_init($ra, $_tags, '--tags/-t', $_prefix . 'tags/');
 747}
 748
 749sub cmd_multi_fetch {
 750        my $remotes = Git::SVN::read_all_remotes();
 751        foreach my $repo_id (sort keys %$remotes) {
 752                if ($remotes->{$repo_id}->{url}) {
 753                        Git::SVN::fetch_all($repo_id, $remotes);
 754                }
 755        }
 756}
 757
 758# this command is special because it requires no metadata
 759sub cmd_commit_diff {
 760        my ($ta, $tb, $url) = @_;
 761        my $usage = "Usage: $0 commit-diff -r<revision> ".
 762                    "<tree-ish> <tree-ish> [<URL>]";
 763        fatal($usage) if (!defined $ta || !defined $tb);
 764        my $svn_path = '';
 765        if (!defined $url) {
 766                my $gs = eval { Git::SVN->new };
 767                if (!$gs) {
 768                        fatal("Needed URL or usable git-svn --id in ",
 769                              "the command-line\n", $usage);
 770                }
 771                $url = $gs->{url};
 772                $svn_path = $gs->{path};
 773        }
 774        unless (defined $_revision) {
 775                fatal("-r|--revision is a required argument\n", $usage);
 776        }
 777        if (defined $_message && defined $_file) {
 778                fatal("Both --message/-m and --file/-F specified ",
 779                      "for the commit message.\n",
 780                      "I have no idea what you mean");
 781        }
 782        if (defined $_file) {
 783                $_message = file_to_s($_file);
 784        } else {
 785                $_message ||= get_commit_entry($tb)->{log};
 786        }
 787        my $ra ||= Git::SVN::Ra->new($url);
 788        my $r = $_revision;
 789        if ($r eq 'HEAD') {
 790                $r = $ra->get_latest_revnum;
 791        } elsif ($r !~ /^\d+$/) {
 792                die "revision argument: $r not understood by git-svn\n";
 793        }
 794        my %ed_opts = ( r => $r,
 795                        log => $_message,
 796                        ra => $ra,
 797                        tree_a => $ta,
 798                        tree_b => $tb,
 799                        editor_cb => sub { print "Committed r$_[0]\n" },
 800                        svn_path => $svn_path );
 801        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
 802                print "No changes\n$ta == $tb\n";
 803        }
 804}
 805
 806sub cmd_info {
 807        my $path = canonicalize_path(defined($_[0]) ? $_[0] : ".");
 808        my $fullpath = canonicalize_path($cmd_dir_prefix . $path);
 809        if (exists $_[1]) {
 810                die "Too many arguments specified\n";
 811        }
 812
 813        my ($file_type, $diff_status) = find_file_type_and_diff_status($path);
 814
 815        if (!$file_type && !$diff_status) {
 816                print STDERR "$path:  (Not a versioned resource)\n\n";
 817                return;
 818        }
 819
 820        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 821        unless ($gs) {
 822                die "Unable to determine upstream SVN information from ",
 823                    "working tree history\n";
 824        }
 825
 826        # canonicalize_path() will return "" to make libsvn 1.5.x happy,
 827        $path = "." if $path eq "";
 828
 829        my $full_url = $url . ($fullpath eq "" ? "" : "/$fullpath");
 830
 831        if ($_url) {
 832                print $full_url, "\n";
 833                return;
 834        }
 835
 836        my $result = "Path: $path\n";
 837        $result .= "Name: " . basename($path) . "\n" if $file_type ne "dir";
 838        $result .= "URL: " . $full_url . "\n";
 839
 840        eval {
 841                my $repos_root = $gs->repos_root;
 842                Git::SVN::remove_username($repos_root);
 843                $result .= "Repository Root: $repos_root\n";
 844        };
 845        if ($@) {
 846                $result .= "Repository Root: (offline)\n";
 847        }
 848        $result .= "Repository UUID: $uuid\n" unless $diff_status eq "A";
 849        $result .= "Revision: " . ($diff_status eq "A" ? 0 : $rev) . "\n";
 850
 851        $result .= "Node Kind: " .
 852                   ($file_type eq "dir" ? "directory" : "file") . "\n";
 853
 854        my $schedule = $diff_status eq "A"
 855                       ? "add"
 856                       : ($diff_status eq "D" ? "delete" : "normal");
 857        $result .= "Schedule: $schedule\n";
 858
 859        if ($diff_status eq "A") {
 860                print $result, "\n";
 861                return;
 862        }
 863
 864        my ($lc_author, $lc_rev, $lc_date_utc);
 865        my @args = Git::SVN::Log::git_svn_log_cmd($rev, $rev, "--", $fullpath);
 866        my $log = command_output_pipe(@args);
 867        my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
 868        while (<$log>) {
 869                if (/^${esc_color}author (.+) <[^>]+> (\d+) ([\-\+]?\d+)$/o) {
 870                        $lc_author = $1;
 871                        $lc_date_utc = Git::SVN::Log::parse_git_date($2, $3);
 872                } elsif (/^${esc_color}    (git-svn-id:.+)$/o) {
 873                        (undef, $lc_rev, undef) = ::extract_metadata($1);
 874                }
 875        }
 876        close $log;
 877
 878        Git::SVN::Log::set_local_timezone();
 879
 880        $result .= "Last Changed Author: $lc_author\n";
 881        $result .= "Last Changed Rev: $lc_rev\n";
 882        $result .= "Last Changed Date: " .
 883                   Git::SVN::Log::format_svn_date($lc_date_utc) . "\n";
 884
 885        if ($file_type ne "dir") {
 886                my $text_last_updated_date =
 887                    ($diff_status eq "D" ? $lc_date_utc : (stat $path)[9]);
 888                $result .=
 889                    "Text Last Updated: " .
 890                    Git::SVN::Log::format_svn_date($text_last_updated_date) .
 891                    "\n";
 892                my $checksum;
 893                if ($diff_status eq "D") {
 894                        my ($fh, $ctx) =
 895                            command_output_pipe(qw(cat-file blob), "HEAD:$path");
 896                        if ($file_type eq "link") {
 897                                my $file_name = <$fh>;
 898                                $checksum = md5sum("link $file_name");
 899                        } else {
 900                                $checksum = md5sum($fh);
 901                        }
 902                        command_close_pipe($fh, $ctx);
 903                } elsif ($file_type eq "link") {
 904                        my $file_name =
 905                            command(qw(cat-file blob), "HEAD:$path");
 906                        $checksum =
 907                            md5sum("link " . $file_name);
 908                } else {
 909                        open FILE, "<", $path or die $!;
 910                        $checksum = md5sum(\*FILE);
 911                        close FILE or die $!;
 912                }
 913                $result .= "Checksum: " . $checksum . "\n";
 914        }
 915
 916        print $result, "\n";
 917}
 918
 919########################### utility functions #########################
 920
 921sub rebase_cmd {
 922        my @cmd = qw/rebase/;
 923        push @cmd, '-v' if $_verbose;
 924        push @cmd, qw/--merge/ if $_merge;
 925        push @cmd, "--strategy=$_strategy" if $_strategy;
 926        @cmd;
 927}
 928
 929sub post_fetch_checkout {
 930        return if $_no_checkout;
 931        my $gs = $Git::SVN::_head or return;
 932        return if verify_ref('refs/heads/master^0');
 933
 934        my $valid_head = verify_ref('HEAD^0');
 935        command_noisy(qw(update-ref refs/heads/master), $gs->refname);
 936        return if ($valid_head || !verify_ref('HEAD^0'));
 937
 938        return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
 939        my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
 940        return if -f $index;
 941
 942        return if command_oneline(qw/rev-parse --is-inside-work-tree/) eq 'false';
 943        return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
 944        command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
 945        print STDERR "Checked out HEAD:\n  ",
 946                     $gs->full_url, " r", $gs->last_rev, "\n";
 947}
 948
 949sub complete_svn_url {
 950        my ($url, $path) = @_;
 951        $path =~ s#/+$##;
 952        if ($path !~ m#^[a-z\+]+://#) {
 953                if (!defined $url || $url !~ m#^[a-z\+]+://#) {
 954                        fatal("E: '$path' is not a complete URL ",
 955                              "and a separate URL is not specified");
 956                }
 957                return ($url, $path);
 958        }
 959        return ($path, '');
 960}
 961
 962sub complete_url_ls_init {
 963        my ($ra, $repo_path, $switch, $pfx) = @_;
 964        unless ($repo_path) {
 965                print STDERR "W: $switch not specified\n";
 966                return;
 967        }
 968        $repo_path =~ s#/+$##;
 969        if ($repo_path =~ m#^[a-z\+]+://#) {
 970                $ra = Git::SVN::Ra->new($repo_path);
 971                $repo_path = '';
 972        } else {
 973                $repo_path =~ s#^/+##;
 974                unless ($ra) {
 975                        fatal("E: '$repo_path' is not a complete URL ",
 976                              "and a separate URL is not specified");
 977                }
 978        }
 979        my $url = $ra->{url};
 980        my $gs = Git::SVN->init($url, undef, undef, undef, 1);
 981        my $k = "svn-remote.$gs->{repo_id}.url";
 982        my $orig_url = eval { command_oneline(qw/config --get/, $k) };
 983        if ($orig_url && ($orig_url ne $gs->{url})) {
 984                die "$k already set: $orig_url\n",
 985                    "wanted to set to: $gs->{url}\n";
 986        }
 987        command_oneline('config', $k, $gs->{url}) unless $orig_url;
 988        my $remote_path = "$ra->{svn_path}/$repo_path";
 989        $remote_path =~ s#/+#/#g;
 990        $remote_path =~ s#^/##g;
 991        $remote_path .= "/*" if $remote_path !~ /\*/;
 992        my ($n) = ($switch =~ /^--(\w+)/);
 993        if (length $pfx && $pfx !~ m#/$#) {
 994                die "--prefix='$pfx' must have a trailing slash '/'\n";
 995        }
 996        command_noisy('config',
 997                      "svn-remote.$gs->{repo_id}.$n",
 998                      "$remote_path:refs/remotes/$pfx*" .
 999                        ('/*' x (($remote_path =~ tr/*/*/) - 1)) );
1000}
1001
1002sub verify_ref {
1003        my ($ref) = @_;
1004        eval { command_oneline([ 'rev-parse', '--verify', $ref ],
1005                               { STDERR => 0 }); };
1006}
1007
1008sub get_tree_from_treeish {
1009        my ($treeish) = @_;
1010        # $treeish can be a symbolic ref, too:
1011        my $type = command_oneline(qw/cat-file -t/, $treeish);
1012        my $expected;
1013        while ($type eq 'tag') {
1014                ($treeish, $type) = command(qw/cat-file tag/, $treeish);
1015        }
1016        if ($type eq 'commit') {
1017                $expected = (grep /^tree /, command(qw/cat-file commit/,
1018                                                    $treeish))[0];
1019                ($expected) = ($expected =~ /^tree ($sha1)$/o);
1020                die "Unable to get tree from $treeish\n" unless $expected;
1021        } elsif ($type eq 'tree') {
1022                $expected = $treeish;
1023        } else {
1024                die "$treeish is a $type, expected tree, tag or commit\n";
1025        }
1026        return $expected;
1027}
1028
1029sub get_commit_entry {
1030        my ($treeish) = shift;
1031        my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
1032        my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
1033        my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
1034        open my $log_fh, '>', $commit_editmsg or croak $!;
1035
1036        my $type = command_oneline(qw/cat-file -t/, $treeish);
1037        if ($type eq 'commit' || $type eq 'tag') {
1038                my ($msg_fh, $ctx) = command_output_pipe('cat-file',
1039                                                         $type, $treeish);
1040                my $in_msg = 0;
1041                my $author;
1042                my $saw_from = 0;
1043                my $msgbuf = "";
1044                while (<$msg_fh>) {
1045                        if (!$in_msg) {
1046                                $in_msg = 1 if (/^\s*$/);
1047                                $author = $1 if (/^author (.*>)/);
1048                        } elsif (/^git-svn-id: /) {
1049                                # skip this for now, we regenerate the
1050                                # correct one on re-fetch anyways
1051                                # TODO: set *:merge properties or like...
1052                        } else {
1053                                if (/^From:/ || /^Signed-off-by:/) {
1054                                        $saw_from = 1;
1055                                }
1056                                $msgbuf .= $_;
1057                        }
1058                }
1059                $msgbuf =~ s/\s+$//s;
1060                if ($Git::SVN::_add_author_from && defined($author)
1061                    && !$saw_from) {
1062                        $msgbuf .= "\n\nFrom: $author";
1063                }
1064                print $log_fh $msgbuf or croak $!;
1065                command_close_pipe($msg_fh, $ctx);
1066        }
1067        close $log_fh or croak $!;
1068
1069        if ($_edit || ($type eq 'tree')) {
1070                my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
1071                # TODO: strip out spaces, comments, like git-commit.sh
1072                system($editor, $commit_editmsg);
1073        }
1074        rename $commit_editmsg, $commit_msg or croak $!;
1075        open $log_fh, '<', $commit_msg or croak $!;
1076        { local $/; chomp($log_entry{log} = <$log_fh>); }
1077        close $log_fh or croak $!;
1078        unlink $commit_msg;
1079        \%log_entry;
1080}
1081
1082sub s_to_file {
1083        my ($str, $file, $mode) = @_;
1084        open my $fd,'>',$file or croak $!;
1085        print $fd $str,"\n" or croak $!;
1086        close $fd or croak $!;
1087        chmod ($mode &~ umask, $file) if (defined $mode);
1088}
1089
1090sub file_to_s {
1091        my $file = shift;
1092        open my $fd,'<',$file or croak "$!: file: $file\n";
1093        local $/;
1094        my $ret = <$fd>;
1095        close $fd or croak $!;
1096        $ret =~ s/\s*$//s;
1097        return $ret;
1098}
1099
1100# '<svn username> = real-name <email address>' mapping based on git-svnimport:
1101sub load_authors {
1102        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1103        my $log = $cmd eq 'log';
1104        while (<$authors>) {
1105                chomp;
1106                next unless /^(.+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
1107                my ($user, $name, $email) = ($1, $2, $3);
1108                if ($log) {
1109                        $Git::SVN::Log::rusers{"$name <$email>"} = $user;
1110                } else {
1111                        $users{$user} = [$name, $email];
1112                }
1113        }
1114        close $authors or croak $!;
1115}
1116
1117# convert GetOpt::Long specs for use by git-config
1118sub read_repo_config {
1119        return unless -d $ENV{GIT_DIR};
1120        my $opts = shift;
1121        my @config_only;
1122        foreach my $o (keys %$opts) {
1123                # if we have mixedCase and a long option-only, then
1124                # it's a config-only variable that we don't need for
1125                # the command-line.
1126                push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
1127                my $v = $opts->{$o};
1128                my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
1129                $key =~ s/-//g;
1130                my $arg = 'git-config';
1131                $arg .= ' --int' if ($o =~ /[:=]i$/);
1132                $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
1133                if (ref $v eq 'ARRAY') {
1134                        chomp(my @tmp = `$arg --get-all svn.$key`);
1135                        @$v = @tmp if @tmp;
1136                } else {
1137                        chomp(my $tmp = `$arg --get svn.$key`);
1138                        if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
1139                                $$v = $tmp;
1140                        }
1141                }
1142        }
1143        delete @$opts{@config_only} if @config_only;
1144}
1145
1146sub extract_metadata {
1147        my $id = shift or return (undef, undef, undef);
1148        my ($url, $rev, $uuid) = ($id =~ /^\s*git-svn-id:\s+(.*)\@(\d+)
1149                                                        \s([a-f\d\-]+)$/x);
1150        if (!defined $rev || !$uuid || !$url) {
1151                # some of the original repositories I made had
1152                # identifiers like this:
1153                ($rev, $uuid) = ($id =~/^\s*git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
1154        }
1155        return ($url, $rev, $uuid);
1156}
1157
1158sub cmt_metadata {
1159        return extract_metadata((grep(/^git-svn-id: /,
1160                command(qw/cat-file commit/, shift)))[-1]);
1161}
1162
1163sub working_head_info {
1164        my ($head, $refs) = @_;
1165        my @args = ('log', '--no-color', '--first-parent', '--pretty=medium');
1166        my ($fh, $ctx) = command_output_pipe(@args, $head);
1167        my $hash;
1168        my %max;
1169        while (<$fh>) {
1170                if ( m{^commit ($::sha1)$} ) {
1171                        unshift @$refs, $hash if $hash and $refs;
1172                        $hash = $1;
1173                        next;
1174                }
1175                next unless s{^\s*(git-svn-id:)}{$1};
1176                my ($url, $rev, $uuid) = extract_metadata($_);
1177                if (defined $url && defined $rev) {
1178                        next if $max{$url} and $max{$url} < $rev;
1179                        if (my $gs = Git::SVN->find_by_url($url)) {
1180                                my $c = $gs->rev_map_get($rev, $uuid);
1181                                if ($c && $c eq $hash) {
1182                                        close $fh; # break the pipe
1183                                        return ($url, $rev, $uuid, $gs);
1184                                } else {
1185                                        $max{$url} ||= $gs->rev_map_max;
1186                                }
1187                        }
1188                }
1189        }
1190        command_close_pipe($fh, $ctx);
1191        (undef, undef, undef, undef);
1192}
1193
1194sub read_commit_parents {
1195        my ($parents, $c) = @_;
1196        chomp(my $p = command_oneline(qw/rev-list --parents -1/, $c));
1197        $p =~ s/^($c)\s*// or die "rev-list --parents -1 $c failed!\n";
1198        @{$parents->{$c}} = split(/ /, $p);
1199}
1200
1201sub linearize_history {
1202        my ($gs, $refs) = @_;
1203        my %parents;
1204        foreach my $c (@$refs) {
1205                read_commit_parents(\%parents, $c);
1206        }
1207
1208        my @linear_refs;
1209        my %skip = ();
1210        my $last_svn_commit = $gs->last_commit;
1211        foreach my $c (reverse @$refs) {
1212                next if $c eq $last_svn_commit;
1213                last if $skip{$c};
1214
1215                unshift @linear_refs, $c;
1216                $skip{$c} = 1;
1217
1218                # we only want the first parent to diff against for linear
1219                # history, we save the rest to inject when we finalize the
1220                # svn commit
1221                my $fp_a = verify_ref("$c~1");
1222                my $fp_b = shift @{$parents{$c}} if $parents{$c};
1223                if (!$fp_a || !$fp_b) {
1224                        die "Commit $c\n",
1225                            "has no parent commit, and therefore ",
1226                            "nothing to diff against.\n",
1227                            "You should be working from a repository ",
1228                            "originally created by git-svn\n";
1229                }
1230                if ($fp_a ne $fp_b) {
1231                        die "$c~1 = $fp_a, however parsing commit $c ",
1232                            "revealed that:\n$c~1 = $fp_b\nBUG!\n";
1233                }
1234
1235                foreach my $p (@{$parents{$c}}) {
1236                        $skip{$p} = 1;
1237                }
1238        }
1239        (\@linear_refs, \%parents);
1240}
1241
1242sub find_file_type_and_diff_status {
1243        my ($path) = @_;
1244        return ('dir', '') if $path eq '';
1245
1246        my $diff_output =
1247            command_oneline(qw(diff --cached --name-status --), $path) || "";
1248        my $diff_status = (split(' ', $diff_output))[0] || "";
1249
1250        my $ls_tree = command_oneline(qw(ls-tree HEAD), $path) || "";
1251
1252        return (undef, undef) if !$diff_status && !$ls_tree;
1253
1254        if ($diff_status eq "A") {
1255                return ("link", $diff_status) if -l $path;
1256                return ("dir", $diff_status) if -d $path;
1257                return ("file", $diff_status);
1258        }
1259
1260        my $mode = (split(' ', $ls_tree))[0] || "";
1261
1262        return ("link", $diff_status) if $mode eq "120000";
1263        return ("dir", $diff_status) if $mode eq "040000";
1264        return ("file", $diff_status);
1265}
1266
1267sub md5sum {
1268        my $arg = shift;
1269        my $ref = ref $arg;
1270        my $md5 = Digest::MD5->new();
1271        if ($ref eq 'GLOB' || $ref eq 'IO::File' || $ref eq 'File::Temp') {
1272                $md5->addfile($arg) or croak $!;
1273        } elsif ($ref eq 'SCALAR') {
1274                $md5->add($$arg) or croak $!;
1275        } elsif (!$ref) {
1276                $md5->add($arg) or croak $!;
1277        } else {
1278                ::fatal "Can't provide MD5 hash for unknown ref type: '", $ref, "'";
1279        }
1280        return $md5->hexdigest();
1281}
1282
1283package Git::SVN;
1284use strict;
1285use warnings;
1286use Fcntl qw/:DEFAULT :seek/;
1287use constant rev_map_fmt => 'NH40';
1288use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
1289            $_repack $_repack_flags $_use_svm_props $_head
1290            $_use_svnsync_props $no_reuse_existing $_minimize_url
1291            $_use_log_author $_add_author_from/;
1292use Carp qw/croak/;
1293use File::Path qw/mkpath/;
1294use File::Copy qw/copy/;
1295use IPC::Open3;
1296
1297my ($_gc_nr, $_gc_period);
1298
1299# properties that we do not log:
1300my %SKIP_PROP;
1301BEGIN {
1302        %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
1303                                        svn:special svn:executable
1304                                        svn:entry:committed-rev
1305                                        svn:entry:last-author
1306                                        svn:entry:uuid
1307                                        svn:entry:committed-date/;
1308
1309        # some options are read globally, but can be overridden locally
1310        # per [svn-remote "..."] section.  Command-line options will *NOT*
1311        # override options set in an [svn-remote "..."] section
1312        no strict 'refs';
1313        for my $option (qw/follow_parent no_metadata use_svm_props
1314                           use_svnsync_props/) {
1315                my $key = $option;
1316                $key =~ tr/_//d;
1317                my $prop = "-$option";
1318                *$option = sub {
1319                        my ($self) = @_;
1320                        return $self->{$prop} if exists $self->{$prop};
1321                        my $k = "svn-remote.$self->{repo_id}.$key";
1322                        eval { command_oneline(qw/config --get/, $k) };
1323                        if ($@) {
1324                                $self->{$prop} = ${"Git::SVN::_$option"};
1325                        } else {
1326                                my $v = command_oneline(qw/config --bool/,$k);
1327                                $self->{$prop} = $v eq 'false' ? 0 : 1;
1328                        }
1329                        return $self->{$prop};
1330                }
1331        }
1332}
1333
1334
1335my (%LOCKFILES, %INDEX_FILES);
1336END {
1337        unlink keys %LOCKFILES if %LOCKFILES;
1338        unlink keys %INDEX_FILES if %INDEX_FILES;
1339}
1340
1341sub resolve_local_globs {
1342        my ($url, $fetch, $glob_spec) = @_;
1343        return unless defined $glob_spec;
1344        my $ref = $glob_spec->{ref};
1345        my $path = $glob_spec->{path};
1346        foreach (command(qw#for-each-ref --format=%(refname) refs/remotes#)) {
1347                next unless m#^refs/remotes/$ref->{regex}$#;
1348                my $p = $1;
1349                my $pathname = desanitize_refname($path->full_path($p));
1350                my $refname = desanitize_refname($ref->full_path($p));
1351                if (my $existing = $fetch->{$pathname}) {
1352                        if ($existing ne $refname) {
1353                                die "Refspec conflict:\n",
1354                                    "existing: refs/remotes/$existing\n",
1355                                    " globbed: refs/remotes/$refname\n";
1356                        }
1357                        my $u = (::cmt_metadata("refs/remotes/$refname"))[0];
1358                        $u =~ s!^\Q$url\E(/|$)!! or die
1359                          "refs/remotes/$refname: '$url' not found in '$u'\n";
1360                        if ($pathname ne $u) {
1361                                warn "W: Refspec glob conflict ",
1362                                     "(ref: refs/remotes/$refname):\n",
1363                                     "expected path: $pathname\n",
1364                                     "    real path: $u\n",
1365                                     "Continuing ahead with $u\n";
1366                                next;
1367                        }
1368                } else {
1369                        $fetch->{$pathname} = $refname;
1370                }
1371        }
1372}
1373
1374sub parse_revision_argument {
1375        my ($base, $head) = @_;
1376        if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
1377                return ($base, $head);
1378        }
1379        return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
1380        return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
1381        return ($head, $head) if ($::_revision eq 'HEAD');
1382        return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
1383        return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
1384        die "revision argument: $::_revision not understood by git-svn\n";
1385}
1386
1387sub fetch_all {
1388        my ($repo_id, $remotes) = @_;
1389        if (ref $repo_id) {
1390                my $gs = $repo_id;
1391                $repo_id = undef;
1392                $repo_id = $gs->{repo_id};
1393        }
1394        $remotes ||= read_all_remotes();
1395        my $remote = $remotes->{$repo_id} or
1396                     die "[svn-remote \"$repo_id\"] unknown\n";
1397        my $fetch = $remote->{fetch};
1398        my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
1399        my (@gs, @globs);
1400        my $ra = Git::SVN::Ra->new($url);
1401        my $uuid = $ra->get_uuid;
1402        my $head = $ra->get_latest_revnum;
1403        my $base = defined $fetch ? $head : 0;
1404
1405        # read the max revs for wildcard expansion (branches/*, tags/*)
1406        foreach my $t (qw/branches tags/) {
1407                defined $remote->{$t} or next;
1408                push @globs, $remote->{$t};
1409                my $max_rev = eval { tmp_config(qw/--int --get/,
1410                                         "svn-remote.$repo_id.${t}-maxRev") };
1411                if (defined $max_rev && ($max_rev < $base)) {
1412                        $base = $max_rev;
1413                } elsif (!defined $max_rev) {
1414                        $base = 0;
1415                }
1416        }
1417
1418        if ($fetch) {
1419                foreach my $p (sort keys %$fetch) {
1420                        my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
1421                        my $lr = $gs->rev_map_max;
1422                        if (defined $lr) {
1423                                $base = $lr if ($lr < $base);
1424                        }
1425                        push @gs, $gs;
1426                }
1427        }
1428
1429        ($base, $head) = parse_revision_argument($base, $head);
1430        $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
1431}
1432
1433sub read_all_remotes {
1434        my $r = {};
1435        my $use_svm_props = eval { command_oneline(qw/config --bool
1436            svn.useSvmProps/) };
1437        $use_svm_props = $use_svm_props eq 'true' if $use_svm_props;
1438        foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
1439                if (m!^(.+)\.fetch=\s*(.*)\s*:\s*(.+)\s*$!) {
1440                        my ($remote, $local_ref, $_remote_ref) = ($1, $2, $3);
1441                        die("svn-remote.$remote: remote ref '$_remote_ref' "
1442                            . "must start with 'refs/remotes/'\n")
1443                                unless $_remote_ref =~ m{^refs/remotes/(.+)};
1444                        my $remote_ref = $1;
1445                        $local_ref =~ s{^/}{};
1446                        $r->{$remote}->{fetch}->{$local_ref} = $remote_ref;
1447                        $r->{$remote}->{svm} = {} if $use_svm_props;
1448                } elsif (m!^(.+)\.usesvmprops=\s*(.*)\s*$!) {
1449                        $r->{$1}->{svm} = {};
1450                } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
1451                        $r->{$1}->{url} = $2;
1452                } elsif (m!^(.+)\.(branches|tags)=
1453                           (.*):refs/remotes/(.+)\s*$/!x) {
1454                        my ($p, $g) = ($3, $4);
1455                        my $rs = $r->{$1}->{$2} = {
1456                                          t => $2,
1457                                          remote => $1,
1458                                          path => Git::SVN::GlobSpec->new($p),
1459                                          ref => Git::SVN::GlobSpec->new($g) };
1460                        if (length($rs->{ref}->{right}) != 0) {
1461                                die "The '*' glob character must be the last ",
1462                                    "character of '$g'\n";
1463                        }
1464                }
1465        }
1466
1467        map {
1468                if (defined $r->{$_}->{svm}) {
1469                        my $svm;
1470                        eval {
1471                                my $section = "svn-remote.$_";
1472                                $svm = {
1473                                        source => tmp_config('--get',
1474                                            "$section.svm-source"),
1475                                        replace => tmp_config('--get',
1476                                            "$section.svm-replace"),
1477                                }
1478                        };
1479                        $r->{$_}->{svm} = $svm;
1480                }
1481        } keys %$r;
1482
1483        $r;
1484}
1485
1486sub init_vars {
1487        $_gc_nr = $_gc_period = 1000;
1488        if (defined $_repack || defined $_repack_flags) {
1489               warn "Repack options are obsolete; they have no effect.\n";
1490        }
1491}
1492
1493sub verify_remotes_sanity {
1494        return unless -d $ENV{GIT_DIR};
1495        my %seen;
1496        foreach (command(qw/config -l/)) {
1497                if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
1498                        if ($seen{$1}) {
1499                                die "Remote ref refs/remote/$1 is tracked by",
1500                                    "\n  \"$_\"\nand\n  \"$seen{$1}\"\n",
1501                                    "Please resolve this ambiguity in ",
1502                                    "your git configuration file before ",
1503                                    "continuing\n";
1504                        }
1505                        $seen{$1} = $_;
1506                }
1507        }
1508}
1509
1510sub find_existing_remote {
1511        my ($url, $remotes) = @_;
1512        return undef if $no_reuse_existing;
1513        my $existing;
1514        foreach my $repo_id (keys %$remotes) {
1515                my $u = $remotes->{$repo_id}->{url} or next;
1516                next if $u ne $url;
1517                $existing = $repo_id;
1518                last;
1519        }
1520        $existing;
1521}
1522
1523sub init_remote_config {
1524        my ($self, $url, $no_write) = @_;
1525        $url =~ s!/+$!!; # strip trailing slash
1526        my $r = read_all_remotes();
1527        my $existing = find_existing_remote($url, $r);
1528        if ($existing) {
1529                unless ($no_write) {
1530                        print STDERR "Using existing ",
1531                                     "[svn-remote \"$existing\"]\n";
1532                }
1533                $self->{repo_id} = $existing;
1534        } elsif ($_minimize_url) {
1535                my $min_url = Git::SVN::Ra->new($url)->minimize_url;
1536                $existing = find_existing_remote($min_url, $r);
1537                if ($existing) {
1538                        unless ($no_write) {
1539                                print STDERR "Using existing ",
1540                                             "[svn-remote \"$existing\"]\n";
1541                        }
1542                        $self->{repo_id} = $existing;
1543                }
1544                if ($min_url ne $url) {
1545                        unless ($no_write) {
1546                                print STDERR "Using higher level of URL: ",
1547                                             "$url => $min_url\n";
1548                        }
1549                        my $old_path = $self->{path};
1550                        $self->{path} = $url;
1551                        $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
1552                        if (length $old_path) {
1553                                $self->{path} .= "/$old_path";
1554                        }
1555                        $url = $min_url;
1556                }
1557        }
1558        my $orig_url;
1559        if (!$existing) {
1560                # verify that we aren't overwriting anything:
1561                $orig_url = eval {
1562                        command_oneline('config', '--get',
1563                                        "svn-remote.$self->{repo_id}.url")
1564                };
1565                if ($orig_url && ($orig_url ne $url)) {
1566                        die "svn-remote.$self->{repo_id}.url already set: ",
1567                            "$orig_url\nwanted to set to: $url\n";
1568                }
1569        }
1570        my ($xrepo_id, $xpath) = find_ref($self->refname);
1571        if (defined $xpath) {
1572                die "svn-remote.$xrepo_id.fetch already set to track ",
1573                    "$xpath:refs/remotes/", $self->refname, "\n";
1574        }
1575        unless ($no_write) {
1576                command_noisy('config',
1577                              "svn-remote.$self->{repo_id}.url", $url);
1578                $self->{path} =~ s{^/}{};
1579                command_noisy('config', '--add',
1580                              "svn-remote.$self->{repo_id}.fetch",
1581                              "$self->{path}:".$self->refname);
1582        }
1583        $self->{url} = $url;
1584}
1585
1586sub find_by_url { # repos_root and, path are optional
1587        my ($class, $full_url, $repos_root, $path) = @_;
1588
1589        return undef unless defined $full_url;
1590        remove_username($full_url);
1591        remove_username($repos_root) if defined $repos_root;
1592        my $remotes = read_all_remotes();
1593        if (defined $full_url && defined $repos_root && !defined $path) {
1594                $path = $full_url;
1595                $path =~ s#^\Q$repos_root\E(?:/|$)##;
1596        }
1597        foreach my $repo_id (keys %$remotes) {
1598                my $u = $remotes->{$repo_id}->{url} or next;
1599                remove_username($u);
1600                next if defined $repos_root && $repos_root ne $u;
1601
1602                my $fetch = $remotes->{$repo_id}->{fetch} || {};
1603                foreach (qw/branches tags/) {
1604                        resolve_local_globs($u, $fetch,
1605                                            $remotes->{$repo_id}->{$_});
1606                }
1607                my $p = $path;
1608                my $rwr = rewrite_root({repo_id => $repo_id});
1609                my $svm = $remotes->{$repo_id}->{svm}
1610                        if defined $remotes->{$repo_id}->{svm};
1611                unless (defined $p) {
1612                        $p = $full_url;
1613                        my $z = $u;
1614                        my $prefix = '';
1615                        if ($rwr) {
1616                                $z = $rwr;
1617                        } elsif (defined $svm) {
1618                                $z = $svm->{source};
1619                                $prefix = $svm->{replace};
1620                                $prefix =~ s#^\Q$u\E(?:/|$)##;
1621                                $prefix =~ s#/$##;
1622                        }
1623                        $p =~ s#^\Q$z\E(?:/|$)#$prefix# or next;
1624                }
1625                foreach my $f (keys %$fetch) {
1626                        next if $f ne $p;
1627                        return Git::SVN->new($fetch->{$f}, $repo_id, $f);
1628                }
1629        }
1630        undef;
1631}
1632
1633sub init {
1634        my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
1635        my $self = _new($class, $repo_id, $ref_id, $path);
1636        if (defined $url) {
1637                $self->init_remote_config($url, $no_write);
1638        }
1639        $self;
1640}
1641
1642sub find_ref {
1643        my ($ref_id) = @_;
1644        foreach (command(qw/config -l/)) {
1645                next unless m!^svn-remote\.(.+)\.fetch=
1646                              \s*(.*)\s*:\s*refs/remotes/(.+)\s*$!x;
1647                my ($repo_id, $path, $ref) = ($1, $2, $3);
1648                if ($ref eq $ref_id) {
1649                        $path = '' if ($path =~ m#^\./?#);
1650                        return ($repo_id, $path);
1651                }
1652        }
1653        (undef, undef, undef);
1654}
1655
1656sub new {
1657        my ($class, $ref_id, $repo_id, $path) = @_;
1658        if (defined $ref_id && !defined $repo_id && !defined $path) {
1659                ($repo_id, $path) = find_ref($ref_id);
1660                if (!defined $repo_id) {
1661                        die "Could not find a \"svn-remote.*.fetch\" key ",
1662                            "in the repository configuration matching: ",
1663                            "refs/remotes/$ref_id\n";
1664                }
1665        }
1666        my $self = _new($class, $repo_id, $ref_id, $path);
1667        if (!defined $self->{path} || !length $self->{path}) {
1668                my $fetch = command_oneline('config', '--get',
1669                                            "svn-remote.$repo_id.fetch",
1670                                            ":refs/remotes/$ref_id\$") or
1671                     die "Failed to read \"svn-remote.$repo_id.fetch\" ",
1672                         "\":refs/remotes/$ref_id\$\" in config\n";
1673                ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
1674        }
1675        $self->{url} = command_oneline('config', '--get',
1676                                       "svn-remote.$repo_id.url") or
1677                  die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
1678        $self->rebuild;
1679        $self;
1680}
1681
1682sub refname {
1683        my ($refname) = "refs/remotes/$_[0]->{ref_id}" ;
1684
1685        # It cannot end with a slash /, we'll throw up on this because
1686        # SVN can't have directories with a slash in their name, either:
1687        if ($refname =~ m{/$}) {
1688                die "ref: '$refname' ends with a trailing slash, this is ",
1689                    "not permitted by git nor Subversion\n";
1690        }
1691
1692        # It cannot have ASCII control character space, tilde ~, caret ^,
1693        # colon :, question-mark ?, asterisk *, space, or open bracket [
1694        # anywhere.
1695        #
1696        # Additionally, % must be escaped because it is used for escaping
1697        # and we want our escaped refname to be reversible
1698        $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
1699
1700        # no slash-separated component can begin with a dot .
1701        # /.* becomes /%2E*
1702        $refname =~ s{/\.}{/%2E}g;
1703
1704        # It cannot have two consecutive dots .. anywhere
1705        # .. becomes %2E%2E
1706        $refname =~ s{\.\.}{%2E%2E}g;
1707
1708        return $refname;
1709}
1710
1711sub desanitize_refname {
1712        my ($refname) = @_;
1713        $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
1714        return $refname;
1715}
1716
1717sub svm_uuid {
1718        my ($self) = @_;
1719        return $self->{svm}->{uuid} if $self->svm;
1720        $self->ra;
1721        unless ($self->{svm}) {
1722                die "SVM UUID not cached, and reading remotely failed\n";
1723        }
1724        $self->{svm}->{uuid};
1725}
1726
1727sub svm {
1728        my ($self) = @_;
1729        return $self->{svm} if $self->{svm};
1730        my $svm;
1731        # see if we have it in our config, first:
1732        eval {
1733                my $section = "svn-remote.$self->{repo_id}";
1734                $svm = {
1735                  source => tmp_config('--get', "$section.svm-source"),
1736                  uuid => tmp_config('--get', "$section.svm-uuid"),
1737                  replace => tmp_config('--get', "$section.svm-replace"),
1738                }
1739        };
1740        if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
1741                $self->{svm} = $svm;
1742        }
1743        $self->{svm};
1744}
1745
1746sub _set_svm_vars {
1747        my ($self, $ra) = @_;
1748        return $ra if $self->svm;
1749
1750        my @err = ( "useSvmProps set, but failed to read SVM properties\n",
1751                    "(svm:source, svm:uuid) ",
1752                    "from the following URLs:\n" );
1753        sub read_svm_props {
1754                my ($self, $ra, $path, $r) = @_;
1755                my $props = ($ra->get_dir($path, $r))[2];
1756                my $src = $props->{'svm:source'};
1757                my $uuid = $props->{'svm:uuid'};
1758                return undef if (!$src || !$uuid);
1759
1760                chomp($src, $uuid);
1761
1762                $uuid =~ m{^[0-9a-f\-]{30,}$}
1763                    or die "doesn't look right - svm:uuid is '$uuid'\n";
1764
1765                # the '!' is used to mark the repos_root!/relative/path
1766                $src =~ s{/?!/?}{/};
1767                $src =~ s{/+$}{}; # no trailing slashes please
1768                # username is of no interest
1769                $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
1770
1771                my $replace = $ra->{url};
1772                $replace .= "/$path" if length $path;
1773
1774                my $section = "svn-remote.$self->{repo_id}";
1775                tmp_config("$section.svm-source", $src);
1776                tmp_config("$section.svm-replace", $replace);
1777                tmp_config("$section.svm-uuid", $uuid);
1778                $self->{svm} = {
1779                        source => $src,
1780                        uuid => $uuid,
1781                        replace => $replace
1782                };
1783        }
1784
1785        my $r = $ra->get_latest_revnum;
1786        my $path = $self->{path};
1787        my %tried;
1788        while (length $path) {
1789                unless ($tried{"$self->{url}/$path"}) {
1790                        return $ra if $self->read_svm_props($ra, $path, $r);
1791                        $tried{"$self->{url}/$path"} = 1;
1792                }
1793                $path =~ s#/?[^/]+$##;
1794        }
1795        die "Path: '$path' should be ''\n" if $path ne '';
1796        return $ra if $self->read_svm_props($ra, $path, $r);
1797        $tried{"$self->{url}/$path"} = 1;
1798
1799        if ($ra->{repos_root} eq $self->{url}) {
1800                die @err, (map { "  $_\n" } keys %tried), "\n";
1801        }
1802
1803        # nope, make sure we're connected to the repository root:
1804        my $ok;
1805        my @tried_b;
1806        $path = $ra->{svn_path};
1807        $ra = Git::SVN::Ra->new($ra->{repos_root});
1808        while (length $path) {
1809                unless ($tried{"$ra->{url}/$path"}) {
1810                        $ok = $self->read_svm_props($ra, $path, $r);
1811                        last if $ok;
1812                        $tried{"$ra->{url}/$path"} = 1;
1813                }
1814                $path =~ s#/?[^/]+$##;
1815        }
1816        die "Path: '$path' should be ''\n" if $path ne '';
1817        $ok ||= $self->read_svm_props($ra, $path, $r);
1818        $tried{"$ra->{url}/$path"} = 1;
1819        if (!$ok) {
1820                die @err, (map { "  $_\n" } keys %tried), "\n";
1821        }
1822        Git::SVN::Ra->new($self->{url});
1823}
1824
1825sub svnsync {
1826        my ($self) = @_;
1827        return $self->{svnsync} if $self->{svnsync};
1828
1829        if ($self->no_metadata) {
1830                die "Can't have both 'noMetadata' and ",
1831                    "'useSvnsyncProps' options set!\n";
1832        }
1833        if ($self->rewrite_root) {
1834                die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
1835                    "options set!\n";
1836        }
1837
1838        my $svnsync;
1839        # see if we have it in our config, first:
1840        eval {
1841                my $section = "svn-remote.$self->{repo_id}";
1842
1843                my $url = tmp_config('--get', "$section.svnsync-url");
1844                ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
1845                   die "doesn't look right - svn:sync-from-url is '$url'\n";
1846
1847                my $uuid = tmp_config('--get', "$section.svnsync-uuid");
1848                ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}) or
1849                   die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
1850
1851                $svnsync = { url => $url, uuid => $uuid }
1852        };
1853        if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
1854                return $self->{svnsync} = $svnsync;
1855        }
1856
1857        my $err = "useSvnsyncProps set, but failed to read " .
1858                  "svnsync property: svn:sync-from-";
1859        my $rp = $self->ra->rev_proplist(0);
1860
1861        my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
1862        ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
1863                   die "doesn't look right - svn:sync-from-url is '$url'\n";
1864
1865        my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
1866        ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}) or
1867                   die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
1868
1869        my $section = "svn-remote.$self->{repo_id}";
1870        tmp_config('--add', "$section.svnsync-uuid", $uuid);
1871        tmp_config('--add', "$section.svnsync-url", $url);
1872        return $self->{svnsync} = { url => $url, uuid => $uuid };
1873}
1874
1875# this allows us to memoize our SVN::Ra UUID locally and avoid a
1876# remote lookup (useful for 'git svn log').
1877sub ra_uuid {
1878        my ($self) = @_;
1879        unless ($self->{ra_uuid}) {
1880                my $key = "svn-remote.$self->{repo_id}.uuid";
1881                my $uuid = eval { tmp_config('--get', $key) };
1882                if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/) {
1883                        $self->{ra_uuid} = $uuid;
1884                } else {
1885                        die "ra_uuid called without URL\n" unless $self->{url};
1886                        $self->{ra_uuid} = $self->ra->get_uuid;
1887                        tmp_config('--add', $key, $self->{ra_uuid});
1888                }
1889        }
1890        $self->{ra_uuid};
1891}
1892
1893sub _set_repos_root {
1894        my ($self, $repos_root) = @_;
1895        my $k = "svn-remote.$self->{repo_id}.reposRoot";
1896        $repos_root ||= $self->ra->{repos_root};
1897        tmp_config($k, $repos_root);
1898        $repos_root;
1899}
1900
1901sub repos_root {
1902        my ($self) = @_;
1903        my $k = "svn-remote.$self->{repo_id}.reposRoot";
1904        eval { tmp_config('--get', $k) } || $self->_set_repos_root;
1905}
1906
1907sub ra {
1908        my ($self) = shift;
1909        my $ra = Git::SVN::Ra->new($self->{url});
1910        $self->_set_repos_root($ra->{repos_root});
1911        if ($self->use_svm_props && !$self->{svm}) {
1912                if ($self->no_metadata) {
1913                        die "Can't have both 'noMetadata' and ",
1914                            "'useSvmProps' options set!\n";
1915                } elsif ($self->use_svnsync_props) {
1916                        die "Can't have both 'useSvnsyncProps' and ",
1917                            "'useSvmProps' options set!\n";
1918                }
1919                $ra = $self->_set_svm_vars($ra);
1920                $self->{-want_revprops} = 1;
1921        }
1922        $ra;
1923}
1924
1925sub rel_path {
1926        my ($self) = @_;
1927        my $repos_root = $self->ra->{repos_root};
1928        return $self->{path} if ($self->{url} eq $repos_root);
1929        my $url = $self->{url} .
1930                  (length $self->{path} ? "/$self->{path}" : $self->{path});
1931        $url =~ s!^\Q$repos_root\E(?:/+|$)!!g;
1932        $url;
1933}
1934
1935# prop_walk(PATH, REV, SUB)
1936# -------------------------
1937# Recursively traverse PATH at revision REV and invoke SUB for each
1938# directory that contains a SVN property.  SUB will be invoked as
1939# follows:  &SUB(gs, path, props);  where `gs' is this instance of
1940# Git::SVN, `path' the path to the directory where the properties
1941# `props' were found.  The `path' will be relative to point of checkout,
1942# that is, if url://repo/trunk is the current Git branch, and that
1943# directory contains a sub-directory `d', SUB will be invoked with `/d/'
1944# as `path' (note the trailing `/').
1945sub prop_walk {
1946        my ($self, $path, $rev, $sub) = @_;
1947
1948        $path =~ s#^/##;
1949        my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev);
1950        $path =~ s#^/*#/#g;
1951        my $p = $path;
1952        # Strip the irrelevant part of the path.
1953        $p =~ s#^/+\Q$self->{path}\E(/|$)#/#;
1954        # Ensure the path is terminated by a `/'.
1955        $p =~ s#/*$#/#;
1956
1957        # The properties contain all the internal SVN stuff nobody
1958        # (usually) cares about.
1959        my $interesting_props = 0;
1960        foreach (keys %{$props}) {
1961                # If it doesn't start with `svn:', it must be a
1962                # user-defined property.
1963                ++$interesting_props and next if $_ !~ /^svn:/;
1964                # FIXME: Fragile, if SVN adds new public properties,
1965                # this needs to be updated.
1966                ++$interesting_props if /^svn:(?:ignore|keywords|executable
1967                                                 |eol-style|mime-type
1968                                                 |externals|needs-lock)$/x;
1969        }
1970        &$sub($self, $p, $props) if $interesting_props;
1971
1972        foreach (sort keys %$dirent) {
1973                next if $dirent->{$_}->{kind} != $SVN::Node::dir;
1974                $self->prop_walk($self->{path} . $p . $_, $rev, $sub);
1975        }
1976}
1977
1978sub last_rev { ($_[0]->last_rev_commit)[0] }
1979sub last_commit { ($_[0]->last_rev_commit)[1] }
1980
1981# returns the newest SVN revision number and newest commit SHA1
1982sub last_rev_commit {
1983        my ($self) = @_;
1984        if (defined $self->{last_rev} && defined $self->{last_commit}) {
1985                return ($self->{last_rev}, $self->{last_commit});
1986        }
1987        my $c = ::verify_ref($self->refname.'^0');
1988        if ($c && !$self->use_svm_props && !$self->no_metadata) {
1989                my $rev = (::cmt_metadata($c))[1];
1990                if (defined $rev) {
1991                        ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
1992                        return ($rev, $c);
1993                }
1994        }
1995        my $map_path = $self->map_path;
1996        unless (-e $map_path) {
1997                ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
1998                return (undef, undef);
1999        }
2000        my ($rev, $commit) = $self->rev_map_max(1);
2001        ($self->{last_rev}, $self->{last_commit}) = ($rev, $commit);
2002        return ($rev, $commit);
2003}
2004
2005sub get_fetch_range {
2006        my ($self, $min, $max) = @_;
2007        $max ||= $self->ra->get_latest_revnum;
2008        $min ||= $self->rev_map_max;
2009        (++$min, $max);
2010}
2011
2012sub tmp_config {
2013        my (@args) = @_;
2014        my $old_def_config = "$ENV{GIT_DIR}/svn/config";
2015        my $config = "$ENV{GIT_DIR}/svn/.metadata";
2016        if (! -f $config && -f $old_def_config) {
2017                rename $old_def_config, $config or
2018                       die "Failed rename $old_def_config => $config: $!\n";
2019        }
2020        my $old_config = $ENV{GIT_CONFIG};
2021        $ENV{GIT_CONFIG} = $config;
2022        $@ = undef;
2023        my @ret = eval {
2024                unless (-f $config) {
2025                        mkfile($config);
2026                        open my $fh, '>', $config or
2027                            die "Can't open $config: $!\n";
2028                        print $fh "; This file is used internally by ",
2029                                  "git-svn\n" or die
2030                                  "Couldn't write to $config: $!\n";
2031                        print $fh "; You should not have to edit it\n" or
2032                              die "Couldn't write to $config: $!\n";
2033                        close $fh or die "Couldn't close $config: $!\n";
2034                }
2035                command('config', @args);
2036        };
2037        my $err = $@;
2038        if (defined $old_config) {
2039                $ENV{GIT_CONFIG} = $old_config;
2040        } else {
2041                delete $ENV{GIT_CONFIG};
2042        }
2043        die $err if $err;
2044        wantarray ? @ret : $ret[0];
2045}
2046
2047sub tmp_index_do {
2048        my ($self, $sub) = @_;
2049        my $old_index = $ENV{GIT_INDEX_FILE};
2050        $ENV{GIT_INDEX_FILE} = $self->{index};
2051        $@ = undef;
2052        my @ret = eval {
2053                my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
2054                mkpath([$dir]) unless -d $dir;
2055                &$sub;
2056        };
2057        my $err = $@;
2058        if (defined $old_index) {
2059                $ENV{GIT_INDEX_FILE} = $old_index;
2060        } else {
2061                delete $ENV{GIT_INDEX_FILE};
2062        }
2063        die $err if $err;
2064        wantarray ? @ret : $ret[0];
2065}
2066
2067sub assert_index_clean {
2068        my ($self, $treeish) = @_;
2069
2070        $self->tmp_index_do(sub {
2071                command_noisy('read-tree', $treeish) unless -e $self->{index};
2072                my $x = command_oneline('write-tree');
2073                my ($y) = (command(qw/cat-file commit/, $treeish) =~
2074                           /^tree ($::sha1)/mo);
2075                return if $y eq $x;
2076
2077                warn "Index mismatch: $y != $x\nrereading $treeish\n";
2078                unlink $self->{index} or die "unlink $self->{index}: $!\n";
2079                command_noisy('read-tree', $treeish);
2080                $x = command_oneline('write-tree');
2081                if ($y ne $x) {
2082                        ::fatal "trees ($treeish) $y != $x\n",
2083                                "Something is seriously wrong...";
2084                }
2085        });
2086}
2087
2088sub get_commit_parents {
2089        my ($self, $log_entry) = @_;
2090        my (%seen, @ret, @tmp);
2091        # legacy support for 'set-tree'; this is only used by set_tree_cb:
2092        if (my $ip = $self->{inject_parents}) {
2093                if (my $commit = delete $ip->{$log_entry->{revision}}) {
2094                        push @tmp, $commit;
2095                }
2096        }
2097        if (my $cur = ::verify_ref($self->refname.'^0')) {
2098                push @tmp, $cur;
2099        }
2100        if (my $ipd = $self->{inject_parents_dcommit}) {
2101                if (my $commit = delete $ipd->{$log_entry->{revision}}) {
2102                        push @tmp, @$commit;
2103                }
2104        }
2105        push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
2106        while (my $p = shift @tmp) {
2107                next if $seen{$p};
2108                $seen{$p} = 1;
2109                push @ret, $p;
2110                # MAXPARENT is defined to 16 in commit-tree.c:
2111                last if @ret >= 16;
2112        }
2113        if (@tmp) {
2114                die "r$log_entry->{revision}: No room for parents:\n\t",
2115                    join("\n\t", @tmp), "\n";
2116        }
2117        @ret;
2118}
2119
2120sub rewrite_root {
2121        my ($self) = @_;
2122        return $self->{-rewrite_root} if exists $self->{-rewrite_root};
2123        my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
2124        my $rwr = eval { command_oneline(qw/config --get/, $k) };
2125        if ($rwr) {
2126                $rwr =~ s#/+$##;
2127                if ($rwr !~ m#^[a-z\+]+://#) {
2128                        die "$rwr is not a valid URL (key: $k)\n";
2129                }
2130        }
2131        $self->{-rewrite_root} = $rwr;
2132}
2133
2134sub metadata_url {
2135        my ($self) = @_;
2136        ($self->rewrite_root || $self->{url}) .
2137           (length $self->{path} ? '/' . $self->{path} : '');
2138}
2139
2140sub full_url {
2141        my ($self) = @_;
2142        $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
2143}
2144
2145
2146sub set_commit_header_env {
2147        my ($log_entry) = @_;
2148        my %env;
2149        foreach my $ned (qw/NAME EMAIL DATE/) {
2150                foreach my $ac (qw/AUTHOR COMMITTER/) {
2151                        $env{"GIT_${ac}_${ned}"} = $ENV{"GIT_${ac}_${ned}"};
2152                }
2153        }
2154
2155        $ENV{GIT_AUTHOR_NAME} = $log_entry->{name};
2156        $ENV{GIT_AUTHOR_EMAIL} = $log_entry->{email};
2157        $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
2158
2159        $ENV{GIT_COMMITTER_NAME} = (defined $log_entry->{commit_name})
2160                                                ? $log_entry->{commit_name}
2161                                                : $log_entry->{name};
2162        $ENV{GIT_COMMITTER_EMAIL} = (defined $log_entry->{commit_email})
2163                                                ? $log_entry->{commit_email}
2164                                                : $log_entry->{email};
2165        \%env;
2166}
2167
2168sub restore_commit_header_env {
2169        my ($env) = @_;
2170        foreach my $ned (qw/NAME EMAIL DATE/) {
2171                foreach my $ac (qw/AUTHOR COMMITTER/) {
2172                        my $k = "GIT_${ac}_${ned}";
2173                        if (defined $env->{$k}) {
2174                                $ENV{$k} = $env->{$k};
2175                        } else {
2176                                delete $ENV{$k};
2177                        }
2178                }
2179        }
2180}
2181
2182sub gc {
2183        command_noisy('gc', '--auto');
2184};
2185
2186sub do_git_commit {
2187        my ($self, $log_entry) = @_;
2188        my $lr = $self->last_rev;
2189        if (defined $lr && $lr >= $log_entry->{revision}) {
2190                die "Last fetched revision of ", $self->refname,
2191                    " was r$lr, but we are about to fetch: ",
2192                    "r$log_entry->{revision}!\n";
2193        }
2194        if (my $c = $self->rev_map_get($log_entry->{revision})) {
2195                croak "$log_entry->{revision} = $c already exists! ",
2196                      "Why are we refetching it?\n";
2197        }
2198        my $old_env = set_commit_header_env($log_entry);
2199        my $tree = $log_entry->{tree};
2200        if (!defined $tree) {
2201                $tree = $self->tmp_index_do(sub {
2202                                            command_oneline('write-tree') });
2203        }
2204        die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
2205
2206        my @exec = ('git-commit-tree', $tree);
2207        foreach ($self->get_commit_parents($log_entry)) {
2208                push @exec, '-p', $_;
2209        }
2210        defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
2211                                                                   or croak $!;
2212        print $msg_fh $log_entry->{log} or croak $!;
2213        restore_commit_header_env($old_env);
2214        unless ($self->no_metadata) {
2215                print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
2216                              or croak $!;
2217        }
2218        $msg_fh->flush == 0 or croak $!;
2219        close $msg_fh or croak $!;
2220        chomp(my $commit = do { local $/; <$out_fh> });
2221        close $out_fh or croak $!;
2222        waitpid $pid, 0;
2223        croak $? if $?;
2224        if ($commit !~ /^$::sha1$/o) {
2225                die "Failed to commit, invalid sha1: $commit\n";
2226        }
2227
2228        $self->rev_map_set($log_entry->{revision}, $commit, 1);
2229
2230        $self->{last_rev} = $log_entry->{revision};
2231        $self->{last_commit} = $commit;
2232        print "r$log_entry->{revision}";
2233        if (defined $log_entry->{svm_revision}) {
2234                 print " (\@$log_entry->{svm_revision})";
2235                 $self->rev_map_set($log_entry->{svm_revision}, $commit,
2236                                   0, $self->svm_uuid);
2237        }
2238        print " = $commit ($self->{ref_id})\n";
2239        if (--$_gc_nr == 0) {
2240                $_gc_nr = $_gc_period;
2241                gc();
2242        }
2243        return $commit;
2244}
2245
2246sub match_paths {
2247        my ($self, $paths, $r) = @_;
2248        return 1 if $self->{path} eq '';
2249        if (my $path = $paths->{"/$self->{path}"}) {
2250                return ($path->{action} eq 'D') ? 0 : 1;
2251        }
2252        $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
2253        if (grep /$self->{path_regex}/, keys %$paths) {
2254                return 1;
2255        }
2256        my $c = '';
2257        foreach (split m#/#, $self->{path}) {
2258                $c .= "/$_";
2259                next unless ($paths->{$c} &&
2260                             ($paths->{$c}->{action} =~ /^[AR]$/));
2261                if ($self->ra->check_path($self->{path}, $r) ==
2262                    $SVN::Node::dir) {
2263                        return 1;
2264                }
2265        }
2266        return 0;
2267}
2268
2269sub find_parent_branch {
2270        my ($self, $paths, $rev) = @_;
2271        return undef unless $self->follow_parent;
2272        unless (defined $paths) {
2273                my $err_handler = $SVN::Error::handler;
2274                $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
2275                $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1, sub {
2276                                   $paths =
2277                                      Git::SVN::Ra::dup_changed_paths($_[0]) });
2278                $SVN::Error::handler = $err_handler;
2279        }
2280        return undef unless defined $paths;
2281
2282        # look for a parent from another branch:
2283        my @b_path_components = split m#/#, $self->rel_path;
2284        my @a_path_components;
2285        my $i;
2286        while (@b_path_components) {
2287                $i = $paths->{'/'.join('/', @b_path_components)};
2288                last if $i && defined $i->{copyfrom_path};
2289                unshift(@a_path_components, pop(@b_path_components));
2290        }
2291        return undef unless defined $i && defined $i->{copyfrom_path};
2292        my $branch_from = $i->{copyfrom_path};
2293        if (@a_path_components) {
2294                print STDERR "branch_from: $branch_from => ";
2295                $branch_from .= '/'.join('/', @a_path_components);
2296                print STDERR $branch_from, "\n";
2297        }
2298        my $r = $i->{copyfrom_rev};
2299        my $repos_root = $self->ra->{repos_root};
2300        my $url = $self->ra->{url};
2301        my $new_url = $repos_root . $branch_from;
2302        print STDERR  "Found possible branch point: ",
2303                      "$new_url => ", $self->full_url, ", $r\n";
2304        $branch_from =~ s#^/##;
2305        my $gs = Git::SVN->find_by_url($new_url, $repos_root, $branch_from);
2306        unless ($gs) {
2307                my $ref_id = $self->{ref_id};
2308                $ref_id =~ s/\@\d+$//;
2309                $ref_id .= "\@$r";
2310                # just grow a tail if we're not unique enough :x
2311                $ref_id .= '-' while find_ref($ref_id);
2312                print STDERR "Initializing parent: $ref_id\n";
2313                my ($u, $p, $repo_id) = ($new_url, '', $ref_id);
2314                if ($u =~ s#^\Q$url\E(/|$)##) {
2315                        $p = $u;
2316                        $u = $url;
2317                        $repo_id = $self->{repo_id};
2318                }
2319                $gs = Git::SVN->init($u, $p, $repo_id, $ref_id, 1);
2320        }
2321        my ($r0, $parent) = $gs->find_rev_before($r, 1);
2322        if (!defined $r0 || !defined $parent) {
2323                my ($base, $head) = parse_revision_argument(0, $r);
2324                if ($base <= $r) {
2325                        $gs->fetch($base, $r);
2326                }
2327                ($r0, $parent) = $gs->last_rev_commit;
2328        }
2329        if (defined $r0 && defined $parent) {
2330                print STDERR "Found branch parent: ($self->{ref_id}) $parent\n";
2331                my $ed;
2332                if ($self->ra->can_do_switch) {
2333                        $self->assert_index_clean($parent);
2334                        print STDERR "Following parent with do_switch\n";
2335                        # do_switch works with svn/trunk >= r22312, but that
2336                        # is not included with SVN 1.4.3 (the latest version
2337                        # at the moment), so we can't rely on it
2338                        $self->{last_commit} = $parent;
2339                        $ed = SVN::Git::Fetcher->new($self);
2340                        $gs->ra->gs_do_switch($r0, $rev, $gs,
2341                                              $self->full_url, $ed)
2342                          or die "SVN connection failed somewhere...\n";
2343                } elsif ($self->ra->trees_match($new_url, $r0,
2344                                                $self->full_url, $rev)) {
2345                        print STDERR "Trees match:\n",
2346                                     "  $new_url\@$r0\n",
2347                                     "  ${\$self->full_url}\@$rev\n",
2348                                     "Following parent with no changes\n";
2349                        $self->tmp_index_do(sub {
2350                            command_noisy('read-tree', $parent);
2351                        });
2352                        $self->{last_commit} = $parent;
2353                } else {
2354                        print STDERR "Following parent with do_update\n";
2355                        $ed = SVN::Git::Fetcher->new($self);
2356                        $self->ra->gs_do_update($rev, $rev, $self, $ed)
2357                          or die "SVN connection failed somewhere...\n";
2358                }
2359                print STDERR "Successfully followed parent\n";
2360                return $self->make_log_entry($rev, [$parent], $ed);
2361        }
2362        return undef;
2363}
2364
2365sub do_fetch {
2366        my ($self, $paths, $rev) = @_;
2367        my $ed;
2368        my ($last_rev, @parents);
2369        if (my $lc = $self->last_commit) {
2370                # we can have a branch that was deleted, then re-added
2371                # under the same name but copied from another path, in
2372                # which case we'll have multiple parents (we don't
2373                # want to break the original ref, nor lose copypath info):
2374                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2375                        push @{$log_entry->{parents}}, $lc;
2376                        return $log_entry;
2377                }
2378                $ed = SVN::Git::Fetcher->new($self);
2379                $last_rev = $self->{last_rev};
2380                $ed->{c} = $lc;
2381                @parents = ($lc);
2382        } else {
2383                $last_rev = $rev;
2384                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2385                        return $log_entry;
2386                }
2387                $ed = SVN::Git::Fetcher->new($self);
2388        }
2389        unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
2390                die "SVN connection failed somewhere...\n";
2391        }
2392        $self->make_log_entry($rev, \@parents, $ed);
2393}
2394
2395sub get_untracked {
2396        my ($self, $ed) = @_;
2397        my @out;
2398        my $h = $ed->{empty};
2399        foreach (sort keys %$h) {
2400                my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
2401                push @out, "  $act: " . uri_encode($_);
2402                warn "W: $act: $_\n";
2403        }
2404        foreach my $t (qw/dir_prop file_prop/) {
2405                $h = $ed->{$t} or next;
2406                foreach my $path (sort keys %$h) {
2407                        my $ppath = $path eq '' ? '.' : $path;
2408                        foreach my $prop (sort keys %{$h->{$path}}) {
2409                                next if $SKIP_PROP{$prop};
2410                                my $v = $h->{$path}->{$prop};
2411                                my $t_ppath_prop = "$t: " .
2412                                                    uri_encode($ppath) . ' ' .
2413                                                    uri_encode($prop);
2414                                if (defined $v) {
2415                                        push @out, "  +$t_ppath_prop " .
2416                                                   uri_encode($v);
2417                                } else {
2418                                        push @out, "  -$t_ppath_prop";
2419                                }
2420                        }
2421                }
2422        }
2423        foreach my $t (qw/absent_file absent_directory/) {
2424                $h = $ed->{$t} or next;
2425                foreach my $parent (sort keys %$h) {
2426                        foreach my $path (sort @{$h->{$parent}}) {
2427                                push @out, "  $t: " .
2428                                           uri_encode("$parent/$path");
2429                                warn "W: $t: $parent/$path ",
2430                                     "Insufficient permissions?\n";
2431                        }
2432                }
2433        }
2434        \@out;
2435}
2436
2437sub parse_svn_date {
2438        my $date = shift || return '+0000 1970-01-01 00:00:00';
2439        my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2440                                            (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x) or
2441                                         croak "Unable to parse date: $date\n";
2442        "+0000 $Y-$m-$d $H:$M:$S";
2443}
2444
2445sub check_author {
2446        my ($author) = @_;
2447        if (!defined $author || length $author == 0) {
2448                $author = '(no author)';
2449        } elsif (defined $::_authors && ! defined $::users{$author}) {
2450                die "Author: $author not defined in $::_authors file\n";
2451        }
2452        $author;
2453}
2454
2455sub make_log_entry {
2456        my ($self, $rev, $parents, $ed) = @_;
2457        my $untracked = $self->get_untracked($ed);
2458
2459        open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
2460        print $un "r$rev\n" or croak $!;
2461        print $un $_, "\n" foreach @$untracked;
2462        my %log_entry = ( parents => $parents || [], revision => $rev,
2463                          log => '');
2464
2465        my $headrev;
2466        my $logged = delete $self->{logged_rev_props};
2467        if (!$logged || $self->{-want_revprops}) {
2468                my $rp = $self->ra->rev_proplist($rev);
2469                foreach (sort keys %$rp) {
2470                        my $v = $rp->{$_};
2471                        if (/^svn:(author|date|log)$/) {
2472                                $log_entry{$1} = $v;
2473                        } elsif ($_ eq 'svm:headrev') {
2474                                $headrev = $v;
2475                        } else {
2476                                print $un "  rev_prop: ", uri_encode($_), ' ',
2477                                          uri_encode($v), "\n";
2478                        }
2479                }
2480        } else {
2481                map { $log_entry{$_} = $logged->{$_} } keys %$logged;
2482        }
2483        close $un or croak $!;
2484
2485        $log_entry{date} = parse_svn_date($log_entry{date});
2486        $log_entry{log} .= "\n";
2487        my $author = $log_entry{author} = check_author($log_entry{author});
2488        my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
2489                                                       : ($author, undef);
2490
2491        my ($commit_name, $commit_email) = ($name, $email);
2492        if ($_use_log_author) {
2493                my $name_field;
2494                if ($log_entry{log} =~ /From:\s+(.*\S)\s*\n/i) {
2495                        $name_field = $1;
2496                } elsif ($log_entry{log} =~ /Signed-off-by:\s+(.*\S)\s*\n/i) {
2497                        $name_field = $1;
2498                }
2499                if (!defined $name_field) {
2500                        if (!defined $email) {
2501                                $email = $name;
2502                        }
2503                } elsif ($name_field =~ /(.*?)\s+<(.*)>/) {
2504                        ($name, $email) = ($1, $2);
2505                } elsif ($name_field =~ /(.*)@/) {
2506                        ($name, $email) = ($1, $name_field);
2507                } else {
2508                        ($name, $email) = ($name_field, $name_field);
2509                }
2510        }
2511        if (defined $headrev && $self->use_svm_props) {
2512                if ($self->rewrite_root) {
2513                        die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
2514                            "options set!\n";
2515                }
2516                my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$};
2517                # we don't want "SVM: initializing mirror for junk" ...
2518                return undef if $r == 0;
2519                my $svm = $self->svm;
2520                if ($uuid ne $svm->{uuid}) {
2521                        die "UUID mismatch on SVM path:\n",
2522                            "expected: $svm->{uuid}\n",
2523                            "     got: $uuid\n";
2524                }
2525                my $full_url = $self->full_url;
2526                $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
2527                             die "Failed to replace '$svm->{replace}' with ",
2528                                 "'$svm->{source}' in $full_url\n";
2529                # throw away username for storing in records
2530                remove_username($full_url);
2531                $log_entry{metadata} = "$full_url\@$r $uuid";
2532                $log_entry{svm_revision} = $r;
2533                $email ||= "$author\@$uuid";
2534                $commit_email ||= "$author\@$uuid";
2535        } elsif ($self->use_svnsync_props) {
2536                my $full_url = $self->svnsync->{url};
2537                $full_url .= "/$self->{path}" if length $self->{path};
2538                remove_username($full_url);
2539                my $uuid = $self->svnsync->{uuid};
2540                $log_entry{metadata} = "$full_url\@$rev $uuid";
2541                $email ||= "$author\@$uuid";
2542                $commit_email ||= "$author\@$uuid";
2543        } else {
2544                my $url = $self->metadata_url;
2545                remove_username($url);
2546                $log_entry{metadata} = "$url\@$rev " .
2547                                       $self->ra->get_uuid;
2548                $email ||= "$author\@" . $self->ra->get_uuid;
2549                $commit_email ||= "$author\@" . $self->ra->get_uuid;
2550        }
2551        $log_entry{name} = $name;
2552        $log_entry{email} = $email;
2553        $log_entry{commit_name} = $commit_name;
2554        $log_entry{commit_email} = $commit_email;
2555        \%log_entry;
2556}
2557
2558sub fetch {
2559        my ($self, $min_rev, $max_rev, @parents) = @_;
2560        my ($last_rev, $last_commit) = $self->last_rev_commit;
2561        my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
2562        $self->ra->gs_fetch_loop_common($base, $head, [$self]);
2563}
2564
2565sub set_tree_cb {
2566        my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
2567        $self->{inject_parents} = { $rev => $tree };
2568        $self->fetch(undef, undef);
2569}
2570
2571sub set_tree {
2572        my ($self, $tree) = (shift, shift);
2573        my $log_entry = ::get_commit_entry($tree);
2574        unless ($self->{last_rev}) {
2575                fatal("Must have an existing revision to commit");
2576        }
2577        my %ed_opts = ( r => $self->{last_rev},
2578                        log => $log_entry->{log},
2579                        ra => $self->ra,
2580                        tree_a => $self->{last_commit},
2581                        tree_b => $tree,
2582                        editor_cb => sub {
2583                               $self->set_tree_cb($log_entry, $tree, @_) },
2584                        svn_path => $self->{path} );
2585        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
2586                print "No changes\nr$self->{last_rev} = $tree\n";
2587        }
2588}
2589
2590sub rebuild_from_rev_db {
2591        my ($self, $path) = @_;
2592        my $r = -1;
2593        open my $fh, '<', $path or croak "open: $!";
2594        binmode $fh or croak "binmode: $!";
2595        while (<$fh>) {
2596                length($_) == 41 or croak "inconsistent size in ($_) != 41";
2597                chomp($_);
2598                ++$r;
2599                next if $_ eq ('0' x 40);
2600                $self->rev_map_set($r, $_);
2601                print "r$r = $_\n";
2602        }
2603        close $fh or croak "close: $!";
2604        unlink $path or croak "unlink: $!";
2605}
2606
2607sub rebuild {
2608        my ($self) = @_;
2609        my $map_path = $self->map_path;
2610        return if (-e $map_path && ! -z $map_path);
2611        return unless ::verify_ref($self->refname.'^0');
2612        if ($self->use_svm_props || $self->no_metadata) {
2613                my $rev_db = $self->rev_db_path;
2614                $self->rebuild_from_rev_db($rev_db);
2615                if ($self->use_svm_props) {
2616                        my $svm_rev_db = $self->rev_db_path($self->svm_uuid);
2617                        $self->rebuild_from_rev_db($svm_rev_db);
2618                }
2619                $self->unlink_rev_db_symlink;
2620                return;
2621        }
2622        print "Rebuilding $map_path ...\n";
2623        my ($log, $ctx) =
2624            command_output_pipe(qw/rev-list --pretty=raw --no-color --reverse/,
2625                                $self->refname, '--');
2626        my $metadata_url = $self->metadata_url;
2627        remove_username($metadata_url);
2628        my $svn_uuid = $self->ra_uuid;
2629        my $c;
2630        while (<$log>) {
2631                if ( m{^commit ($::sha1)$} ) {
2632                        $c = $1;
2633                        next;
2634                }
2635                next unless s{^\s*(git-svn-id:)}{$1};
2636                my ($url, $rev, $uuid) = ::extract_metadata($_);
2637                remove_username($url);
2638
2639                # ignore merges (from set-tree)
2640                next if (!defined $rev || !$uuid);
2641
2642                # if we merged or otherwise started elsewhere, this is
2643                # how we break out of it
2644                if (($uuid ne $svn_uuid) ||
2645                    ($metadata_url && $url && ($url ne $metadata_url))) {
2646                        next;
2647                }
2648
2649                $self->rev_map_set($rev, $c);
2650                print "r$rev = $c\n";
2651        }
2652        command_close_pipe($log, $ctx);
2653        print "Done rebuilding $map_path\n";
2654        my $rev_db_path = $self->rev_db_path;
2655        if (-f $self->rev_db_path) {
2656                unlink $self->rev_db_path or croak "unlink: $!";
2657        }
2658        $self->unlink_rev_db_symlink;
2659}
2660
2661# rev_map:
2662# Tie::File seems to be prone to offset errors if revisions get sparse,
2663# it's not that fast, either.  Tie::File is also not in Perl 5.6.  So
2664# one of my favorite modules is out :<  Next up would be one of the DBM
2665# modules, but I'm not sure which is most portable...
2666#
2667# This is the replacement for the rev_db format, which was too big
2668# and inefficient for large repositories with a lot of sparse history
2669# (mainly tags)
2670#
2671# The format is this:
2672#   - 24 bytes for every record,
2673#     * 4 bytes for the integer representing an SVN revision number
2674#     * 20 bytes representing the sha1 of a git commit
2675#   - No empty padding records like the old format
2676#     (except the last record, which can be overwritten)
2677#   - new records are written append-only since SVN revision numbers
2678#     increase monotonically
2679#   - lookups on SVN revision number are done via a binary search
2680#   - Piping the file to xxd -c24 is a good way of dumping it for
2681#     viewing or editing (piped back through xxd -r), should the need
2682#     ever arise.
2683#   - The last record can be padding revision with an all-zero sha1
2684#     This is used to optimize fetch performance when using multiple
2685#     "fetch" directives in .git/config
2686#
2687# These files are disposable unless noMetadata or useSvmProps is set
2688
2689sub _rev_map_set {
2690        my ($fh, $rev, $commit) = @_;
2691
2692        binmode $fh or croak "binmode: $!";
2693        my $size = (stat($fh))[7];
2694        ($size % 24) == 0 or croak "inconsistent size: $size";
2695
2696        my $wr_offset = 0;
2697        if ($size > 0) {
2698                sysseek($fh, -24, SEEK_END) or croak "seek: $!";
2699                my $read = sysread($fh, my $buf, 24) or croak "read: $!";
2700                $read == 24 or croak "read only $read bytes (!= 24)";
2701                my ($last_rev, $last_commit) = unpack(rev_map_fmt, $buf);
2702                if ($last_commit eq ('0' x40)) {
2703                        if ($size >= 48) {
2704                                sysseek($fh, -48, SEEK_END) or croak "seek: $!";
2705                                $read = sysread($fh, $buf, 24) or
2706                                    croak "read: $!";
2707                                $read == 24 or
2708                                    croak "read only $read bytes (!= 24)";
2709                                ($last_rev, $last_commit) =
2710                                    unpack(rev_map_fmt, $buf);
2711                                if ($last_commit eq ('0' x40)) {
2712                                        croak "inconsistent .rev_map\n";
2713                                }
2714                        }
2715                        if ($last_rev >= $rev) {
2716                                croak "last_rev is higher!: $last_rev >= $rev";
2717                        }
2718                        $wr_offset = -24;
2719                }
2720        }
2721        sysseek($fh, $wr_offset, SEEK_END) or croak "seek: $!";
2722        syswrite($fh, pack(rev_map_fmt, $rev, $commit), 24) == 24 or
2723          croak "write: $!";
2724}
2725
2726sub mkfile {
2727        my ($path) = @_;
2728        unless (-e $path) {
2729                my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
2730                mkpath([$dir]) unless -d $dir;
2731                open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
2732                close $fh or die "Couldn't close (create) $path: $!\n";
2733        }
2734}
2735
2736sub rev_map_set {
2737        my ($self, $rev, $commit, $update_ref, $uuid) = @_;
2738        length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
2739        my $db = $self->map_path($uuid);
2740        my $db_lock = "$db.lock";
2741        my $sig;
2742        if ($update_ref) {
2743                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
2744                            $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
2745        }
2746        mkfile($db);
2747
2748        $LOCKFILES{$db_lock} = 1;
2749        my $sync;
2750        # both of these options make our .rev_db file very, very important
2751        # and we can't afford to lose it because rebuild() won't work
2752        if ($self->use_svm_props || $self->no_metadata) {
2753                $sync = 1;
2754                copy($db, $db_lock) or die "rev_map_set(@_): ",
2755                                           "Failed to copy: ",
2756                                           "$db => $db_lock ($!)\n";
2757        } else {
2758                rename $db, $db_lock or die "rev_map_set(@_): ",
2759                                            "Failed to rename: ",
2760                                            "$db => $db_lock ($!)\n";
2761        }
2762
2763        sysopen(my $fh, $db_lock, O_RDWR | O_CREAT)
2764             or croak "Couldn't open $db_lock: $!\n";
2765        _rev_map_set($fh, $rev, $commit);
2766        if ($sync) {
2767                $fh->flush or die "Couldn't flush $db_lock: $!\n";
2768                $fh->sync or die "Couldn't sync $db_lock: $!\n";
2769        }
2770        close $fh or croak $!;
2771        if ($update_ref) {
2772                $_head = $self;
2773                command_noisy('update-ref', '-m', "r$rev",
2774                              $self->refname, $commit);
2775        }
2776        rename $db_lock, $db or die "rev_map_set(@_): ", "Failed to rename: ",
2777                                    "$db_lock => $db ($!)\n";
2778        delete $LOCKFILES{$db_lock};
2779        if ($update_ref) {
2780                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
2781                            $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
2782                kill $sig, $$ if defined $sig;
2783        }
2784}
2785
2786# If want_commit, this will return an array of (rev, commit) where
2787# commit _must_ be a valid commit in the archive.
2788# Otherwise, it'll return the max revision (whether or not the
2789# commit is valid or just a 0x40 placeholder).
2790sub rev_map_max {
2791        my ($self, $want_commit) = @_;
2792        $self->rebuild;
2793        my $map_path = $self->map_path;
2794        stat $map_path or return $want_commit ? (0, undef) : 0;
2795        sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
2796        binmode $fh or croak "binmode: $!";
2797        my $size = (stat($fh))[7];
2798        ($size % 24) == 0 or croak "inconsistent size: $size";
2799
2800        if ($size == 0) {
2801                close $fh or croak "close: $!";
2802                return $want_commit ? (0, undef) : 0;
2803        }
2804
2805        sysseek($fh, -24, SEEK_END) or croak "seek: $!";
2806        sysread($fh, my $buf, 24) == 24 or croak "read: $!";
2807        my ($r, $c) = unpack(rev_map_fmt, $buf);
2808        if ($want_commit && $c eq ('0' x40)) {
2809                if ($size < 48) {
2810                        return $want_commit ? (0, undef) : 0;
2811                }
2812                sysseek($fh, -48, SEEK_END) or croak "seek: $!";
2813                sysread($fh, $buf, 24) == 24 or croak "read: $!";
2814                ($r, $c) = unpack(rev_map_fmt, $buf);
2815                if ($c eq ('0'x40)) {
2816                        croak "Penultimate record is all-zeroes in $map_path";
2817                }
2818        }
2819        close $fh or croak "close: $!";
2820        $want_commit ? ($r, $c) : $r;
2821}
2822
2823sub rev_map_get {
2824        my ($self, $rev, $uuid) = @_;
2825        my $map_path = $self->map_path($uuid);
2826        return undef unless -e $map_path;
2827
2828        sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
2829        binmode $fh or croak "binmode: $!";
2830        my $size = (stat($fh))[7];
2831        ($size % 24) == 0 or croak "inconsistent size: $size";
2832
2833        if ($size == 0) {
2834                close $fh or croak "close: $fh";
2835                return undef;
2836        }
2837
2838        my ($l, $u) = (0, $size - 24);
2839        my ($r, $c, $buf);
2840
2841        while ($l <= $u) {
2842                my $i = int(($l/24 + $u/24) / 2) * 24;
2843                sysseek($fh, $i, SEEK_SET) or croak "seek: $!";
2844                sysread($fh, my $buf, 24) == 24 or croak "read: $!";
2845                my ($r, $c) = unpack('NH40', $buf);
2846
2847                if ($r < $rev) {
2848                        $l = $i + 24;
2849                } elsif ($r > $rev) {
2850                        $u = $i - 24;
2851                } else { # $r == $rev
2852                        close($fh) or croak "close: $!";
2853                        return $c eq ('0' x 40) ? undef : $c;
2854                }
2855        }
2856        close($fh) or croak "close: $!";
2857        undef;
2858}
2859
2860# Finds the first svn revision that exists on (if $eq_ok is true) or
2861# before $rev for the current branch.  It will not search any lower
2862# than $min_rev.  Returns the git commit hash and svn revision number
2863# if found, else (undef, undef).
2864sub find_rev_before {
2865        my ($self, $rev, $eq_ok, $min_rev) = @_;
2866        --$rev unless $eq_ok;
2867        $min_rev ||= 1;
2868        while ($rev >= $min_rev) {
2869                if (my $c = $self->rev_map_get($rev)) {
2870                        return ($rev, $c);
2871                }
2872                --$rev;
2873        }
2874        return (undef, undef);
2875}
2876
2877# Finds the first svn revision that exists on (if $eq_ok is true) or
2878# after $rev for the current branch.  It will not search any higher
2879# than $max_rev.  Returns the git commit hash and svn revision number
2880# if found, else (undef, undef).
2881sub find_rev_after {
2882        my ($self, $rev, $eq_ok, $max_rev) = @_;
2883        ++$rev unless $eq_ok;
2884        $max_rev ||= $self->rev_map_max;
2885        while ($rev <= $max_rev) {
2886                if (my $c = $self->rev_map_get($rev)) {
2887                        return ($rev, $c);
2888                }
2889                ++$rev;
2890        }
2891        return (undef, undef);
2892}
2893
2894sub _new {
2895        my ($class, $repo_id, $ref_id, $path) = @_;
2896        unless (defined $repo_id && length $repo_id) {
2897                $repo_id = $Git::SVN::default_repo_id;
2898        }
2899        unless (defined $ref_id && length $ref_id) {
2900                $_[2] = $ref_id = $Git::SVN::default_ref_id;
2901        }
2902        $_[1] = $repo_id;
2903        my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
2904        $_[3] = $path = '' unless (defined $path);
2905        mkpath(["$ENV{GIT_DIR}/svn"]);
2906        bless {
2907                ref_id => $ref_id, dir => $dir, index => "$dir/index",
2908                path => $path, config => "$ENV{GIT_DIR}/svn/config",
2909                map_root => "$dir/.rev_map", repo_id => $repo_id }, $class;
2910}
2911
2912# for read-only access of old .rev_db formats
2913sub unlink_rev_db_symlink {
2914        my ($self) = @_;
2915        my $link = $self->rev_db_path;
2916        $link =~ s/\.[\w-]+$// or croak "missing UUID at the end of $link";
2917        if (-l $link) {
2918                unlink $link or croak "unlink: $link failed!";
2919        }
2920}
2921
2922sub rev_db_path {
2923        my ($self, $uuid) = @_;
2924        my $db_path = $self->map_path($uuid);
2925        $db_path =~ s{/\.rev_map\.}{/\.rev_db\.}
2926            or croak "map_path: $db_path does not contain '/.rev_map.' !";
2927        $db_path;
2928}
2929
2930# the new replacement for .rev_db
2931sub map_path {
2932        my ($self, $uuid) = @_;
2933        $uuid ||= $self->ra_uuid;
2934        "$self->{map_root}.$uuid";
2935}
2936
2937sub uri_encode {
2938        my ($f) = @_;
2939        $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
2940        $f
2941}
2942
2943sub remove_username {
2944        $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
2945}
2946
2947package Git::SVN::Prompt;
2948use strict;
2949use warnings;
2950require SVN::Core;
2951use vars qw/$_no_auth_cache $_username/;
2952
2953sub simple {
2954        my ($cred, $realm, $default_username, $may_save, $pool) = @_;
2955        $may_save = undef if $_no_auth_cache;
2956        $default_username = $_username if defined $_username;
2957        if (defined $default_username && length $default_username) {
2958                if (defined $realm && length $realm) {
2959                        print STDERR "Authentication realm: $realm\n";
2960                        STDERR->flush;
2961                }
2962                $cred->username($default_username);
2963        } else {
2964                username($cred, $realm, $may_save, $pool);
2965        }
2966        $cred->password(_read_password("Password for '" .
2967                                       $cred->username . "': ", $realm));
2968        $cred->may_save($may_save);
2969        $SVN::_Core::SVN_NO_ERROR;
2970}
2971
2972sub ssl_server_trust {
2973        my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
2974        $may_save = undef if $_no_auth_cache;
2975        print STDERR "Error validating server certificate for '$realm':\n";
2976        {
2977                no warnings 'once';
2978                # All variables SVN::Auth::SSL::* are used only once,
2979                # so we're shutting up Perl warnings about this.
2980                if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
2981                        print STDERR " - The certificate is not issued ",
2982                            "by a trusted authority. Use the\n",
2983                            "   fingerprint to validate ",
2984                            "the certificate manually!\n";
2985                }
2986                if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
2987                        print STDERR " - The certificate hostname ",
2988                            "does not match.\n";
2989                }
2990                if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
2991                        print STDERR " - The certificate is not yet valid.\n";
2992                }
2993                if ($failures & $SVN::Auth::SSL::EXPIRED) {
2994                        print STDERR " - The certificate has expired.\n";
2995                }
2996                if ($failures & $SVN::Auth::SSL::OTHER) {
2997                        print STDERR " - The certificate has ",
2998                            "an unknown error.\n";
2999                }
3000        } # no warnings 'once'
3001        printf STDERR
3002                "Certificate information:\n".
3003                " - Hostname: %s\n".
3004                " - Valid: from %s until %s\n".
3005                " - Issuer: %s\n".
3006                " - Fingerprint: %s\n",
3007                map $cert_info->$_, qw(hostname valid_from valid_until
3008                                       issuer_dname fingerprint);
3009        my $choice;
3010prompt:
3011        print STDERR $may_save ?
3012              "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
3013              "(R)eject or accept (t)emporarily? ";
3014        STDERR->flush;
3015        $choice = lc(substr(<STDIN> || 'R', 0, 1));
3016        if ($choice =~ /^t$/i) {
3017                $cred->may_save(undef);
3018        } elsif ($choice =~ /^r$/i) {
3019                return -1;
3020        } elsif ($may_save && $choice =~ /^p$/i) {
3021                $cred->may_save($may_save);
3022        } else {
3023                goto prompt;
3024        }
3025        $cred->accepted_failures($failures);
3026        $SVN::_Core::SVN_NO_ERROR;
3027}
3028
3029sub ssl_client_cert {
3030        my ($cred, $realm, $may_save, $pool) = @_;
3031        $may_save = undef if $_no_auth_cache;
3032        print STDERR "Client certificate filename: ";
3033        STDERR->flush;
3034        chomp(my $filename = <STDIN>);
3035        $cred->cert_file($filename);
3036        $cred->may_save($may_save);
3037        $SVN::_Core::SVN_NO_ERROR;
3038}
3039
3040sub ssl_client_cert_pw {
3041        my ($cred, $realm, $may_save, $pool) = @_;
3042        $may_save = undef if $_no_auth_cache;
3043        $cred->password(_read_password("Password: ", $realm));
3044        $cred->may_save($may_save);
3045        $SVN::_Core::SVN_NO_ERROR;
3046}
3047
3048sub username {
3049        my ($cred, $realm, $may_save, $pool) = @_;
3050        $may_save = undef if $_no_auth_cache;
3051        if (defined $realm && length $realm) {
3052                print STDERR "Authentication realm: $realm\n";
3053        }
3054        my $username;
3055        if (defined $_username) {
3056                $username = $_username;
3057        } else {
3058                print STDERR "Username: ";
3059                STDERR->flush;
3060                chomp($username = <STDIN>);
3061        }
3062        $cred->username($username);
3063        $cred->may_save($may_save);
3064        $SVN::_Core::SVN_NO_ERROR;
3065}
3066
3067sub _read_password {
3068        my ($prompt, $realm) = @_;
3069        print STDERR $prompt;
3070        STDERR->flush;
3071        require Term::ReadKey;
3072        Term::ReadKey::ReadMode('noecho');
3073        my $password = '';
3074        while (defined(my $key = Term::ReadKey::ReadKey(0))) {
3075                last if $key =~ /[\012\015]/; # \n\r
3076                $password .= $key;
3077        }
3078        Term::ReadKey::ReadMode('restore');
3079        print STDERR "\n";
3080        STDERR->flush;
3081        $password;
3082}
3083
3084package SVN::Git::Fetcher;
3085use vars qw/@ISA/;
3086use strict;
3087use warnings;
3088use Carp qw/croak/;
3089use File::Temp qw/tempfile/;
3090use IO::File qw//;
3091
3092# file baton members: path, mode_a, mode_b, pool, fh, blob, base
3093sub new {
3094        my ($class, $git_svn) = @_;
3095        my $self = SVN::Delta::Editor->new;
3096        bless $self, $class;
3097        $self->{c} = $git_svn->{last_commit} if exists $git_svn->{last_commit};
3098        $self->{empty} = {};
3099        $self->{dir_prop} = {};
3100        $self->{file_prop} = {};
3101        $self->{absent_dir} = {};
3102        $self->{absent_file} = {};
3103        $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
3104        $self;
3105}
3106
3107sub set_path_strip {
3108        my ($self, $path) = @_;
3109        $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
3110}
3111
3112sub open_root {
3113        { path => '' };
3114}
3115
3116sub open_directory {
3117        my ($self, $path, $pb, $rev) = @_;
3118        { path => $path };
3119}
3120
3121sub git_path {
3122        my ($self, $path) = @_;
3123        if ($self->{path_strip}) {
3124                $path =~ s!$self->{path_strip}!! or
3125                  die "Failed to strip path '$path' ($self->{path_strip})\n";
3126        }
3127        $path;
3128}
3129
3130sub delete_entry {
3131        my ($self, $path, $rev, $pb) = @_;
3132
3133        my $gpath = $self->git_path($path);
3134        return undef if ($gpath eq '');
3135
3136        # remove entire directories.
3137        if (command('ls-tree', $self->{c}, '--', $gpath) =~ /^040000 tree/) {
3138                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3139                                                     -r --name-only -z/,
3140                                                     $self->{c}, '--', $gpath);
3141                local $/ = "\0";
3142                while (<$ls>) {
3143                        chomp;
3144                        $self->{gii}->remove($_);
3145                        print "\tD\t$_\n" unless $::_q;
3146                }
3147                print "\tD\t$gpath/\n" unless $::_q;
3148                command_close_pipe($ls, $ctx);
3149                $self->{empty}->{$path} = 0
3150        } else {
3151                $self->{gii}->remove($gpath);
3152                print "\tD\t$gpath\n" unless $::_q;
3153        }
3154        undef;
3155}
3156
3157sub open_file {
3158        my ($self, $path, $pb, $rev) = @_;
3159        my $gpath = $self->git_path($path);
3160        my ($mode, $blob) = (command('ls-tree', $self->{c}, '--', $gpath)
3161                             =~ /^(\d{6}) blob ([a-f\d]{40})\t/);
3162        unless (defined $mode && defined $blob) {
3163                die "$path was not found in commit $self->{c} (r$rev)\n";
3164        }
3165        { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
3166          pool => SVN::Pool->new, action => 'M' };
3167}
3168
3169sub add_file {
3170        my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
3171        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3172        delete $self->{empty}->{$dir};
3173        { path => $path, mode_a => 100644, mode_b => 100644,
3174          pool => SVN::Pool->new, action => 'A' };
3175}
3176
3177sub add_directory {
3178        my ($self, $path, $cp_path, $cp_rev) = @_;
3179        my $gpath = $self->git_path($path);
3180        if ($gpath eq '') {
3181                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3182                                                     -r --name-only -z/,
3183                                                     $self->{c});
3184                local $/ = "\0";
3185                while (<$ls>) {
3186                        chomp;
3187                        $self->{gii}->remove($_);
3188                        print "\tD\t$_\n" unless $::_q;
3189                }
3190                command_close_pipe($ls, $ctx);
3191                $self->{empty}->{$path} = 0;
3192        }
3193        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3194        delete $self->{empty}->{$dir};
3195        $self->{empty}->{$path} = 1;
3196        { path => $path };
3197}
3198
3199sub change_dir_prop {
3200        my ($self, $db, $prop, $value) = @_;
3201        $self->{dir_prop}->{$db->{path}} ||= {};
3202        $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
3203        undef;
3204}
3205
3206sub absent_directory {
3207        my ($self, $path, $pb) = @_;
3208        $self->{absent_dir}->{$pb->{path}} ||= [];
3209        push @{$self->{absent_dir}->{$pb->{path}}}, $path;
3210        undef;
3211}
3212
3213sub absent_file {
3214        my ($self, $path, $pb) = @_;
3215        $self->{absent_file}->{$pb->{path}} ||= [];
3216        push @{$self->{absent_file}->{$pb->{path}}}, $path;
3217        undef;
3218}
3219
3220sub change_file_prop {
3221        my ($self, $fb, $prop, $value) = @_;
3222        if ($prop eq 'svn:executable') {
3223                if ($fb->{mode_b} != 120000) {
3224                        $fb->{mode_b} = defined $value ? 100755 : 100644;
3225                }
3226        } elsif ($prop eq 'svn:special') {
3227                $fb->{mode_b} = defined $value ? 120000 : 100644;
3228        } else {
3229                $self->{file_prop}->{$fb->{path}} ||= {};
3230                $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
3231        }
3232        undef;
3233}
3234
3235sub apply_textdelta {
3236        my ($self, $fb, $exp) = @_;
3237        my $fh = Git::temp_acquire('svn_delta');
3238        # $fh gets auto-closed() by SVN::TxDelta::apply(),
3239        # (but $base does not,) so dup() it for reading in close_file
3240        open my $dup, '<&', $fh or croak $!;
3241        my $base = Git::temp_acquire('git_blob');
3242        if ($fb->{blob}) {
3243                print $base 'link ' if ($fb->{mode_a} == 120000);
3244                my $size = $::_repository->cat_blob($fb->{blob}, $base);
3245                die "Failed to read object $fb->{blob}" if ($size < 0);
3246
3247                if (defined $exp) {
3248                        seek $base, 0, 0 or croak $!;
3249                        my $got = ::md5sum($base);
3250                        die "Checksum mismatch: $fb->{path} $fb->{blob}\n",
3251                            "expected: $exp\n",
3252                            "     got: $got\n" if ($got ne $exp);
3253                }
3254        }
3255        seek $base, 0, 0 or croak $!;
3256        $fb->{fh} = $fh;
3257        $fb->{base} = $base;
3258        [ SVN::TxDelta::apply($base, $dup, undef, $fb->{path}, $fb->{pool}) ];
3259}
3260
3261sub close_file {
3262        my ($self, $fb, $exp) = @_;
3263        my $hash;
3264        my $path = $self->git_path($fb->{path});
3265        if (my $fh = $fb->{fh}) {
3266                if (defined $exp) {
3267                        seek($fh, 0, 0) or croak $!;
3268                        my $got = ::md5sum($fh);
3269                        if ($got ne $exp) {
3270                                die "Checksum mismatch: $path\n",
3271                                    "expected: $exp\n    got: $got\n";
3272                        }
3273                }
3274                if ($fb->{mode_b} == 120000) {
3275                        sysseek($fh, 0, 0) or croak $!;
3276                        sysread($fh, my $buf, 5) == 5 or croak $!;
3277
3278                        unless ($buf eq 'link ') {
3279                                warn "$path has mode 120000",
3280                                                " but is not a link\n";
3281                        } else {
3282                                my $tmp_fh = Git::temp_acquire('svn_hash');
3283                                my $res;
3284                                while ($res = sysread($fh, my $str, 1024)) {
3285                                        my $out = syswrite($tmp_fh, $str, $res);
3286                                        defined($out) && $out == $res
3287                                                or croak("write ",
3288                                                        $tmp_fh->filename,
3289                                                        ": $!\n");
3290                                }
3291                                defined $res or croak $!;
3292
3293                                ($fh, $tmp_fh) = ($tmp_fh, $fh);
3294                                Git::temp_release($tmp_fh, 1);
3295                        }
3296                }
3297
3298                $hash = $::_repository->hash_and_insert_object(
3299                                $fh->filename);
3300                $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
3301
3302                Git::temp_release($fb->{base}, 1);
3303                Git::temp_release($fh, 1);
3304        } else {
3305                $hash = $fb->{blob} or die "no blob information\n";
3306        }
3307        $fb->{pool}->clear;
3308        $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
3309        print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
3310        undef;
3311}
3312
3313sub abort_edit {
3314        my $self = shift;
3315        $self->{nr} = $self->{gii}->{nr};
3316        delete $self->{gii};
3317        $self->SUPER::abort_edit(@_);
3318}
3319
3320sub close_edit {
3321        my $self = shift;
3322        $self->{git_commit_ok} = 1;
3323        $self->{nr} = $self->{gii}->{nr};
3324        delete $self->{gii};
3325        $self->SUPER::close_edit(@_);
3326}
3327
3328package SVN::Git::Editor;
3329use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
3330use strict;
3331use warnings;
3332use Carp qw/croak/;
3333use IO::File;
3334
3335sub new {
3336        my ($class, $opts) = @_;
3337        foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
3338                die "$_ required!\n" unless (defined $opts->{$_});
3339        }
3340
3341        my $pool = SVN::Pool->new;
3342        my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
3343        my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
3344                                     $opts->{r}, $mods);
3345
3346        # $opts->{ra} functions should not be used after this:
3347        my @ce  = $opts->{ra}->get_commit_editor($opts->{log},
3348                                                $opts->{editor_cb}, $pool);
3349        my $self = SVN::Delta::Editor->new(@ce, $pool);
3350        bless $self, $class;
3351        foreach (qw/svn_path r tree_a tree_b/) {
3352                $self->{$_} = $opts->{$_};
3353        }
3354        $self->{url} = $opts->{ra}->{url};
3355        $self->{mods} = $mods;
3356        $self->{types} = $types;
3357        $self->{pool} = $pool;
3358        $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
3359        $self->{rm} = { };
3360        $self->{path_prefix} = length $self->{svn_path} ?
3361                               "$self->{svn_path}/" : '';
3362        $self->{config} = $opts->{config};
3363        return $self;
3364}
3365
3366sub generate_diff {
3367        my ($tree_a, $tree_b) = @_;
3368        my @diff_tree = qw(diff-tree -z -r);
3369        if ($_cp_similarity) {
3370                push @diff_tree, "-C$_cp_similarity";
3371        } else {
3372                push @diff_tree, '-C';
3373        }
3374        push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
3375        push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
3376        push @diff_tree, $tree_a, $tree_b;
3377        my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
3378        local $/ = "\0";
3379        my $state = 'meta';
3380        my @mods;
3381        while (<$diff_fh>) {
3382                chomp $_; # this gets rid of the trailing "\0"
3383                if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
3384                                        ($::sha1)\s($::sha1)\s
3385                                        ([MTCRAD])\d*$/xo) {
3386                        push @mods, {   mode_a => $1, mode_b => $2,
3387                                        sha1_a => $3, sha1_b => $4,
3388                                        chg => $5 };
3389                        if ($5 =~ /^(?:C|R)$/) {
3390                                $state = 'file_a';
3391                        } else {
3392                                $state = 'file_b';
3393                        }
3394                } elsif ($state eq 'file_a') {
3395                        my $x = $mods[$#mods] or croak "Empty array\n";
3396                        if ($x->{chg} !~ /^(?:C|R)$/) {
3397                                croak "Error parsing $_, $x->{chg}\n";
3398                        }
3399                        $x->{file_a} = $_;
3400                        $state = 'file_b';
3401                } elsif ($state eq 'file_b') {
3402                        my $x = $mods[$#mods] or croak "Empty array\n";
3403                        if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
3404                                croak "Error parsing $_, $x->{chg}\n";
3405                        }
3406                        if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
3407                                croak "Error parsing $_, $x->{chg}\n";
3408                        }
3409                        $x->{file_b} = $_;
3410                        $state = 'meta';
3411                } else {
3412                        croak "Error parsing $_\n";
3413                }
3414        }
3415        command_close_pipe($diff_fh, $ctx);
3416        \@mods;
3417}
3418
3419sub check_diff_paths {
3420        my ($ra, $pfx, $rev, $mods) = @_;
3421        my %types;
3422        $pfx .= '/' if length $pfx;
3423
3424        sub type_diff_paths {
3425                my ($ra, $types, $path, $rev) = @_;
3426                my @p = split m#/+#, $path;
3427                my $c = shift @p;
3428                unless (defined $types->{$c}) {
3429                        $types->{$c} = $ra->check_path($c, $rev);
3430                }
3431                while (@p) {
3432                        $c .= '/' . shift @p;
3433                        next if defined $types->{$c};
3434                        $types->{$c} = $ra->check_path($c, $rev);
3435                }
3436        }
3437
3438        foreach my $m (@$mods) {
3439                foreach my $f (qw/file_a file_b/) {
3440                        next unless defined $m->{$f};
3441                        my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
3442                        if (length $pfx.$dir && ! defined $types{$dir}) {
3443                                type_diff_paths($ra, \%types, $pfx.$dir, $rev);
3444                        }
3445                }
3446        }
3447        \%types;
3448}
3449
3450sub split_path {
3451        return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
3452}
3453
3454sub repo_path {
3455        my ($self, $path) = @_;
3456        $self->{path_prefix}.(defined $path ? $path : '');
3457}
3458
3459sub url_path {
3460        my ($self, $path) = @_;
3461        if ($self->{url} =~ m#^https?://#) {
3462                $path =~ s/([^a-zA-Z0-9_.-])/uc sprintf("%%%02x",ord($1))/eg;
3463        }
3464        $self->{url} . '/' . $self->repo_path($path);
3465}
3466
3467sub rmdirs {
3468        my ($self) = @_;
3469        my $rm = $self->{rm};
3470        delete $rm->{''}; # we never delete the url we're tracking
3471        return unless %$rm;
3472
3473        foreach (keys %$rm) {
3474                my @d = split m#/#, $_;
3475                my $c = shift @d;
3476                $rm->{$c} = 1;
3477                while (@d) {
3478                        $c .= '/' . shift @d;
3479                        $rm->{$c} = 1;
3480                }
3481        }
3482        delete $rm->{$self->{svn_path}};
3483        delete $rm->{''}; # we never delete the url we're tracking
3484        return unless %$rm;
3485
3486        my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
3487                                             $self->{tree_b});
3488        local $/ = "\0";
3489        while (<$fh>) {
3490                chomp;
3491                my @dn = split m#/#, $_;
3492                while (pop @dn) {
3493                        delete $rm->{join '/', @dn};
3494                }
3495                unless (%$rm) {
3496                        close $fh;
3497                        return;
3498                }
3499        }
3500        command_close_pipe($fh, $ctx);
3501
3502        my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
3503        foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
3504                $self->close_directory($bat->{$d}, $p);
3505                my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
3506                print "\tD+\t$d/\n" unless $::_q;
3507                $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
3508                delete $bat->{$d};
3509        }
3510}
3511
3512sub open_or_add_dir {
3513        my ($self, $full_path, $baton) = @_;
3514        my $t = $self->{types}->{$full_path};
3515        if (!defined $t) {
3516                die "$full_path not known in r$self->{r} or we have a bug!\n";
3517        }
3518        {
3519                no warnings 'once';
3520                # SVN::Node::none and SVN::Node::file are used only once,
3521                # so we're shutting up Perl's warnings about them.
3522                if ($t == $SVN::Node::none) {
3523                        return $self->add_directory($full_path, $baton,
3524                            undef, -1, $self->{pool});
3525                } elsif ($t == $SVN::Node::dir) {
3526                        return $self->open_directory($full_path, $baton,
3527                            $self->{r}, $self->{pool});
3528                } # no warnings 'once'
3529                print STDERR "$full_path already exists in repository at ",
3530                    "r$self->{r} and it is not a directory (",
3531                    ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
3532        } # no warnings 'once'
3533        exit 1;
3534}
3535
3536sub ensure_path {
3537        my ($self, $path) = @_;
3538        my $bat = $self->{bat};
3539        my $repo_path = $self->repo_path($path);
3540        return $bat->{''} unless (length $repo_path);
3541        my @p = split m#/+#, $repo_path;
3542        my $c = shift @p;
3543        $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
3544        while (@p) {
3545                my $c0 = $c;
3546                $c .= '/' . shift @p;
3547                $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
3548        }
3549        return $bat->{$c};
3550}
3551
3552# Subroutine to convert a globbing pattern to a regular expression.
3553# From perl cookbook.
3554sub glob2pat {
3555        my $globstr = shift;
3556        my %patmap = ('*' => '.*', '?' => '.', '[' => '[', ']' => ']');
3557        $globstr =~ s{(.)} { $patmap{$1} || "\Q$1" }ge;
3558        return '^' . $globstr . '$';
3559}
3560
3561sub check_autoprop {
3562        my ($self, $pattern, $properties, $file, $fbat) = @_;
3563        # Convert the globbing pattern to a regular expression.
3564        my $regex = glob2pat($pattern);
3565        # Check if the pattern matches the file name.
3566        if($file =~ m/($regex)/) {
3567                # Parse the list of properties to set.
3568                my @props = split(/;/, $properties);
3569                foreach my $prop (@props) {
3570                        # Parse 'name=value' syntax and set the property.
3571                        if ($prop =~ /([^=]+)=(.*)/) {
3572                                my ($n,$v) = ($1,$2);
3573                                for ($n, $v) {
3574                                        s/^\s+//; s/\s+$//;
3575                                }
3576                                $self->change_file_prop($fbat, $n, $v);
3577                        }
3578                }
3579        }
3580}
3581
3582sub apply_autoprops {
3583        my ($self, $file, $fbat) = @_;
3584        my $conf_t = ${$self->{config}}{'config'};
3585        no warnings 'once';
3586        # Check [miscellany]/enable-auto-props in svn configuration.
3587        if (SVN::_Core::svn_config_get_bool(
3588                $conf_t,
3589                $SVN::_Core::SVN_CONFIG_SECTION_MISCELLANY,
3590                $SVN::_Core::SVN_CONFIG_OPTION_ENABLE_AUTO_PROPS,
3591                0)) {
3592                # Auto-props are enabled.  Enumerate them to look for matches.
3593                my $callback = sub {
3594                        $self->check_autoprop($_[0], $_[1], $file, $fbat);
3595                };
3596                SVN::_Core::svn_config_enumerate(
3597                        $conf_t,
3598                        $SVN::_Core::SVN_CONFIG_SECTION_AUTO_PROPS,
3599                        $callback);
3600        }
3601}
3602
3603sub A {
3604        my ($self, $m) = @_;
3605        my ($dir, $file) = split_path($m->{file_b});
3606        my $pbat = $self->ensure_path($dir);
3607        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3608                                        undef, -1);
3609        print "\tA\t$m->{file_b}\n" unless $::_q;
3610        $self->apply_autoprops($file, $fbat);
3611        $self->chg_file($fbat, $m);
3612        $self->close_file($fbat,undef,$self->{pool});
3613}
3614
3615sub C {
3616        my ($self, $m) = @_;
3617        my ($dir, $file) = split_path($m->{file_b});
3618        my $pbat = $self->ensure_path($dir);
3619        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3620                                $self->url_path($m->{file_a}), $self->{r});
3621        print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
3622        $self->chg_file($fbat, $m);
3623        $self->close_file($fbat,undef,$self->{pool});
3624}
3625
3626sub delete_entry {
3627        my ($self, $path, $pbat) = @_;
3628        my $rpath = $self->repo_path($path);
3629        my ($dir, $file) = split_path($rpath);
3630        $self->{rm}->{$dir} = 1;
3631        $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
3632}
3633
3634sub R {
3635        my ($self, $m) = @_;
3636        my ($dir, $file) = split_path($m->{file_b});
3637        my $pbat = $self->ensure_path($dir);
3638        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3639                                $self->url_path($m->{file_a}), $self->{r});
3640        print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
3641        $self->chg_file($fbat, $m);
3642        $self->close_file($fbat,undef,$self->{pool});
3643
3644        ($dir, $file) = split_path($m->{file_a});
3645        $pbat = $self->ensure_path($dir);
3646        $self->delete_entry($m->{file_a}, $pbat);
3647}
3648
3649sub M {
3650        my ($self, $m) = @_;
3651        my ($dir, $file) = split_path($m->{file_b});
3652        my $pbat = $self->ensure_path($dir);
3653        my $fbat = $self->open_file($self->repo_path($m->{file_b}),
3654                                $pbat,$self->{r},$self->{pool});
3655        print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
3656        $self->chg_file($fbat, $m);
3657        $self->close_file($fbat,undef,$self->{pool});
3658}
3659
3660sub T { shift->M(@_) }
3661
3662sub change_file_prop {
3663        my ($self, $fbat, $pname, $pval) = @_;
3664        $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
3665}
3666
3667sub _chg_file_get_blob ($$$$) {
3668        my ($self, $fbat, $m, $which) = @_;
3669        my $fh = Git::temp_acquire("git_blob_$which");
3670        if ($m->{"mode_$which"} =~ /^120/) {
3671                print $fh 'link ' or croak $!;
3672                $self->change_file_prop($fbat,'svn:special','*');
3673        } elsif ($m->{mode_a} =~ /^120/ && $m->{"mode_$which"} !~ /^120/) {
3674                $self->change_file_prop($fbat,'svn:special',undef);
3675        }
3676        my $blob = $m->{"sha1_$which"};
3677        return ($fh,) if ($blob =~ /^0{40}$/);
3678        my $size = $::_repository->cat_blob($blob, $fh);
3679        croak "Failed to read object $blob" if ($size < 0);
3680        $fh->flush == 0 or croak $!;
3681        seek $fh, 0, 0 or croak $!;
3682
3683        my $exp = ::md5sum($fh);
3684        seek $fh, 0, 0 or croak $!;
3685        return ($fh, $exp);
3686}
3687
3688sub chg_file {
3689        my ($self, $fbat, $m) = @_;
3690        if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
3691                $self->change_file_prop($fbat,'svn:executable','*');
3692        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
3693                $self->change_file_prop($fbat,'svn:executable',undef);
3694        }
3695        my ($fh_a, $exp_a) = _chg_file_get_blob $self, $fbat, $m, 'a';
3696        my ($fh_b, $exp_b) = _chg_file_get_blob $self, $fbat, $m, 'b';
3697        my $pool = SVN::Pool->new;
3698        my $atd = $self->apply_textdelta($fbat, $exp_a, $pool);
3699        if (-s $fh_a) {
3700                my $txstream = SVN::TxDelta::new ($fh_a, $fh_b, $pool);
3701                my $res = SVN::TxDelta::send_txstream($txstream, @$atd, $pool);
3702                if (defined $res) {
3703                        die "Unexpected result from send_txstream: $res\n",
3704                            "(SVN::Core::VERSION: $SVN::Core::VERSION)\n";
3705                }
3706        } else {
3707                my $got = SVN::TxDelta::send_stream($fh_b, @$atd, $pool);
3708                die "Checksum mismatch\nexpected: $exp_b\ngot: $got\n"
3709                    if ($got ne $exp_b);
3710        }
3711        Git::temp_release($fh_b, 1);
3712        Git::temp_release($fh_a, 1);
3713        $pool->clear;
3714}
3715
3716sub D {
3717        my ($self, $m) = @_;
3718        my ($dir, $file) = split_path($m->{file_b});
3719        my $pbat = $self->ensure_path($dir);
3720        print "\tD\t$m->{file_b}\n" unless $::_q;
3721        $self->delete_entry($m->{file_b}, $pbat);
3722}
3723
3724sub close_edit {
3725        my ($self) = @_;
3726        my ($p,$bat) = ($self->{pool}, $self->{bat});
3727        foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
3728                next if $_ eq '';
3729                $self->close_directory($bat->{$_}, $p);
3730        }
3731        $self->close_directory($bat->{''}, $p);
3732        $self->SUPER::close_edit($p);
3733        $p->clear;
3734}
3735
3736sub abort_edit {
3737        my ($self) = @_;
3738        $self->SUPER::abort_edit($self->{pool});
3739}
3740
3741sub DESTROY {
3742        my $self = shift;
3743        $self->SUPER::DESTROY(@_);
3744        $self->{pool}->clear;
3745}
3746
3747# this drives the editor
3748sub apply_diff {
3749        my ($self) = @_;
3750        my $mods = $self->{mods};
3751        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
3752        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
3753                my $f = $m->{chg};
3754                if (defined $o{$f}) {
3755                        $self->$f($m);
3756                } else {
3757                        fatal("Invalid change type: $f");
3758                }
3759        }
3760        $self->rmdirs if $_rmdir;
3761        if (@$mods == 0) {
3762                $self->abort_edit;
3763        } else {
3764                $self->close_edit;
3765        }
3766        return scalar @$mods;
3767}
3768
3769package Git::SVN::Ra;
3770use vars qw/@ISA $config_dir $_log_window_size/;
3771use strict;
3772use warnings;
3773my ($ra_invalid, $can_do_switch, %ignored_err, $RA);
3774
3775BEGIN {
3776        # enforce temporary pool usage for some simple functions
3777        no strict 'refs';
3778        for my $f (qw/rev_proplist get_latest_revnum get_uuid get_repos_root/) {
3779                my $SUPER = "SUPER::$f";
3780                *$f = sub {
3781                        my $self = shift;
3782                        my $pool = SVN::Pool->new;
3783                        my @ret = $self->$SUPER(@_,$pool);
3784                        $pool->clear;
3785                        wantarray ? @ret : $ret[0];
3786                };
3787        }
3788}
3789
3790sub _auth_providers () {
3791        [
3792          SVN::Client::get_simple_provider(),
3793          SVN::Client::get_ssl_server_trust_file_provider(),
3794          SVN::Client::get_simple_prompt_provider(
3795            \&Git::SVN::Prompt::simple, 2),
3796          SVN::Client::get_ssl_client_cert_file_provider(),
3797          SVN::Client::get_ssl_client_cert_prompt_provider(
3798            \&Git::SVN::Prompt::ssl_client_cert, 2),
3799          SVN::Client::get_ssl_client_cert_pw_file_provider(),
3800          SVN::Client::get_ssl_client_cert_pw_prompt_provider(
3801            \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
3802          SVN::Client::get_username_provider(),
3803          SVN::Client::get_ssl_server_trust_prompt_provider(
3804            \&Git::SVN::Prompt::ssl_server_trust),
3805          SVN::Client::get_username_prompt_provider(
3806            \&Git::SVN::Prompt::username, 2)
3807        ]
3808}
3809
3810sub escape_uri_only {
3811        my ($uri) = @_;
3812        my @tmp;
3813        foreach (split m{/}, $uri) {
3814                s/([^\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
3815                push @tmp, $_;
3816        }
3817        join('/', @tmp);
3818}
3819
3820sub escape_url {
3821        my ($url) = @_;
3822        if ($url =~ m#^(https?)://([^/]+)(.*)$#) {
3823                my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
3824                $url = "$scheme://$domain$uri";
3825        }
3826        $url;
3827}
3828
3829sub new {
3830        my ($class, $url) = @_;
3831        $url =~ s!/+$!!;
3832        return $RA if ($RA && $RA->{url} eq $url);
3833
3834        SVN::_Core::svn_config_ensure($config_dir, undef);
3835        my ($baton, $callbacks) = SVN::Core::auth_open_helper(_auth_providers);
3836        my $config = SVN::Core::config_get_config($config_dir);
3837        $RA = undef;
3838        my $dont_store_passwords = 1;
3839        my $conf_t = ${$config}{'config'};
3840        {
3841                no warnings 'once';
3842                # The usage of $SVN::_Core::SVN_CONFIG_* variables
3843                # produces warnings that variables are used only once.
3844                # I had not found the better way to shut them up, so
3845                # the warnings of type 'once' are disabled in this block.
3846                if (SVN::_Core::svn_config_get_bool($conf_t,
3847                    $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
3848                    $SVN::_Core::SVN_CONFIG_OPTION_STORE_PASSWORDS,
3849                    1) == 0) {
3850                        SVN::_Core::svn_auth_set_parameter($baton,
3851                            $SVN::_Core::SVN_AUTH_PARAM_DONT_STORE_PASSWORDS,
3852                            bless (\$dont_store_passwords, "_p_void"));
3853                }
3854                if (SVN::_Core::svn_config_get_bool($conf_t,
3855                    $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
3856                    $SVN::_Core::SVN_CONFIG_OPTION_STORE_AUTH_CREDS,
3857                    1) == 0) {
3858                        $Git::SVN::Prompt::_no_auth_cache = 1;
3859                }
3860        } # no warnings 'once'
3861        my $self = SVN::Ra->new(url => escape_url($url), auth => $baton,
3862                              config => $config,
3863                              pool => SVN::Pool->new,
3864                              auth_provider_callbacks => $callbacks);
3865        $self->{url} = $url;
3866        $self->{svn_path} = $url;
3867        $self->{repos_root} = $self->get_repos_root;
3868        $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
3869        $self->{cache} = { check_path => { r => 0, data => {} },
3870                           get_dir => { r => 0, data => {} } };
3871        $RA = bless $self, $class;
3872}
3873
3874sub check_path {
3875        my ($self, $path, $r) = @_;
3876        my $cache = $self->{cache}->{check_path};
3877        if ($r == $cache->{r} && exists $cache->{data}->{$path}) {
3878                return $cache->{data}->{$path};
3879        }
3880        my $pool = SVN::Pool->new;
3881        my $t = $self->SUPER::check_path($path, $r, $pool);
3882        $pool->clear;
3883        if ($r != $cache->{r}) {
3884                %{$cache->{data}} = ();
3885                $cache->{r} = $r;
3886        }
3887        $cache->{data}->{$path} = $t;
3888}
3889
3890sub get_dir {
3891        my ($self, $dir, $r) = @_;
3892        my $cache = $self->{cache}->{get_dir};
3893        if ($r == $cache->{r}) {
3894                if (my $x = $cache->{data}->{$dir}) {
3895                        return wantarray ? @$x : $x->[0];
3896                }
3897        }
3898        my $pool = SVN::Pool->new;
3899        my ($d, undef, $props) = $self->SUPER::get_dir($dir, $r, $pool);
3900        my %dirents = map { $_ => { kind => $d->{$_}->kind } } keys %$d;
3901        $pool->clear;
3902        if ($r != $cache->{r}) {
3903                %{$cache->{data}} = ();
3904                $cache->{r} = $r;
3905        }
3906        $cache->{data}->{$dir} = [ \%dirents, $r, $props ];
3907        wantarray ? (\%dirents, $r, $props) : \%dirents;
3908}
3909
3910sub DESTROY {
3911        # do not call the real DESTROY since we store ourselves in $RA
3912}
3913
3914sub get_log {
3915        my ($self, @args) = @_;
3916        my $pool = SVN::Pool->new;
3917        splice(@args, 3, 1) if ($SVN::Core::VERSION le '1.2.0');
3918        my $ret = $self->SUPER::get_log(@args, $pool);
3919        $pool->clear;
3920        $ret;
3921}
3922
3923sub trees_match {
3924        my ($self, $url1, $rev1, $url2, $rev2) = @_;
3925        my $ctx = SVN::Client->new(auth => _auth_providers);
3926        my $out = IO::File->new_tmpfile;
3927
3928        # older SVN (1.1.x) doesn't take $pool as the last parameter for
3929        # $ctx->diff(), so we'll create a default one
3930        my $pool = SVN::Pool->new_default_sub;
3931
3932        $ra_invalid = 1; # this will open a new SVN::Ra connection to $url1
3933        $ctx->diff([], $url1, $rev1, $url2, $rev2, 1, 1, 0, $out, $out);
3934        $out->flush;
3935        my $ret = (($out->stat)[7] == 0);
3936        close $out or croak $!;
3937
3938        $ret;
3939}
3940
3941sub get_commit_editor {
3942        my ($self, $log, $cb, $pool) = @_;
3943        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
3944        $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
3945}
3946
3947sub gs_do_update {
3948        my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
3949        my $new = ($rev_a == $rev_b);
3950        my $path = $gs->{path};
3951
3952        if ($new && -e $gs->{index}) {
3953                unlink $gs->{index} or die
3954                  "Couldn't unlink index: $gs->{index}: $!\n";
3955        }
3956        my $pool = SVN::Pool->new;
3957        $editor->set_path_strip($path);
3958        my (@pc) = split m#/#, $path;
3959        my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
3960                                        1, $editor, $pool);
3961        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
3962
3963        # Since we can't rely on svn_ra_reparent being available, we'll
3964        # just have to do some magic with set_path to make it so
3965        # we only want a partial path.
3966        my $sp = '';
3967        my $final = join('/', @pc);
3968        while (@pc) {
3969                $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
3970                $sp .= '/' if length $sp;
3971                $sp .= shift @pc;
3972        }
3973        die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
3974
3975        $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
3976
3977        $reporter->finish_report($pool);
3978        $pool->clear;
3979        $editor->{git_commit_ok};
3980}
3981
3982# this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
3983# svn_ra_reparent didn't work before 1.4)
3984sub gs_do_switch {
3985        my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
3986        my $path = $gs->{path};
3987        my $pool = SVN::Pool->new;
3988
3989        my $full_url = $self->{url};
3990        my $old_url = $full_url;
3991        $full_url .= '/' . escape_uri_only($path) if length $path;
3992        my ($ra, $reparented);
3993        if ($old_url ne $full_url) {
3994                if ($old_url !~ m#^svn(\+ssh)?://#) {
3995                        SVN::_Ra::svn_ra_reparent($self->{session}, $full_url,
3996                                                  $pool);
3997                        $self->{url} = $full_url;
3998                        $reparented = 1;
3999                } else {
4000                        $_[0] = undef;
4001                        $self = undef;
4002                        $RA = undef;
4003                        $ra = Git::SVN::Ra->new($full_url);
4004                        $ra_invalid = 1;
4005                }
4006        }
4007        $ra ||= $self;
4008        my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
4009        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
4010        $reporter->set_path('', $rev_a, 0, @lock, $pool);
4011        $reporter->finish_report($pool);
4012
4013        if ($reparented) {
4014                SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
4015                $self->{url} = $old_url;
4016        }
4017
4018        $pool->clear;
4019        $editor->{git_commit_ok};
4020}
4021
4022sub longest_common_path {
4023        my ($gsv, $globs) = @_;
4024        my %common;
4025        my $common_max = scalar @$gsv;
4026
4027        foreach my $gs (@$gsv) {
4028                my @tmp = split m#/#, $gs->{path};
4029                my $p = '';
4030                foreach (@tmp) {
4031                        $p .= length($p) ? "/$_" : $_;
4032                        $common{$p} ||= 0;
4033                        $common{$p}++;
4034                }
4035        }
4036        $globs ||= [];
4037        $common_max += scalar @$globs;
4038        foreach my $glob (@$globs) {
4039                my @tmp = split m#/#, $glob->{path}->{left};
4040                my $p = '';
4041                foreach (@tmp) {
4042                        $p .= length($p) ? "/$_" : $_;
4043                        $common{$p} ||= 0;
4044                        $common{$p}++;
4045                }
4046        }
4047
4048        my $longest_path = '';
4049        foreach (sort {length $b <=> length $a} keys %common) {
4050                if ($common{$_} == $common_max) {
4051                        $longest_path = $_;
4052                        last;
4053                }
4054        }
4055        $longest_path;
4056}
4057
4058sub gs_fetch_loop_common {
4059        my ($self, $base, $head, $gsv, $globs) = @_;
4060        return if ($base > $head);
4061        my $inc = $_log_window_size;
4062        my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
4063        my $longest_path = longest_common_path($gsv, $globs);
4064        my $ra_url = $self->{url};
4065        while (1) {
4066                my %revs;
4067                my $err;
4068                my $err_handler = $SVN::Error::handler;
4069                $SVN::Error::handler = sub {
4070                        ($err) = @_;
4071                        skip_unknown_revs($err);
4072                };
4073                sub _cb {
4074                        my ($paths, $r, $author, $date, $log) = @_;
4075                        [ dup_changed_paths($paths),
4076                          { author => $author, date => $date, log => $log } ];
4077                }
4078                $self->get_log([$longest_path], $min, $max, 0, 1, 1,
4079                               sub { $revs{$_[1]} = _cb(@_) });
4080                if ($err && $max >= $head) {
4081                        print STDERR "Path '$longest_path' ",
4082                                     "was probably deleted:\n",
4083                                     $err->expanded_message,
4084                                     "\nWill attempt to follow ",
4085                                     "revisions r$min .. r$max ",
4086                                     "committed before the deletion\n";
4087                        my $hi = $max;
4088                        while (--$hi >= $min) {
4089                                my $ok;
4090                                $self->get_log([$longest_path], $min, $hi,
4091                                               0, 1, 1, sub {
4092                                               $ok ||= $_[1];
4093                                               $revs{$_[1]} = _cb(@_) });
4094                                if ($ok) {
4095                                        print STDERR "r$min .. r$ok OK\n";
4096                                        last;
4097                                }
4098                        }
4099                }
4100                $SVN::Error::handler = $err_handler;
4101
4102                my %exists = map { $_->{path} => $_ } @$gsv;
4103                foreach my $r (sort {$a <=> $b} keys %revs) {
4104                        my ($paths, $logged) = @{$revs{$r}};
4105
4106                        foreach my $gs ($self->match_globs(\%exists, $paths,
4107                                                           $globs, $r)) {
4108                                if ($gs->rev_map_max >= $r) {
4109                                        next;
4110                                }
4111                                next unless $gs->match_paths($paths, $r);
4112                                $gs->{logged_rev_props} = $logged;
4113                                if (my $last_commit = $gs->last_commit) {
4114                                        $gs->assert_index_clean($last_commit);
4115                                }
4116                                my $log_entry = $gs->do_fetch($paths, $r);
4117                                if ($log_entry) {
4118                                        $gs->do_git_commit($log_entry);
4119                                }
4120                                $INDEX_FILES{$gs->{index}} = 1;
4121                        }
4122                        foreach my $g (@$globs) {
4123                                my $k = "svn-remote.$g->{remote}." .
4124                                        "$g->{t}-maxRev";
4125                                Git::SVN::tmp_config($k, $r);
4126                        }
4127                        if ($ra_invalid) {
4128                                $_[0] = undef;
4129                                $self = undef;
4130                                $RA = undef;
4131                                $self = Git::SVN::Ra->new($ra_url);
4132                                $ra_invalid = undef;
4133                        }
4134                }
4135                # pre-fill the .rev_db since it'll eventually get filled in
4136                # with '0' x40 if something new gets committed
4137                foreach my $gs (@$gsv) {
4138                        next if $gs->rev_map_max >= $max;
4139                        next if defined $gs->rev_map_get($max);
4140                        $gs->rev_map_set($max, 0 x40);
4141                }
4142                foreach my $g (@$globs) {
4143                        my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
4144                        Git::SVN::tmp_config($k, $max);
4145                }
4146                last if $max >= $head;
4147                $min = $max + 1;
4148                $max += $inc;
4149                $max = $head if ($max > $head);
4150        }
4151        Git::SVN::gc();
4152}
4153
4154sub get_dir_globbed {
4155        my ($self, $left, $depth, $r) = @_;
4156
4157        my @x = eval { $self->get_dir($left, $r) };
4158        return unless scalar @x == 3;
4159        my $dirents = $x[0];
4160        my @finalents;
4161        foreach my $de (keys %$dirents) {
4162                next if $dirents->{$de}->{kind} != $SVN::Node::dir;
4163                if ($depth > 1) {
4164                        my @args = ("$left/$de", $depth - 1, $r);
4165                        foreach my $dir ($self->get_dir_globbed(@args)) {
4166                                push @finalents, "$de/$dir";
4167                        }
4168                } else {
4169                        push @finalents, $de;
4170                }
4171        }
4172        @finalents;
4173}
4174
4175sub match_globs {
4176        my ($self, $exists, $paths, $globs, $r) = @_;
4177
4178        sub get_dir_check {
4179                my ($self, $exists, $g, $r) = @_;
4180
4181                my @dirs = $self->get_dir_globbed($g->{path}->{left},
4182                                                  $g->{path}->{depth},
4183                                                  $r);
4184
4185                foreach my $de (@dirs) {
4186                        my $p = $g->{path}->full_path($de);
4187                        next if $exists->{$p};
4188                        next if (length $g->{path}->{right} &&
4189                                 ($self->check_path($p, $r) !=
4190                                  $SVN::Node::dir));
4191                        $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
4192                                         $g->{ref}->full_path($de), 1);
4193                }
4194        }
4195        foreach my $g (@$globs) {
4196                if (my $path = $paths->{"/$g->{path}->{left}"}) {
4197                        if ($path->{action} =~ /^[AR]$/) {
4198                                get_dir_check($self, $exists, $g, $r);
4199                        }
4200                }
4201                foreach (keys %$paths) {
4202                        if (/$g->{path}->{left_regex}/ &&
4203                            !/$g->{path}->{regex}/) {
4204                                next if $paths->{$_}->{action} !~ /^[AR]$/;
4205                                get_dir_check($self, $exists, $g, $r);
4206                        }
4207                        next unless /$g->{path}->{regex}/;
4208                        my $p = $1;
4209                        my $pathname = $g->{path}->full_path($p);
4210                        next if $exists->{$pathname};
4211                        next if ($self->check_path($pathname, $r) !=
4212                                 $SVN::Node::dir);
4213                        $exists->{$pathname} = Git::SVN->init(
4214                                              $self->{url}, $pathname, undef,
4215                                              $g->{ref}->full_path($p), 1);
4216                }
4217                my $c = '';
4218                foreach (split m#/#, $g->{path}->{left}) {
4219                        $c .= "/$_";
4220                        next unless ($paths->{$c} &&
4221                                     ($paths->{$c}->{action} =~ /^[AR]$/));
4222                        get_dir_check($self, $exists, $g, $r);
4223                }
4224        }
4225        values %$exists;
4226}
4227
4228sub minimize_url {
4229        my ($self) = @_;
4230        return $self->{url} if ($self->{url} eq $self->{repos_root});
4231        my $url = $self->{repos_root};
4232        my @components = split(m!/!, $self->{svn_path});
4233        my $c = '';
4234        do {
4235                $url .= "/$c" if length $c;
4236                eval { (ref $self)->new($url)->get_latest_revnum };
4237        } while ($@ && ($c = shift @components));
4238        $url;
4239}
4240
4241sub can_do_switch {
4242        my $self = shift;
4243        unless (defined $can_do_switch) {
4244                my $pool = SVN::Pool->new;
4245                my $rep = eval {
4246                        $self->do_switch(1, '', 0, $self->{url},
4247                                         SVN::Delta::Editor->new, $pool);
4248                };
4249                if ($@) {
4250                        $can_do_switch = 0;
4251                } else {
4252                        $rep->abort_report($pool);
4253                        $can_do_switch = 1;
4254                }
4255                $pool->clear;
4256        }
4257        $can_do_switch;
4258}
4259
4260sub skip_unknown_revs {
4261        my ($err) = @_;
4262        my $errno = $err->apr_err();
4263        # Maybe the branch we're tracking didn't
4264        # exist when the repo started, so it's
4265        # not an error if it doesn't, just continue
4266        #
4267        # Wonderfully consistent library, eh?
4268        # 160013 - svn:// and file://
4269        # 175002 - http(s)://
4270        # 175007 - http(s):// (this repo required authorization, too...)
4271        #   More codes may be discovered later...
4272        if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
4273                my $err_key = $err->expanded_message;
4274                # revision numbers change every time, filter them out
4275                $err_key =~ s/\d+/\0/g;
4276                $err_key = "$errno\0$err_key";
4277                unless ($ignored_err{$err_key}) {
4278                        warn "W: Ignoring error from SVN, path probably ",
4279                             "does not exist: ($errno): ",
4280                             $err->expanded_message,"\n";
4281                        warn "W: Do not be alarmed at the above message ",
4282                             "git-svn is just searching aggressively for ",
4283                             "old history.\n",
4284                             "This may take a while on large repositories\n";
4285                        $ignored_err{$err_key} = 1;
4286                }
4287                return;
4288        }
4289        die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
4290}
4291
4292# svn_log_changed_path_t objects passed to get_log are likely to be
4293# overwritten even if only the refs are copied to an external variable,
4294# so we should dup the structures in their entirety.  Using an externally
4295# passed pool (instead of our temporary and quickly cleared pool in
4296# Git::SVN::Ra) does not help matters at all...
4297sub dup_changed_paths {
4298        my ($paths) = @_;
4299        return undef unless $paths;
4300        my %ret;
4301        foreach my $p (keys %$paths) {
4302                my $i = $paths->{$p};
4303                my %s = map { $_ => $i->$_ }
4304                              qw/copyfrom_path copyfrom_rev action/;
4305                $ret{$p} = \%s;
4306        }
4307        \%ret;
4308}
4309
4310package Git::SVN::Log;
4311use strict;
4312use warnings;
4313use POSIX qw/strftime/;
4314use constant commit_log_separator => ('-' x 72) . "\n";
4315use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
4316            %rusers $show_commit $incremental/;
4317my $l_fmt;
4318
4319sub cmt_showable {
4320        my ($c) = @_;
4321        return 1 if defined $c->{r};
4322
4323        # big commit message got truncated by the 16k pretty buffer in rev-list
4324        if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
4325                                $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
4326                @{$c->{l}} = ();
4327                my @log = command(qw/cat-file commit/, $c->{c});
4328
4329                # shift off the headers
4330                shift @log while ($log[0] ne '');
4331                shift @log;
4332
4333                # TODO: make $c->{l} not have a trailing newline in the future
4334                @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
4335
4336                (undef, $c->{r}, undef) = ::extract_metadata(
4337                                (grep(/^git-svn-id: /, @log))[-1]);
4338        }
4339        return defined $c->{r};
4340}
4341
4342sub log_use_color {
4343        return $color || Git->repository->get_colorbool('color.diff');
4344}
4345
4346sub git_svn_log_cmd {
4347        my ($r_min, $r_max, @args) = @_;
4348        my $head = 'HEAD';
4349        my (@files, @log_opts);
4350        foreach my $x (@args) {
4351                if ($x eq '--' || @files) {
4352                        push @files, $x;
4353                } else {
4354                        if (::verify_ref("$x^0")) {
4355                                $head = $x;
4356                        } else {
4357                                push @log_opts, $x;
4358                        }
4359                }
4360        }
4361
4362        my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
4363        $gs ||= Git::SVN->_new;
4364        my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
4365                   $gs->refname);
4366        push @cmd, '-r' unless $non_recursive;
4367        push @cmd, qw/--raw --name-status/ if $verbose;
4368        push @cmd, '--color' if log_use_color();
4369        push @cmd, @log_opts;
4370        if (defined $r_max && $r_max == $r_min) {
4371                push @cmd, '--max-count=1';
4372                if (my $c = $gs->rev_map_get($r_max)) {
4373                        push @cmd, $c;
4374                }
4375        } elsif (defined $r_max) {
4376                if ($r_max < $r_min) {
4377                        ($r_min, $r_max) = ($r_max, $r_min);
4378                }
4379                my (undef, $c_max) = $gs->find_rev_before($r_max, 1, $r_min);
4380                my (undef, $c_min) = $gs->find_rev_after($r_min, 1, $r_max);
4381                # If there are no commits in the range, both $c_max and $c_min
4382                # will be undefined.  If there is at least 1 commit in the
4383                # range, both will be defined.
4384                return () if !defined $c_min || !defined $c_max;
4385                if ($c_min eq $c_max) {
4386                        push @cmd, '--max-count=1', $c_min;
4387                } else {
4388                        push @cmd, '--boundary', "$c_min..$c_max";
4389                }
4390        }
4391        return (@cmd, @files);
4392}
4393
4394# adapted from pager.c
4395sub config_pager {
4396        $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
4397        if (!defined $pager) {
4398                $pager = 'less';
4399        } elsif (length $pager == 0 || $pager eq 'cat') {
4400                $pager = undef;
4401        }
4402        $ENV{GIT_PAGER_IN_USE} = defined($pager);
4403}
4404
4405sub run_pager {
4406        return unless -t *STDOUT && defined $pager;
4407        pipe my $rfd, my $wfd or return;
4408        defined(my $pid = fork) or ::fatal "Can't fork: $!";
4409        if (!$pid) {
4410                open STDOUT, '>&', $wfd or
4411                                     ::fatal "Can't redirect to stdout: $!";
4412                return;
4413        }
4414        open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!";
4415        $ENV{LESS} ||= 'FRSX';
4416        exec $pager or ::fatal "Can't run pager: $! ($pager)";
4417}
4418
4419sub format_svn_date {
4420        return strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)", localtime(shift));
4421}
4422
4423sub parse_git_date {
4424        my ($t, $tz) = @_;
4425        # Date::Parse isn't in the standard Perl distro :(
4426        if ($tz =~ s/^\+//) {
4427                $t += tz_to_s_offset($tz);
4428        } elsif ($tz =~ s/^\-//) {
4429                $t -= tz_to_s_offset($tz);
4430        }
4431        return $t;
4432}
4433
4434sub set_local_timezone {
4435        if (defined $TZ) {
4436                $ENV{TZ} = $TZ;
4437        } else {
4438                delete $ENV{TZ};
4439        }
4440}
4441
4442sub tz_to_s_offset {
4443        my ($tz) = @_;
4444        $tz =~ s/(\d\d)$//;
4445        return ($1 * 60) + ($tz * 3600);
4446}
4447
4448sub get_author_info {
4449        my ($dest, $author, $t, $tz) = @_;
4450        $author =~ s/(?:^\s*|\s*$)//g;
4451        $dest->{a_raw} = $author;
4452        my $au;
4453        if ($::_authors) {
4454                $au = $rusers{$author} || undef;
4455        }
4456        if (!$au) {
4457                ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
4458        }
4459        $dest->{t} = $t;
4460        $dest->{tz} = $tz;
4461        $dest->{a} = $au;
4462        $dest->{t_utc} = parse_git_date($t, $tz);
4463}
4464
4465sub process_commit {
4466        my ($c, $r_min, $r_max, $defer) = @_;
4467        if (defined $r_min && defined $r_max) {
4468                if ($r_min == $c->{r} && $r_min == $r_max) {
4469                        show_commit($c);
4470                        return 0;
4471                }
4472                return 1 if $r_min == $r_max;
4473                if ($r_min < $r_max) {
4474                        # we need to reverse the print order
4475                        return 0 if (defined $limit && --$limit < 0);
4476                        push @$defer, $c;
4477                        return 1;
4478                }
4479                if ($r_min != $r_max) {
4480                        return 1 if ($r_min < $c->{r});
4481                        return 1 if ($r_max > $c->{r});
4482                }
4483        }
4484        return 0 if (defined $limit && --$limit < 0);
4485        show_commit($c);
4486        return 1;
4487}
4488
4489sub show_commit {
4490        my $c = shift;
4491        if ($oneline) {
4492                my $x = "\n";
4493                if (my $l = $c->{l}) {
4494                        while ($l->[0] =~ /^\s*$/) { shift @$l }
4495                        $x = $l->[0];
4496                }
4497                $l_fmt ||= 'A' . length($c->{r});
4498                print 'r',pack($l_fmt, $c->{r}),' | ';
4499                print "$c->{c} | " if $show_commit;
4500                print $x;
4501        } else {
4502                show_commit_normal($c);
4503        }
4504}
4505
4506sub show_commit_changed_paths {
4507        my ($c) = @_;
4508        return unless $c->{changed};
4509        print "Changed paths:\n", @{$c->{changed}};
4510}
4511
4512sub show_commit_normal {
4513        my ($c) = @_;
4514        print commit_log_separator, "r$c->{r} | ";
4515        print "$c->{c} | " if $show_commit;
4516        print "$c->{a} | ", format_svn_date($c->{t_utc}), ' | ';
4517        my $nr_line = 0;
4518
4519        if (my $l = $c->{l}) {
4520                while ($l->[$#$l] eq "\n" && $#$l > 0
4521                                          && $l->[($#$l - 1)] eq "\n") {
4522                        pop @$l;
4523                }
4524                $nr_line = scalar @$l;
4525                if (!$nr_line) {
4526                        print "1 line\n\n\n";
4527                } else {
4528                        if ($nr_line == 1) {
4529                                $nr_line = '1 line';
4530                        } else {
4531                                $nr_line .= ' lines';
4532                        }
4533                        print $nr_line, "\n";
4534                        show_commit_changed_paths($c);
4535                        print "\n";
4536                        print $_ foreach @$l;
4537                }
4538        } else {
4539                print "1 line\n";
4540                show_commit_changed_paths($c);
4541                print "\n";
4542
4543        }
4544        foreach my $x (qw/raw stat diff/) {
4545                if ($c->{$x}) {
4546                        print "\n";
4547                        print $_ foreach @{$c->{$x}}
4548                }
4549        }
4550}
4551
4552sub cmd_show_log {
4553        my (@args) = @_;
4554        my ($r_min, $r_max);
4555        my $r_last = -1; # prevent dupes
4556        set_local_timezone();
4557        if (defined $::_revision) {
4558                if ($::_revision =~ /^(\d+):(\d+)$/) {
4559                        ($r_min, $r_max) = ($1, $2);
4560                } elsif ($::_revision =~ /^\d+$/) {
4561                        $r_min = $r_max = $::_revision;
4562                } else {
4563                        ::fatal "-r$::_revision is not supported, use ",
4564                                "standard 'git log' arguments instead";
4565                }
4566        }
4567
4568        config_pager();
4569        @args = git_svn_log_cmd($r_min, $r_max, @args);
4570        if (!@args) {
4571                print commit_log_separator unless $incremental || $oneline;
4572                return;
4573        }
4574        my $log = command_output_pipe(@args);
4575        run_pager();
4576        my (@k, $c, $d, $stat);
4577        my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
4578        while (<$log>) {
4579                if (/^${esc_color}commit -?($::sha1_short)/o) {
4580                        my $cmt = $1;
4581                        if ($c && cmt_showable($c) && $c->{r} != $r_last) {
4582                                $r_last = $c->{r};
4583                                process_commit($c, $r_min, $r_max, \@k) or
4584                                                                goto out;
4585                        }
4586                        $d = undef;
4587                        $c = { c => $cmt };
4588                } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
4589                        get_author_info($c, $1, $2, $3);
4590                } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
4591                        # ignore
4592                } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
4593                        push @{$c->{raw}}, $_;
4594                } elsif (/^${esc_color}[ACRMDT]\t/) {
4595                        # we could add $SVN->{svn_path} here, but that requires
4596                        # remote access at the moment (repo_path_split)...
4597                        s#^(${esc_color})([ACRMDT])\t#$1   $2 #o;
4598                        push @{$c->{changed}}, $_;
4599                } elsif (/^${esc_color}diff /o) {
4600                        $d = 1;
4601                        push @{$c->{diff}}, $_;
4602                } elsif ($d) {
4603                        push @{$c->{diff}}, $_;
4604                } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
4605                          $esc_color*[\+\-]*$esc_color$/x) {
4606                        $stat = 1;
4607                        push @{$c->{stat}}, $_;
4608                } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
4609                        push @{$c->{stat}}, $_;
4610                        $stat = undef;
4611                } elsif (/^${esc_color}    (git-svn-id:.+)$/o) {
4612                        ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
4613                } elsif (s/^${esc_color}    //o) {
4614                        push @{$c->{l}}, $_;
4615                }
4616        }
4617        if ($c && defined $c->{r} && $c->{r} != $r_last) {
4618                $r_last = $c->{r};
4619                process_commit($c, $r_min, $r_max, \@k);
4620        }
4621        if (@k) {
4622                ($r_min, $r_max) = ($r_max, $r_min);
4623                process_commit($_, $r_min, $r_max) foreach reverse @k;
4624        }
4625out:
4626        close $log;
4627        print commit_log_separator unless $incremental || $oneline;
4628}
4629
4630sub cmd_blame {
4631        my $path = pop;
4632
4633        config_pager();
4634        run_pager();
4635
4636        my ($fh, $ctx, $rev);
4637
4638        if ($_git_format) {
4639                ($fh, $ctx) = command_output_pipe('blame', @_, $path);
4640                while (my $line = <$fh>) {
4641                        if ($line =~ /^\^?([[:xdigit:]]+)\s/) {
4642                                # Uncommitted edits show up as a rev ID of
4643                                # all zeros, which we can't look up with
4644                                # cmt_metadata
4645                                if ($1 !~ /^0+$/) {
4646                                        (undef, $rev, undef) =
4647                                                ::cmt_metadata($1);
4648                                        $rev = '0' if (!$rev);
4649                                } else {
4650                                        $rev = '0';
4651                                }
4652                                $rev = sprintf('%-10s', $rev);
4653                                $line =~ s/^\^?[[:xdigit:]]+(\s)/$rev$1/;
4654                        }
4655                        print $line;
4656                }
4657        } else {
4658                ($fh, $ctx) = command_output_pipe('blame', '-p', @_, 'HEAD',
4659                                                  '--', $path);
4660                my ($sha1);
4661                my %authors;
4662                while (my $line = <$fh>) {
4663                        if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
4664                                $sha1 = $1;
4665                                (undef, $rev, undef) = ::cmt_metadata($1);
4666                                $rev = '0' if (!$rev);
4667                        }
4668                        elsif ($line =~ /^author (.*)/) {
4669                                $authors{$rev} = $1;
4670                                $authors{$rev} =~ s/\s/_/g;
4671                        }
4672                        elsif ($line =~ /^\t(.*)$/) {
4673                                printf("%6s %10s %s\n", $rev, $authors{$rev}, $1);
4674                        }
4675                }
4676        }
4677        command_close_pipe($fh, $ctx);
4678}
4679
4680package Git::SVN::Migration;
4681# these version numbers do NOT correspond to actual version numbers
4682# of git nor git-svn.  They are just relative.
4683#
4684# v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
4685#
4686# v1 layout: .git/$id/info/url, refs/remotes/$id
4687#
4688# v2 layout: .git/svn/$id/info/url, refs/remotes/$id
4689#
4690# v3 layout: .git/svn/$id, refs/remotes/$id
4691#            - info/url may remain for backwards compatibility
4692#            - this is what we migrate up to this layout automatically,
4693#            - this will be used by git svn init on single branches
4694# v3.1 layout (auto migrated):
4695#            - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
4696#              for backwards compatibility
4697#
4698# v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
4699#            - this is only created for newly multi-init-ed
4700#              repositories.  Similar in spirit to the
4701#              --use-separate-remotes option in git-clone (now default)
4702#            - we do not automatically migrate to this (following
4703#              the example set by core git)
4704#
4705# v5 layout: .rev_db.$UUID => .rev_map.$UUID
4706#            - newer, more-efficient format that uses 24-bytes per record
4707#              with no filler space.
4708#            - use xxd -c24 < .rev_map.$UUID to view and debug
4709#            - This is a one-way migration, repositories updated to the
4710#              new format will not be able to use old git-svn without
4711#              rebuilding the .rev_db.  Rebuilding the rev_db is not
4712#              possible if noMetadata or useSvmProps are set; but should
4713#              be no problem for users that use the (sensible) defaults.
4714use strict;
4715use warnings;
4716use Carp qw/croak/;
4717use File::Path qw/mkpath/;
4718use File::Basename qw/dirname basename/;
4719use vars qw/$_minimize/;
4720
4721sub migrate_from_v0 {
4722        my $git_dir = $ENV{GIT_DIR};
4723        return undef unless -d $git_dir;
4724        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
4725        my $migrated = 0;
4726        while (<$fh>) {
4727                chomp;
4728                my ($id, $orig_ref) = ($_, $_);
4729                next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
4730                next unless -f "$git_dir/$id/info/url";
4731                my $new_ref = "refs/remotes/$id";
4732                if (::verify_ref("$new_ref^0")) {
4733                        print STDERR "W: $orig_ref is probably an old ",
4734                                     "branch used by an ancient version of ",
4735                                     "git-svn.\n",
4736                                     "However, $new_ref also exists.\n",
4737                                     "We will not be able ",
4738                                     "to use this branch until this ",
4739                                     "ambiguity is resolved.\n";
4740                        next;
4741                }
4742                print STDERR "Migrating from v0 layout...\n" if !$migrated;
4743                print STDERR "Renaming ref: $orig_ref => $new_ref\n";
4744                command_noisy('update-ref', $new_ref, $orig_ref);
4745                command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
4746                $migrated++;
4747        }
4748        command_close_pipe($fh, $ctx);
4749        print STDERR "Done migrating from v0 layout...\n" if $migrated;
4750        $migrated;
4751}
4752
4753sub migrate_from_v1 {
4754        my $git_dir = $ENV{GIT_DIR};
4755        my $migrated = 0;
4756        return $migrated unless -d $git_dir;
4757        my $svn_dir = "$git_dir/svn";
4758
4759        # just in case somebody used 'svn' as their $id at some point...
4760        return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
4761
4762        print STDERR "Migrating from a git-svn v1 layout...\n";
4763        mkpath([$svn_dir]);
4764        print STDERR "Data from a previous version of git-svn exists, but\n\t",
4765                     "$svn_dir\n\t(required for this version ",
4766                     "($::VERSION) of git-svn) does not exist.\n";
4767        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
4768        while (<$fh>) {
4769                my $x = $_;
4770                next unless $x =~ s#^refs/remotes/##;
4771                chomp $x;
4772                next unless -f "$git_dir/$x/info/url";
4773                my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
4774                next unless $u;
4775                my $dn = dirname("$git_dir/svn/$x");
4776                mkpath([$dn]) unless -d $dn;
4777                if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
4778                        mkpath(["$git_dir/svn/svn"]);
4779                        print STDERR " - $git_dir/$x/info => ",
4780                                        "$git_dir/svn/$x/info\n";
4781                        rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
4782                               croak "$!: $x";
4783                        # don't worry too much about these, they probably
4784                        # don't exist with repos this old (save for index,
4785                        # and we can easily regenerate that)
4786                        foreach my $f (qw/unhandled.log index .rev_db/) {
4787                                rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
4788                        }
4789                } else {
4790                        print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
4791                        rename "$git_dir/$x", "$git_dir/svn/$x" or
4792                               croak "$!: $x";
4793                }
4794                $migrated++;
4795        }
4796        command_close_pipe($fh, $ctx);
4797        print STDERR "Done migrating from a git-svn v1 layout\n";
4798        $migrated;
4799}
4800
4801sub read_old_urls {
4802        my ($l_map, $pfx, $path) = @_;
4803        my @dir;
4804        foreach (<$path/*>) {
4805                if (-r "$_/info/url") {
4806                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
4807                        my $ref_id = $pfx . basename $_;
4808                        my $url = ::file_to_s("$_/info/url");
4809                        $l_map->{$ref_id} = $url;
4810                } elsif (-d $_) {
4811                        push @dir, $_;
4812                }
4813        }
4814        foreach (@dir) {
4815                my $x = $_;
4816                $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
4817                read_old_urls($l_map, $x, $_);
4818        }
4819}
4820
4821sub migrate_from_v2 {
4822        my @cfg = command(qw/config -l/);
4823        return if grep /^svn-remote\..+\.url=/, @cfg;
4824        my %l_map;
4825        read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
4826        my $migrated = 0;
4827
4828        foreach my $ref_id (sort keys %l_map) {
4829                eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
4830                if ($@) {
4831                        Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
4832                }
4833                $migrated++;
4834        }
4835        $migrated;
4836}
4837
4838sub minimize_connections {
4839        my $r = Git::SVN::read_all_remotes();
4840        my $new_urls = {};
4841        my $root_repos = {};
4842        foreach my $repo_id (keys %$r) {
4843                my $url = $r->{$repo_id}->{url} or next;
4844                my $fetch = $r->{$repo_id}->{fetch} or next;
4845                my $ra = Git::SVN::Ra->new($url);
4846
4847                # skip existing cases where we already connect to the root
4848                if (($ra->{url} eq $ra->{repos_root}) ||
4849                    ($ra->{repos_root} eq $repo_id)) {
4850                        $root_repos->{$ra->{url}} = $repo_id;
4851                        next;
4852                }
4853
4854                my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
4855                my $root_path = $ra->{url};
4856                $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
4857                foreach my $path (keys %$fetch) {
4858                        my $ref_id = $fetch->{$path};
4859                        my $gs = Git::SVN->new($ref_id, $repo_id, $path);
4860
4861                        # make sure we can read when connecting to
4862                        # a higher level of a repository
4863                        my ($last_rev, undef) = $gs->last_rev_commit;
4864                        if (!defined $last_rev) {
4865                                $last_rev = eval {
4866                                        $root_ra->get_latest_revnum;
4867                                };
4868                                next if $@;
4869                        }
4870                        my $new = $root_path;
4871                        $new .= length $path ? "/$path" : '';
4872                        eval {
4873                                $root_ra->get_log([$new], $last_rev, $last_rev,
4874                                                  0, 0, 1, sub { });
4875                        };
4876                        next if $@;
4877                        $new_urls->{$ra->{repos_root}}->{$new} =
4878                                { ref_id => $ref_id,
4879                                  old_repo_id => $repo_id,
4880                                  old_path => $path };
4881                }
4882        }
4883
4884        my @emptied;
4885        foreach my $url (keys %$new_urls) {
4886                # see if we can re-use an existing [svn-remote "repo_id"]
4887                # instead of creating a(n ugly) new section:
4888                my $repo_id = $root_repos->{$url} || $url;
4889
4890                my $fetch = $new_urls->{$url};
4891                foreach my $path (keys %$fetch) {
4892                        my $x = $fetch->{$path};
4893                        Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
4894                        my $pfx = "svn-remote.$x->{old_repo_id}";
4895
4896                        my $old_fetch = quotemeta("$x->{old_path}:".
4897                                                  "refs/remotes/$x->{ref_id}");
4898                        command_noisy(qw/config --unset/,
4899                                      "$pfx.fetch", '^'. $old_fetch . '$');
4900                        delete $r->{$x->{old_repo_id}}->
4901                               {fetch}->{$x->{old_path}};
4902                        if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
4903                                command_noisy(qw/config --unset/,
4904                                              "$pfx.url");
4905                                push @emptied, $x->{old_repo_id}
4906                        }
4907                }
4908        }
4909        if (@emptied) {
4910                my $file = $ENV{GIT_CONFIG} || $ENV{GIT_CONFIG_LOCAL} ||
4911                           "$ENV{GIT_DIR}/config";
4912                print STDERR <<EOF;
4913The following [svn-remote] sections in your config file ($file) are empty
4914and can be safely removed:
4915EOF
4916                print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
4917        }
4918}
4919
4920sub migration_check {
4921        migrate_from_v0();
4922        migrate_from_v1();
4923        migrate_from_v2();
4924        minimize_connections() if $_minimize;
4925}
4926
4927package Git::IndexInfo;
4928use strict;
4929use warnings;
4930use Git qw/command_input_pipe command_close_pipe/;
4931
4932sub new {
4933        my ($class) = @_;
4934        my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
4935        bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
4936}
4937
4938sub remove {
4939        my ($self, $path) = @_;
4940        if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
4941                return ++$self->{nr};
4942        }
4943        undef;
4944}
4945
4946sub update {
4947        my ($self, $mode, $hash, $path) = @_;
4948        if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
4949                return ++$self->{nr};
4950        }
4951        undef;
4952}
4953
4954sub DESTROY {
4955        my ($self) = @_;
4956        command_close_pipe($self->{gui}, $self->{ctx});
4957}
4958
4959package Git::SVN::GlobSpec;
4960use strict;
4961use warnings;
4962
4963sub new {
4964        my ($class, $glob) = @_;
4965        my $re = $glob;
4966        $re =~ s!/+$!!g; # no need for trailing slashes
4967        $re =~ m!^([^*]*)(\*(?:/\*)*)([^*]*)$!;
4968        my $temp = $re;
4969        my ($left, $right) = ($1, $3);
4970        $re = $2;
4971        my $depth = $re =~ tr/*/*/;
4972        if ($depth != $temp =~ tr/*/*/) {
4973                die "Only one set of wildcard directories " .
4974                        "(e.g. '*' or '*/*/*') is supported: '$glob'\n";
4975        }
4976        if ($depth == 0) {
4977                die "One '*' is needed for glob: '$glob'\n";
4978        }
4979        $re =~ s!\*!\[^/\]*!g;
4980        $re = quotemeta($left) . "($re)" . quotemeta($right);
4981        if (length $left && !($left =~ s!/+$!!g)) {
4982                die "Missing trailing '/' on left side of: '$glob' ($left)\n";
4983        }
4984        if (length $right && !($right =~ s!^/+!!g)) {
4985                die "Missing leading '/' on right side of: '$glob' ($right)\n";
4986        }
4987        my $left_re = qr/^\/\Q$left\E(\/|$)/;
4988        bless { left => $left, right => $right, left_regex => $left_re,
4989                regex => qr/$re/, glob => $glob, depth => $depth }, $class;
4990}
4991
4992sub full_path {
4993        my ($self, $path) = @_;
4994        return (length $self->{left} ? "$self->{left}/" : '') .
4995               $path . (length $self->{right} ? "/$self->{right}" : '');
4996}
4997
4998__END__
4999
5000Data structures:
5001
5002
5003$remotes = { # returned by read_all_remotes()
5004        'svn' => {
5005                # svn-remote.svn.url=https://svn.musicpd.org
5006                url => 'https://svn.musicpd.org',
5007                # svn-remote.svn.fetch=mpd/trunk:trunk
5008                fetch => {
5009                        'mpd/trunk' => 'trunk',
5010                },
5011                # svn-remote.svn.tags=mpd/tags/*:tags/*
5012                tags => {
5013                        path => {
5014                                left => 'mpd/tags',
5015                                right => '',
5016                                regex => qr!mpd/tags/([^/]+)$!,
5017                                glob => 'tags/*',
5018                        },
5019                        ref => {
5020                                left => 'tags',
5021                                right => '',
5022                                regex => qr!tags/([^/]+)$!,
5023                                glob => 'tags/*',
5024                        },
5025                }
5026        }
5027};
5028
5029$log_entry hashref as returned by libsvn_log_entry()
5030{
5031        log => 'whitespace-formatted log entry
5032',                                              # trailing newline is preserved
5033        revision => '8',                        # integer
5034        date => '2004-02-24T17:01:44.108345Z',  # commit date
5035        author => 'committer name'
5036};
5037
5038
5039# this is generated by generate_diff();
5040@mods = array of diff-index line hashes, each element represents one line
5041        of diff-index output
5042
5043diff-index line ($m hash)
5044{
5045        mode_a => first column of diff-index output, no leading ':',
5046        mode_b => second column of diff-index output,
5047        sha1_b => sha1sum of the final blob,
5048        chg => change type [MCRADT],
5049        file_a => original file name of a file (iff chg is 'C' or 'R')
5050        file_b => new/current file name of a file (any chg)
5051}
5052;
5053
5054# retval of read_url_paths{,_all}();
5055$l_map = {
5056        # repository root url
5057        'https://svn.musicpd.org' => {
5058                # repository path               # GIT_SVN_ID
5059                'mpd/trunk'             =>      'trunk',
5060                'mpd/tags/0.11.5'       =>      'tags/0.11.5',
5061        },
5062}
5063
5064Notes:
5065        I don't trust the each() function on unless I created %hash myself
5066        because the internal iterator may not have started at base.