git-svn.perlon commit git-svn: respect i18n.commitencoding config (16fc08e)
   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, $_tag);
  70$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        branch => [ \&cmd_branch,
 135                    'Create a branch in the SVN repository',
 136                    { 'message|m=s' => \$_message,
 137                      'dry-run|n' => \$_dry_run,
 138                      'tag|t' => \$_tag } ],
 139        tag => [ sub { $_tag = 1; cmd_branch(@_) },
 140                 'Create a tag in the SVN repository',
 141                 { 'message|m=s' => \$_message,
 142                   'dry-run|n' => \$_dry_run } ],
 143        'set-tree' => [ \&cmd_set_tree,
 144                        "Set an SVN repository to a git tree-ish",
 145                        { 'stdin|' => \$_stdin, %cmt_opts, %fc_opts, } ],
 146        'create-ignore' => [ \&cmd_create_ignore,
 147                             'Create a .gitignore per svn:ignore',
 148                             { 'revision|r=i' => \$_revision
 149                             } ],
 150        'propget' => [ \&cmd_propget,
 151                       'Print the value of a property on a file or directory',
 152                       { 'revision|r=i' => \$_revision } ],
 153        'proplist' => [ \&cmd_proplist,
 154                       'List all properties of a file or directory',
 155                       { 'revision|r=i' => \$_revision } ],
 156        'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
 157                        { 'revision|r=i' => \$_revision
 158                        } ],
 159        'show-externals' => [ \&cmd_show_externals, "Show svn:externals listings",
 160                        { 'revision|r=i' => \$_revision
 161                        } ],
 162        'multi-fetch' => [ \&cmd_multi_fetch,
 163                           "Deprecated alias for $0 fetch --all",
 164                           { 'revision|r=s' => \$_revision, %fc_opts } ],
 165        'migrate' => [ sub { },
 166                       # no-op, we automatically run this anyways,
 167                       'Migrate configuration/metadata/layout from
 168                        previous versions of git-svn',
 169                       { 'minimize' => \$Git::SVN::Migration::_minimize,
 170                         %remote_opts } ],
 171        'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
 172                        { 'limit=i' => \$Git::SVN::Log::limit,
 173                          'revision|r=s' => \$_revision,
 174                          'verbose|v' => \$Git::SVN::Log::verbose,
 175                          'incremental' => \$Git::SVN::Log::incremental,
 176                          'oneline' => \$Git::SVN::Log::oneline,
 177                          'show-commit' => \$Git::SVN::Log::show_commit,
 178                          'non-recursive' => \$Git::SVN::Log::non_recursive,
 179                          'authors-file|A=s' => \$_authors,
 180                          'color' => \$Git::SVN::Log::color,
 181                          'pager=s' => \$Git::SVN::Log::pager
 182                        } ],
 183        'find-rev' => [ \&cmd_find_rev,
 184                        "Translate between SVN revision numbers and tree-ish",
 185                        {} ],
 186        'rebase' => [ \&cmd_rebase, "Fetch and rebase your working directory",
 187                        { 'merge|m|M' => \$_merge,
 188                          'verbose|v' => \$_verbose,
 189                          'strategy|s=s' => \$_strategy,
 190                          'local|l' => \$_local,
 191                          'fetch-all|all' => \$_fetch_all,
 192                          'dry-run|n' => \$_dry_run,
 193                          %fc_opts } ],
 194        'commit-diff' => [ \&cmd_commit_diff,
 195                           'Commit a diff between two trees',
 196                        { 'message|m=s' => \$_message,
 197                          'file|F=s' => \$_file,
 198                          'revision|r=s' => \$_revision,
 199                        %cmt_opts } ],
 200        'info' => [ \&cmd_info,
 201                    "Show info about the latest SVN revision
 202                     on the current branch",
 203                    { 'url' => \$_url, } ],
 204        'blame' => [ \&Git::SVN::Log::cmd_blame,
 205                    "Show what revision and author last modified each line of a file",
 206                    { 'git-format' => \$_git_format } ],
 207);
 208
 209my $cmd;
 210for (my $i = 0; $i < @ARGV; $i++) {
 211        if (defined $cmd{$ARGV[$i]}) {
 212                $cmd = $ARGV[$i];
 213                splice @ARGV, $i, 1;
 214                last;
 215        }
 216};
 217
 218# make sure we're always running at the top-level working directory
 219unless ($cmd && $cmd =~ /(?:clone|init|multi-init)$/) {
 220        unless (-d $ENV{GIT_DIR}) {
 221                if ($git_dir_user_set) {
 222                        die "GIT_DIR=$ENV{GIT_DIR} explicitly set, ",
 223                            "but it is not a directory\n";
 224                }
 225                my $git_dir = delete $ENV{GIT_DIR};
 226                chomp(my $cdup = command_oneline(qw/rev-parse --show-cdup/));
 227                unless (length $cdup) {
 228                        die "Already at toplevel, but $git_dir ",
 229                            "not found '$cdup'\n";
 230                }
 231                chdir $cdup or die "Unable to chdir up to '$cdup'\n";
 232                unless (-d $git_dir) {
 233                        die "$git_dir still not found after going to ",
 234                            "'$cdup'\n";
 235                }
 236                $ENV{GIT_DIR} = $git_dir;
 237        }
 238        $_repository = Git->repository(Repository => $ENV{GIT_DIR});
 239}
 240
 241my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
 242
 243read_repo_config(\%opts);
 244if ($cmd && ($cmd eq 'log' || $cmd eq 'blame')) {
 245        Getopt::Long::Configure('pass_through');
 246}
 247my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
 248                    'minimize-connections' => \$Git::SVN::Migration::_minimize,
 249                    'id|i=s' => \$Git::SVN::default_ref_id,
 250                    'svn-remote|remote|R=s' => sub {
 251                       $Git::SVN::no_reuse_existing = 1;
 252                       $Git::SVN::default_repo_id = $_[1] });
 253exit 1 if (!$rv && $cmd && $cmd ne 'log');
 254
 255usage(0) if $_help;
 256version() if $_version;
 257usage(1) unless defined $cmd;
 258load_authors() if $_authors;
 259
 260unless ($cmd =~ /^(?:clone|init|multi-init|commit-diff)$/) {
 261        Git::SVN::Migration::migration_check();
 262}
 263Git::SVN::init_vars();
 264eval {
 265        Git::SVN::verify_remotes_sanity();
 266        $cmd{$cmd}->[0]->(@ARGV);
 267};
 268fatal $@ if $@;
 269post_fetch_checkout();
 270exit 0;
 271
 272####################### primary functions ######################
 273sub usage {
 274        my $exit = shift || 0;
 275        my $fd = $exit ? \*STDERR : \*STDOUT;
 276        print $fd <<"";
 277git-svn - bidirectional operations between a single Subversion tree and git
 278Usage: git svn <command> [options] [arguments]\n
 279
 280        print $fd "Available commands:\n" unless $cmd;
 281
 282        foreach (sort keys %cmd) {
 283                next if $cmd && $cmd ne $_;
 284                next if /^multi-/; # don't show deprecated commands
 285                print $fd '  ',pack('A17',$_),$cmd{$_}->[1],"\n";
 286                foreach (sort keys %{$cmd{$_}->[2]}) {
 287                        # mixed-case options are for .git/config only
 288                        next if /[A-Z]/ && /^[a-z]+$/i;
 289                        # prints out arguments as they should be passed:
 290                        my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
 291                        print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
 292                                                        "--$_" : "-$_" }
 293                                                split /\|/,$_)," $x\n";
 294                }
 295        }
 296        print $fd <<"";
 297\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
 298arbitrary identifier if you're tracking multiple SVN branches/repositories in
 299one git repository and want to keep them separate.  See git-svn(1) for more
 300information.
 301
 302        exit $exit;
 303}
 304
 305sub version {
 306        print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
 307        exit 0;
 308}
 309
 310sub do_git_init_db {
 311        unless (-d $ENV{GIT_DIR}) {
 312                my @init_db = ('init');
 313                push @init_db, "--template=$_template" if defined $_template;
 314                if (defined $_shared) {
 315                        if ($_shared =~ /[a-z]/) {
 316                                push @init_db, "--shared=$_shared";
 317                        } else {
 318                                push @init_db, "--shared";
 319                        }
 320                }
 321                command_noisy(@init_db);
 322                $_repository = Git->repository(Repository => ".git");
 323        }
 324        my $set;
 325        my $pfx = "svn-remote.$Git::SVN::default_repo_id";
 326        foreach my $i (keys %icv) {
 327                die "'$set' and '$i' cannot both be set\n" if $set;
 328                next unless defined $icv{$i};
 329                command_noisy('config', "$pfx.$i", $icv{$i});
 330                $set = $i;
 331        }
 332}
 333
 334sub init_subdir {
 335        my $repo_path = shift or return;
 336        mkpath([$repo_path]) unless -d $repo_path;
 337        chdir $repo_path or die "Couldn't chdir to $repo_path: $!\n";
 338        $ENV{GIT_DIR} = '.git';
 339        $_repository = Git->repository(Repository => $ENV{GIT_DIR});
 340}
 341
 342sub cmd_clone {
 343        my ($url, $path) = @_;
 344        if (!defined $path &&
 345            (defined $_trunk || defined $_branches || defined $_tags ||
 346             defined $_stdlayout) &&
 347            $url !~ m#^[a-z\+]+://#) {
 348                $path = $url;
 349        }
 350        $path = basename($url) if !defined $path || !length $path;
 351        cmd_init($url, $path);
 352        Git::SVN::fetch_all($Git::SVN::default_repo_id);
 353}
 354
 355sub cmd_init {
 356        if (defined $_stdlayout) {
 357                $_trunk = 'trunk' if (!defined $_trunk);
 358                $_tags = 'tags' if (!defined $_tags);
 359                $_branches = 'branches' if (!defined $_branches);
 360        }
 361        if (defined $_trunk || defined $_branches || defined $_tags) {
 362                return cmd_multi_init(@_);
 363        }
 364        my $url = shift or die "SVN repository location required ",
 365                               "as a command-line argument\n";
 366        init_subdir(@_);
 367        do_git_init_db();
 368
 369        Git::SVN->init($url);
 370}
 371
 372sub cmd_fetch {
 373        if (grep /^\d+=./, @_) {
 374                die "'<rev>=<commit>' fetch arguments are ",
 375                    "no longer supported.\n";
 376        }
 377        my ($remote) = @_;
 378        if (@_ > 1) {
 379                die "Usage: $0 fetch [--all] [svn-remote]\n";
 380        }
 381        $remote ||= $Git::SVN::default_repo_id;
 382        if ($_fetch_all) {
 383                cmd_multi_fetch();
 384        } else {
 385                Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
 386        }
 387}
 388
 389sub cmd_set_tree {
 390        my (@commits) = @_;
 391        if ($_stdin || !@commits) {
 392                print "Reading from stdin...\n";
 393                @commits = ();
 394                while (<STDIN>) {
 395                        if (/\b($sha1_short)\b/o) {
 396                                unshift @commits, $1;
 397                        }
 398                }
 399        }
 400        my @revs;
 401        foreach my $c (@commits) {
 402                my @tmp = command('rev-parse',$c);
 403                if (scalar @tmp == 1) {
 404                        push @revs, $tmp[0];
 405                } elsif (scalar @tmp > 1) {
 406                        push @revs, reverse(command('rev-list',@tmp));
 407                } else {
 408                        fatal "Failed to rev-parse $c";
 409                }
 410        }
 411        my $gs = Git::SVN->new;
 412        my ($r_last, $cmt_last) = $gs->last_rev_commit;
 413        $gs->fetch;
 414        if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
 415                fatal "There are new revisions that were fetched ",
 416                      "and need to be merged (or acknowledged) ",
 417                      "before committing.\nlast rev: $r_last\n",
 418                      " current: $gs->{last_rev}";
 419        }
 420        $gs->set_tree($_) foreach @revs;
 421        print "Done committing ",scalar @revs," revisions to SVN\n";
 422        unlink $gs->{index};
 423}
 424
 425sub cmd_dcommit {
 426        my $head = shift;
 427        git_cmd_try { command_oneline(qw/diff-index --quiet HEAD/) }
 428                'Cannot dcommit with a dirty index.  Commit your changes first, '
 429                . "or stash them with `git stash'.\n";
 430        $head ||= 'HEAD';
 431        my @refs;
 432        my ($url, $rev, $uuid, $gs) = working_head_info($head, \@refs);
 433        unless ($gs) {
 434                die "Unable to determine upstream SVN information from ",
 435                    "$head history.\nPerhaps the repository is empty.";
 436        }
 437        $url = defined $_commit_url ? $_commit_url : $gs->full_url;
 438        my $last_rev = $_revision if defined $_revision;
 439        if ($url) {
 440                print "Committing to $url ...\n";
 441        }
 442        my ($linear_refs, $parents) = linearize_history($gs, \@refs);
 443        if ($_no_rebase && scalar(@$linear_refs) > 1) {
 444                warn "Attempting to commit more than one change while ",
 445                     "--no-rebase is enabled.\n",
 446                     "If these changes depend on each other, re-running ",
 447                     "without --no-rebase may be required."
 448        }
 449        my $expect_url = $url;
 450        Git::SVN::remove_username($expect_url);
 451        while (1) {
 452                my $d = shift @$linear_refs or last;
 453                unless (defined $last_rev) {
 454                        (undef, $last_rev, undef) = cmt_metadata("$d~1");
 455                        unless (defined $last_rev) {
 456                                fatal "Unable to extract revision information ",
 457                                      "from commit $d~1";
 458                        }
 459                }
 460                if ($_dry_run) {
 461                        print "diff-tree $d~1 $d\n";
 462                } else {
 463                        my $cmt_rev;
 464                        my %ed_opts = ( r => $last_rev,
 465                                        log => get_commit_entry($d)->{log},
 466                                        ra => Git::SVN::Ra->new($url),
 467                                        config => SVN::Core::config_get_config(
 468                                                $Git::SVN::Ra::config_dir
 469                                        ),
 470                                        tree_a => "$d~1",
 471                                        tree_b => $d,
 472                                        editor_cb => sub {
 473                                               print "Committed r$_[0]\n";
 474                                               $cmt_rev = $_[0];
 475                                        },
 476                                        svn_path => '');
 477                        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
 478                                print "No changes\n$d~1 == $d\n";
 479                        } elsif ($parents->{$d} && @{$parents->{$d}}) {
 480                                $gs->{inject_parents_dcommit}->{$cmt_rev} =
 481                                                               $parents->{$d};
 482                        }
 483                        $_fetch_all ? $gs->fetch_all : $gs->fetch;
 484                        $last_rev = $cmt_rev;
 485                        next if $_no_rebase;
 486
 487                        # we always want to rebase against the current HEAD,
 488                        # not any head that was passed to us
 489                        my @diff = command('diff-tree', $d,
 490                                           $gs->refname, '--');
 491                        my @finish;
 492                        if (@diff) {
 493                                @finish = rebase_cmd();
 494                                print STDERR "W: $d and ", $gs->refname,
 495                                             " differ, using @finish:\n",
 496                                             join("\n", @diff), "\n";
 497                        } else {
 498                                print "No changes between current HEAD and ",
 499                                      $gs->refname,
 500                                      "\nResetting to the latest ",
 501                                      $gs->refname, "\n";
 502                                @finish = qw/reset --mixed/;
 503                        }
 504                        command_noisy(@finish, $gs->refname);
 505                        if (@diff) {
 506                                @refs = ();
 507                                my ($url_, $rev_, $uuid_, $gs_) =
 508                                              working_head_info($head, \@refs);
 509                                my ($linear_refs_, $parents_) =
 510                                              linearize_history($gs_, \@refs);
 511                                if (scalar(@$linear_refs) !=
 512                                    scalar(@$linear_refs_)) {
 513                                        fatal "# of revisions changed ",
 514                                          "\nbefore:\n",
 515                                          join("\n", @$linear_refs),
 516                                          "\n\nafter:\n",
 517                                          join("\n", @$linear_refs_), "\n",
 518                                          'If you are attempting to commit ',
 519                                          "merges, try running:\n\t",
 520                                          'git rebase --interactive',
 521                                          '--preserve-merges ',
 522                                          $gs->refname,
 523                                          "\nBefore dcommitting";
 524                                }
 525                                if ($url_ ne $expect_url) {
 526                                        fatal "URL mismatch after rebase: ",
 527                                              "$url_ != $expect_url";
 528                                }
 529                                if ($uuid_ ne $uuid) {
 530                                        fatal "uuid mismatch after rebase: ",
 531                                              "$uuid_ != $uuid";
 532                                }
 533                                # remap parents
 534                                my (%p, @l, $i);
 535                                for ($i = 0; $i < scalar @$linear_refs; $i++) {
 536                                        my $new = $linear_refs_->[$i] or next;
 537                                        $p{$new} =
 538                                                $parents->{$linear_refs->[$i]};
 539                                        push @l, $new;
 540                                }
 541                                $parents = \%p;
 542                                $linear_refs = \@l;
 543                        }
 544                }
 545        }
 546        unlink $gs->{index};
 547}
 548
 549sub cmd_branch {
 550        my ($branch_name, $head) = @_;
 551
 552        unless (defined $branch_name && length $branch_name) {
 553                die(($_tag ? "tag" : "branch") . " name required\n");
 554        }
 555        $head ||= 'HEAD';
 556
 557        my ($src, $rev, undef, $gs) = working_head_info($head);
 558
 559        my $remote = Git::SVN::read_all_remotes()->{svn};
 560        my $glob = $remote->{ $_tag ? 'tags' : 'branches' };
 561        my ($lft, $rgt) = @{ $glob->{path} }{qw/left right/};
 562        my $dst = join '/', $remote->{url}, $lft, $branch_name, ($rgt || ());
 563
 564        my $ctx = SVN::Client->new(
 565                auth    => Git::SVN::Ra::_auth_providers(),
 566                log_msg => sub {
 567                        ${ $_[0] } = defined $_message
 568                                ? $_message
 569                                : 'Create ' . ($_tag ? 'tag ' : 'branch ' )
 570                                . $branch_name;
 571                },
 572        );
 573
 574        eval {
 575                $ctx->ls($dst, 'HEAD', 0);
 576        } and die "branch ${branch_name} already exists\n";
 577
 578        print "Copying ${src} at r${rev} to ${dst}...\n";
 579        $ctx->copy($src, $rev, $dst)
 580                unless $_dry_run;
 581
 582        $gs->fetch_all;
 583}
 584
 585sub cmd_find_rev {
 586        my $revision_or_hash = shift or die "SVN or git revision required ",
 587                                            "as a command-line argument\n";
 588        my $result;
 589        if ($revision_or_hash =~ /^r\d+$/) {
 590                my $head = shift;
 591                $head ||= 'HEAD';
 592                my @refs;
 593                my (undef, undef, $uuid, $gs) = working_head_info($head, \@refs);
 594                unless ($gs) {
 595                        die "Unable to determine upstream SVN information from ",
 596                            "$head history\n";
 597                }
 598                my $desired_revision = substr($revision_or_hash, 1);
 599                $result = $gs->rev_map_get($desired_revision, $uuid);
 600        } else {
 601                my (undef, $rev, undef) = cmt_metadata($revision_or_hash);
 602                $result = $rev;
 603        }
 604        print "$result\n" if $result;
 605}
 606
 607sub cmd_rebase {
 608        command_noisy(qw/update-index --refresh/);
 609        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 610        unless ($gs) {
 611                die "Unable to determine upstream SVN information from ",
 612                    "working tree history\n";
 613        }
 614        if ($_dry_run) {
 615                print "Remote Branch: " . $gs->refname . "\n";
 616                print "SVN URL: " . $url . "\n";
 617                return;
 618        }
 619        if (command(qw/diff-index HEAD --/)) {
 620                print STDERR "Cannot rebase with uncommited changes:\n";
 621                command_noisy('status');
 622                exit 1;
 623        }
 624        unless ($_local) {
 625                # rebase will checkout for us, so no need to do it explicitly
 626                $_no_checkout = 'true';
 627                $_fetch_all ? $gs->fetch_all : $gs->fetch;
 628        }
 629        command_noisy(rebase_cmd(), $gs->refname);
 630}
 631
 632sub cmd_show_ignore {
 633        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 634        $gs ||= Git::SVN->new;
 635        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 636        $gs->prop_walk($gs->{path}, $r, sub {
 637                my ($gs, $path, $props) = @_;
 638                print STDOUT "\n# $path\n";
 639                my $s = $props->{'svn:ignore'} or return;
 640                $s =~ s/[\r\n]+/\n/g;
 641                chomp $s;
 642                $s =~ s#^#$path#gm;
 643                print STDOUT "$s\n";
 644        });
 645}
 646
 647sub cmd_show_externals {
 648        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 649        $gs ||= Git::SVN->new;
 650        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 651        $gs->prop_walk($gs->{path}, $r, sub {
 652                my ($gs, $path, $props) = @_;
 653                print STDOUT "\n# $path\n";
 654                my $s = $props->{'svn:externals'} or return;
 655                $s =~ s/[\r\n]+/\n/g;
 656                chomp $s;
 657                $s =~ s#^#$path#gm;
 658                print STDOUT "$s\n";
 659        });
 660}
 661
 662sub cmd_create_ignore {
 663        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 664        $gs ||= Git::SVN->new;
 665        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 666        $gs->prop_walk($gs->{path}, $r, sub {
 667                my ($gs, $path, $props) = @_;
 668                # $path is of the form /path/to/dir/
 669                my $ignore = '.' . $path . '.gitignore';
 670                my $s = $props->{'svn:ignore'} or return;
 671                open(GITIGNORE, '>', $ignore)
 672                  or fatal("Failed to open `$ignore' for writing: $!");
 673                $s =~ s/[\r\n]+/\n/g;
 674                chomp $s;
 675                # Prefix all patterns so that the ignore doesn't apply
 676                # to sub-directories.
 677                $s =~ s#^#/#gm;
 678                print GITIGNORE "$s\n";
 679                close(GITIGNORE)
 680                  or fatal("Failed to close `$ignore': $!");
 681                command_noisy('add', '-f', $ignore);
 682        });
 683}
 684
 685sub canonicalize_path {
 686        my ($path) = @_;
 687        my $dot_slash_added = 0;
 688        if (substr($path, 0, 1) ne "/") {
 689                $path = "./" . $path;
 690                $dot_slash_added = 1;
 691        }
 692        # File::Spec->canonpath doesn't collapse x/../y into y (for a
 693        # good reason), so let's do this manually.
 694        $path =~ s#/+#/#g;
 695        $path =~ s#/\.(?:/|$)#/#g;
 696        $path =~ s#/[^/]+/\.\.##g;
 697        $path =~ s#/$##g;
 698        $path =~ s#^\./## if $dot_slash_added;
 699        $path =~ s#^/##;
 700        $path =~ s#^\.$##;
 701        return $path;
 702}
 703
 704# get_svnprops(PATH)
 705# ------------------
 706# Helper for cmd_propget and cmd_proplist below.
 707sub get_svnprops {
 708        my $path = shift;
 709        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 710        $gs ||= Git::SVN->new;
 711
 712        # prefix THE PATH by the sub-directory from which the user
 713        # invoked us.
 714        $path = $cmd_dir_prefix . $path;
 715        fatal("No such file or directory: $path") unless -e $path;
 716        my $is_dir = -d $path ? 1 : 0;
 717        $path = $gs->{path} . '/' . $path;
 718
 719        # canonicalize the path (otherwise libsvn will abort or fail to
 720        # find the file)
 721        $path = canonicalize_path($path);
 722
 723        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 724        my $props;
 725        if ($is_dir) {
 726                (undef, undef, $props) = $gs->ra->get_dir($path, $r);
 727        }
 728        else {
 729                (undef, $props) = $gs->ra->get_file($path, $r, undef);
 730        }
 731        return $props;
 732}
 733
 734# cmd_propget (PROP, PATH)
 735# ------------------------
 736# Print the SVN property PROP for PATH.
 737sub cmd_propget {
 738        my ($prop, $path) = @_;
 739        $path = '.' if not defined $path;
 740        usage(1) if not defined $prop;
 741        my $props = get_svnprops($path);
 742        if (not defined $props->{$prop}) {
 743                fatal("`$path' does not have a `$prop' SVN property.");
 744        }
 745        print $props->{$prop} . "\n";
 746}
 747
 748# cmd_proplist (PATH)
 749# -------------------
 750# Print the list of SVN properties for PATH.
 751sub cmd_proplist {
 752        my $path = shift;
 753        $path = '.' if not defined $path;
 754        my $props = get_svnprops($path);
 755        print "Properties on '$path':\n";
 756        foreach (sort keys %{$props}) {
 757                print "  $_\n";
 758        }
 759}
 760
 761sub cmd_multi_init {
 762        my $url = shift;
 763        unless (defined $_trunk || defined $_branches || defined $_tags) {
 764                usage(1);
 765        }
 766
 767        # there are currently some bugs that prevent multi-init/multi-fetch
 768        # setups from working well without this.
 769        $Git::SVN::_minimize_url = 1;
 770
 771        $_prefix = '' unless defined $_prefix;
 772        if (defined $url) {
 773                $url =~ s#/+$##;
 774                init_subdir(@_);
 775        }
 776        do_git_init_db();
 777        if (defined $_trunk) {
 778                my $trunk_ref = $_prefix . 'trunk';
 779                # try both old-style and new-style lookups:
 780                my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
 781                unless ($gs_trunk) {
 782                        my ($trunk_url, $trunk_path) =
 783                                              complete_svn_url($url, $_trunk);
 784                        $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
 785                                                   undef, $trunk_ref);
 786                }
 787        }
 788        return unless defined $_branches || defined $_tags;
 789        my $ra = $url ? Git::SVN::Ra->new($url) : undef;
 790        complete_url_ls_init($ra, $_branches, '--branches/-b', $_prefix);
 791        complete_url_ls_init($ra, $_tags, '--tags/-t', $_prefix . 'tags/');
 792}
 793
 794sub cmd_multi_fetch {
 795        my $remotes = Git::SVN::read_all_remotes();
 796        foreach my $repo_id (sort keys %$remotes) {
 797                if ($remotes->{$repo_id}->{url}) {
 798                        Git::SVN::fetch_all($repo_id, $remotes);
 799                }
 800        }
 801}
 802
 803# this command is special because it requires no metadata
 804sub cmd_commit_diff {
 805        my ($ta, $tb, $url) = @_;
 806        my $usage = "Usage: $0 commit-diff -r<revision> ".
 807                    "<tree-ish> <tree-ish> [<URL>]";
 808        fatal($usage) if (!defined $ta || !defined $tb);
 809        my $svn_path = '';
 810        if (!defined $url) {
 811                my $gs = eval { Git::SVN->new };
 812                if (!$gs) {
 813                        fatal("Needed URL or usable git-svn --id in ",
 814                              "the command-line\n", $usage);
 815                }
 816                $url = $gs->{url};
 817                $svn_path = $gs->{path};
 818        }
 819        unless (defined $_revision) {
 820                fatal("-r|--revision is a required argument\n", $usage);
 821        }
 822        if (defined $_message && defined $_file) {
 823                fatal("Both --message/-m and --file/-F specified ",
 824                      "for the commit message.\n",
 825                      "I have no idea what you mean");
 826        }
 827        if (defined $_file) {
 828                $_message = file_to_s($_file);
 829        } else {
 830                $_message ||= get_commit_entry($tb)->{log};
 831        }
 832        my $ra ||= Git::SVN::Ra->new($url);
 833        my $r = $_revision;
 834        if ($r eq 'HEAD') {
 835                $r = $ra->get_latest_revnum;
 836        } elsif ($r !~ /^\d+$/) {
 837                die "revision argument: $r not understood by git-svn\n";
 838        }
 839        my %ed_opts = ( r => $r,
 840                        log => $_message,
 841                        ra => $ra,
 842                        tree_a => $ta,
 843                        tree_b => $tb,
 844                        editor_cb => sub { print "Committed r$_[0]\n" },
 845                        svn_path => $svn_path );
 846        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
 847                print "No changes\n$ta == $tb\n";
 848        }
 849}
 850
 851sub escape_uri_only {
 852        my ($uri) = @_;
 853        my @tmp;
 854        foreach (split m{/}, $uri) {
 855                s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
 856                push @tmp, $_;
 857        }
 858        join('/', @tmp);
 859}
 860
 861sub escape_url {
 862        my ($url) = @_;
 863        if ($url =~ m#^([^:]+)://([^/]*)(.*)$#) {
 864                my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
 865                $url = "$scheme://$domain$uri";
 866        }
 867        $url;
 868}
 869
 870sub cmd_info {
 871        my $path = canonicalize_path(defined($_[0]) ? $_[0] : ".");
 872        my $fullpath = canonicalize_path($cmd_dir_prefix . $path);
 873        if (exists $_[1]) {
 874                die "Too many arguments specified\n";
 875        }
 876
 877        my ($file_type, $diff_status) = find_file_type_and_diff_status($path);
 878
 879        if (!$file_type && !$diff_status) {
 880                print STDERR "svn: '$path' is not under version control\n";
 881                exit 1;
 882        }
 883
 884        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 885        unless ($gs) {
 886                die "Unable to determine upstream SVN information from ",
 887                    "working tree history\n";
 888        }
 889
 890        # canonicalize_path() will return "" to make libsvn 1.5.x happy,
 891        $path = "." if $path eq "";
 892
 893        my $full_url = $url . ($fullpath eq "" ? "" : "/$fullpath");
 894
 895        if ($_url) {
 896                print escape_url($full_url), "\n";
 897                return;
 898        }
 899
 900        my $result = "Path: $path\n";
 901        $result .= "Name: " . basename($path) . "\n" if $file_type ne "dir";
 902        $result .= "URL: " . escape_url($full_url) . "\n";
 903
 904        eval {
 905                my $repos_root = $gs->repos_root;
 906                Git::SVN::remove_username($repos_root);
 907                $result .= "Repository Root: " . escape_url($repos_root) . "\n";
 908        };
 909        if ($@) {
 910                $result .= "Repository Root: (offline)\n";
 911        }
 912        $result .= "Repository UUID: $uuid\n" unless $diff_status eq "A";
 913        $result .= "Revision: " . ($diff_status eq "A" ? 0 : $rev) . "\n";
 914
 915        $result .= "Node Kind: " .
 916                   ($file_type eq "dir" ? "directory" : "file") . "\n";
 917
 918        my $schedule = $diff_status eq "A"
 919                       ? "add"
 920                       : ($diff_status eq "D" ? "delete" : "normal");
 921        $result .= "Schedule: $schedule\n";
 922
 923        if ($diff_status eq "A") {
 924                print $result, "\n";
 925                return;
 926        }
 927
 928        my ($lc_author, $lc_rev, $lc_date_utc);
 929        my @args = Git::SVN::Log::git_svn_log_cmd($rev, $rev, "--", $fullpath);
 930        my $log = command_output_pipe(@args);
 931        my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
 932        while (<$log>) {
 933                if (/^${esc_color}author (.+) <[^>]+> (\d+) ([\-\+]?\d+)$/o) {
 934                        $lc_author = $1;
 935                        $lc_date_utc = Git::SVN::Log::parse_git_date($2, $3);
 936                } elsif (/^${esc_color}    (git-svn-id:.+)$/o) {
 937                        (undef, $lc_rev, undef) = ::extract_metadata($1);
 938                }
 939        }
 940        close $log;
 941
 942        Git::SVN::Log::set_local_timezone();
 943
 944        $result .= "Last Changed Author: $lc_author\n";
 945        $result .= "Last Changed Rev: $lc_rev\n";
 946        $result .= "Last Changed Date: " .
 947                   Git::SVN::Log::format_svn_date($lc_date_utc) . "\n";
 948
 949        if ($file_type ne "dir") {
 950                my $text_last_updated_date =
 951                    ($diff_status eq "D" ? $lc_date_utc : (stat $path)[9]);
 952                $result .=
 953                    "Text Last Updated: " .
 954                    Git::SVN::Log::format_svn_date($text_last_updated_date) .
 955                    "\n";
 956                my $checksum;
 957                if ($diff_status eq "D") {
 958                        my ($fh, $ctx) =
 959                            command_output_pipe(qw(cat-file blob), "HEAD:$path");
 960                        if ($file_type eq "link") {
 961                                my $file_name = <$fh>;
 962                                $checksum = md5sum("link $file_name");
 963                        } else {
 964                                $checksum = md5sum($fh);
 965                        }
 966                        command_close_pipe($fh, $ctx);
 967                } elsif ($file_type eq "link") {
 968                        my $file_name =
 969                            command(qw(cat-file blob), "HEAD:$path");
 970                        $checksum =
 971                            md5sum("link " . $file_name);
 972                } else {
 973                        open FILE, "<", $path or die $!;
 974                        $checksum = md5sum(\*FILE);
 975                        close FILE or die $!;
 976                }
 977                $result .= "Checksum: " . $checksum . "\n";
 978        }
 979
 980        print $result, "\n";
 981}
 982
 983########################### utility functions #########################
 984
 985sub rebase_cmd {
 986        my @cmd = qw/rebase/;
 987        push @cmd, '-v' if $_verbose;
 988        push @cmd, qw/--merge/ if $_merge;
 989        push @cmd, "--strategy=$_strategy" if $_strategy;
 990        @cmd;
 991}
 992
 993sub post_fetch_checkout {
 994        return if $_no_checkout;
 995        my $gs = $Git::SVN::_head or return;
 996        return if verify_ref('refs/heads/master^0');
 997
 998        my $valid_head = verify_ref('HEAD^0');
 999        command_noisy(qw(update-ref refs/heads/master), $gs->refname);
1000        return if ($valid_head || !verify_ref('HEAD^0'));
1001
1002        return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
1003        my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
1004        return if -f $index;
1005
1006        return if command_oneline(qw/rev-parse --is-inside-work-tree/) eq 'false';
1007        return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
1008        command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
1009        print STDERR "Checked out HEAD:\n  ",
1010                     $gs->full_url, " r", $gs->last_rev, "\n";
1011}
1012
1013sub complete_svn_url {
1014        my ($url, $path) = @_;
1015        $path =~ s#/+$##;
1016        if ($path !~ m#^[a-z\+]+://#) {
1017                if (!defined $url || $url !~ m#^[a-z\+]+://#) {
1018                        fatal("E: '$path' is not a complete URL ",
1019                              "and a separate URL is not specified");
1020                }
1021                return ($url, $path);
1022        }
1023        return ($path, '');
1024}
1025
1026sub complete_url_ls_init {
1027        my ($ra, $repo_path, $switch, $pfx) = @_;
1028        unless ($repo_path) {
1029                print STDERR "W: $switch not specified\n";
1030                return;
1031        }
1032        $repo_path =~ s#/+$##;
1033        if ($repo_path =~ m#^[a-z\+]+://#) {
1034                $ra = Git::SVN::Ra->new($repo_path);
1035                $repo_path = '';
1036        } else {
1037                $repo_path =~ s#^/+##;
1038                unless ($ra) {
1039                        fatal("E: '$repo_path' is not a complete URL ",
1040                              "and a separate URL is not specified");
1041                }
1042        }
1043        my $url = $ra->{url};
1044        my $gs = Git::SVN->init($url, undef, undef, undef, 1);
1045        my $k = "svn-remote.$gs->{repo_id}.url";
1046        my $orig_url = eval { command_oneline(qw/config --get/, $k) };
1047        if ($orig_url && ($orig_url ne $gs->{url})) {
1048                die "$k already set: $orig_url\n",
1049                    "wanted to set to: $gs->{url}\n";
1050        }
1051        command_oneline('config', $k, $gs->{url}) unless $orig_url;
1052        my $remote_path = "$ra->{svn_path}/$repo_path";
1053        $remote_path =~ s#/+#/#g;
1054        $remote_path =~ s#^/##g;
1055        $remote_path .= "/*" if $remote_path !~ /\*/;
1056        my ($n) = ($switch =~ /^--(\w+)/);
1057        if (length $pfx && $pfx !~ m#/$#) {
1058                die "--prefix='$pfx' must have a trailing slash '/'\n";
1059        }
1060        command_noisy('config',
1061                      "svn-remote.$gs->{repo_id}.$n",
1062                      "$remote_path:refs/remotes/$pfx*" .
1063                        ('/*' x (($remote_path =~ tr/*/*/) - 1)) );
1064}
1065
1066sub verify_ref {
1067        my ($ref) = @_;
1068        eval { command_oneline([ 'rev-parse', '--verify', $ref ],
1069                               { STDERR => 0 }); };
1070}
1071
1072sub get_tree_from_treeish {
1073        my ($treeish) = @_;
1074        # $treeish can be a symbolic ref, too:
1075        my $type = command_oneline(qw/cat-file -t/, $treeish);
1076        my $expected;
1077        while ($type eq 'tag') {
1078                ($treeish, $type) = command(qw/cat-file tag/, $treeish);
1079        }
1080        if ($type eq 'commit') {
1081                $expected = (grep /^tree /, command(qw/cat-file commit/,
1082                                                    $treeish))[0];
1083                ($expected) = ($expected =~ /^tree ($sha1)$/o);
1084                die "Unable to get tree from $treeish\n" unless $expected;
1085        } elsif ($type eq 'tree') {
1086                $expected = $treeish;
1087        } else {
1088                die "$treeish is a $type, expected tree, tag or commit\n";
1089        }
1090        return $expected;
1091}
1092
1093sub get_commit_entry {
1094        my ($treeish) = shift;
1095        my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
1096        my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
1097        my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
1098        open my $log_fh, '>', $commit_editmsg or croak $!;
1099
1100        my $type = command_oneline(qw/cat-file -t/, $treeish);
1101        if ($type eq 'commit' || $type eq 'tag') {
1102                my ($msg_fh, $ctx) = command_output_pipe('cat-file',
1103                                                         $type, $treeish);
1104                my $in_msg = 0;
1105                my $author;
1106                my $saw_from = 0;
1107                my $msgbuf = "";
1108                while (<$msg_fh>) {
1109                        if (!$in_msg) {
1110                                $in_msg = 1 if (/^\s*$/);
1111                                $author = $1 if (/^author (.*>)/);
1112                        } elsif (/^git-svn-id: /) {
1113                                # skip this for now, we regenerate the
1114                                # correct one on re-fetch anyways
1115                                # TODO: set *:merge properties or like...
1116                        } else {
1117                                if (/^From:/ || /^Signed-off-by:/) {
1118                                        $saw_from = 1;
1119                                }
1120                                $msgbuf .= $_;
1121                        }
1122                }
1123                $msgbuf =~ s/\s+$//s;
1124                if ($Git::SVN::_add_author_from && defined($author)
1125                    && !$saw_from) {
1126                        $msgbuf .= "\n\nFrom: $author";
1127                }
1128                print $log_fh $msgbuf or croak $!;
1129                command_close_pipe($msg_fh, $ctx);
1130        }
1131        close $log_fh or croak $!;
1132
1133        if ($_edit || ($type eq 'tree')) {
1134                my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
1135                # TODO: strip out spaces, comments, like git-commit.sh
1136                system($editor, $commit_editmsg);
1137        }
1138        rename $commit_editmsg, $commit_msg or croak $!;
1139        {
1140                # SVN requires messages to be UTF-8 when entering the repo
1141                local $/;
1142                open $log_fh, '<', $commit_msg or croak $!;
1143                binmode $log_fh;
1144                chomp($log_entry{log} = <$log_fh>);
1145
1146                if (my $enc = Git::config('i18n.commitencoding')) {
1147                        require Encode;
1148                        Encode::from_to($log_entry{log}, $enc, 'UTF-8');
1149                }
1150                close $log_fh or croak $!;
1151        }
1152        unlink $commit_msg;
1153        \%log_entry;
1154}
1155
1156sub s_to_file {
1157        my ($str, $file, $mode) = @_;
1158        open my $fd,'>',$file or croak $!;
1159        print $fd $str,"\n" or croak $!;
1160        close $fd or croak $!;
1161        chmod ($mode &~ umask, $file) if (defined $mode);
1162}
1163
1164sub file_to_s {
1165        my $file = shift;
1166        open my $fd,'<',$file or croak "$!: file: $file\n";
1167        local $/;
1168        my $ret = <$fd>;
1169        close $fd or croak $!;
1170        $ret =~ s/\s*$//s;
1171        return $ret;
1172}
1173
1174# '<svn username> = real-name <email address>' mapping based on git-svnimport:
1175sub load_authors {
1176        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1177        my $log = $cmd eq 'log';
1178        while (<$authors>) {
1179                chomp;
1180                next unless /^(.+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
1181                my ($user, $name, $email) = ($1, $2, $3);
1182                if ($log) {
1183                        $Git::SVN::Log::rusers{"$name <$email>"} = $user;
1184                } else {
1185                        $users{$user} = [$name, $email];
1186                }
1187        }
1188        close $authors or croak $!;
1189}
1190
1191# convert GetOpt::Long specs for use by git-config
1192sub read_repo_config {
1193        return unless -d $ENV{GIT_DIR};
1194        my $opts = shift;
1195        my @config_only;
1196        foreach my $o (keys %$opts) {
1197                # if we have mixedCase and a long option-only, then
1198                # it's a config-only variable that we don't need for
1199                # the command-line.
1200                push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
1201                my $v = $opts->{$o};
1202                my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
1203                $key =~ s/-//g;
1204                my $arg = 'git config';
1205                $arg .= ' --int' if ($o =~ /[:=]i$/);
1206                $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
1207                if (ref $v eq 'ARRAY') {
1208                        chomp(my @tmp = `$arg --get-all svn.$key`);
1209                        @$v = @tmp if @tmp;
1210                } else {
1211                        chomp(my $tmp = `$arg --get svn.$key`);
1212                        if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
1213                                $$v = $tmp;
1214                        }
1215                }
1216        }
1217        delete @$opts{@config_only} if @config_only;
1218}
1219
1220sub extract_metadata {
1221        my $id = shift or return (undef, undef, undef);
1222        my ($url, $rev, $uuid) = ($id =~ /^\s*git-svn-id:\s+(.*)\@(\d+)
1223                                                        \s([a-f\d\-]+)$/x);
1224        if (!defined $rev || !$uuid || !$url) {
1225                # some of the original repositories I made had
1226                # identifiers like this:
1227                ($rev, $uuid) = ($id =~/^\s*git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
1228        }
1229        return ($url, $rev, $uuid);
1230}
1231
1232sub cmt_metadata {
1233        return extract_metadata((grep(/^git-svn-id: /,
1234                command(qw/cat-file commit/, shift)))[-1]);
1235}
1236
1237sub working_head_info {
1238        my ($head, $refs) = @_;
1239        my @args = ('log', '--no-color', '--first-parent', '--pretty=medium');
1240        my ($fh, $ctx) = command_output_pipe(@args, $head);
1241        my $hash;
1242        my %max;
1243        while (<$fh>) {
1244                if ( m{^commit ($::sha1)$} ) {
1245                        unshift @$refs, $hash if $hash and $refs;
1246                        $hash = $1;
1247                        next;
1248                }
1249                next unless s{^\s*(git-svn-id:)}{$1};
1250                my ($url, $rev, $uuid) = extract_metadata($_);
1251                if (defined $url && defined $rev) {
1252                        next if $max{$url} and $max{$url} < $rev;
1253                        if (my $gs = Git::SVN->find_by_url($url)) {
1254                                my $c = $gs->rev_map_get($rev, $uuid);
1255                                if ($c && $c eq $hash) {
1256                                        close $fh; # break the pipe
1257                                        return ($url, $rev, $uuid, $gs);
1258                                } else {
1259                                        $max{$url} ||= $gs->rev_map_max;
1260                                }
1261                        }
1262                }
1263        }
1264        command_close_pipe($fh, $ctx);
1265        (undef, undef, undef, undef);
1266}
1267
1268sub read_commit_parents {
1269        my ($parents, $c) = @_;
1270        chomp(my $p = command_oneline(qw/rev-list --parents -1/, $c));
1271        $p =~ s/^($c)\s*// or die "rev-list --parents -1 $c failed!\n";
1272        @{$parents->{$c}} = split(/ /, $p);
1273}
1274
1275sub linearize_history {
1276        my ($gs, $refs) = @_;
1277        my %parents;
1278        foreach my $c (@$refs) {
1279                read_commit_parents(\%parents, $c);
1280        }
1281
1282        my @linear_refs;
1283        my %skip = ();
1284        my $last_svn_commit = $gs->last_commit;
1285        foreach my $c (reverse @$refs) {
1286                next if $c eq $last_svn_commit;
1287                last if $skip{$c};
1288
1289                unshift @linear_refs, $c;
1290                $skip{$c} = 1;
1291
1292                # we only want the first parent to diff against for linear
1293                # history, we save the rest to inject when we finalize the
1294                # svn commit
1295                my $fp_a = verify_ref("$c~1");
1296                my $fp_b = shift @{$parents{$c}} if $parents{$c};
1297                if (!$fp_a || !$fp_b) {
1298                        die "Commit $c\n",
1299                            "has no parent commit, and therefore ",
1300                            "nothing to diff against.\n",
1301                            "You should be working from a repository ",
1302                            "originally created by git-svn\n";
1303                }
1304                if ($fp_a ne $fp_b) {
1305                        die "$c~1 = $fp_a, however parsing commit $c ",
1306                            "revealed that:\n$c~1 = $fp_b\nBUG!\n";
1307                }
1308
1309                foreach my $p (@{$parents{$c}}) {
1310                        $skip{$p} = 1;
1311                }
1312        }
1313        (\@linear_refs, \%parents);
1314}
1315
1316sub find_file_type_and_diff_status {
1317        my ($path) = @_;
1318        return ('dir', '') if $path eq '';
1319
1320        my $diff_output =
1321            command_oneline(qw(diff --cached --name-status --), $path) || "";
1322        my $diff_status = (split(' ', $diff_output))[0] || "";
1323
1324        my $ls_tree = command_oneline(qw(ls-tree HEAD), $path) || "";
1325
1326        return (undef, undef) if !$diff_status && !$ls_tree;
1327
1328        if ($diff_status eq "A") {
1329                return ("link", $diff_status) if -l $path;
1330                return ("dir", $diff_status) if -d $path;
1331                return ("file", $diff_status);
1332        }
1333
1334        my $mode = (split(' ', $ls_tree))[0] || "";
1335
1336        return ("link", $diff_status) if $mode eq "120000";
1337        return ("dir", $diff_status) if $mode eq "040000";
1338        return ("file", $diff_status);
1339}
1340
1341sub md5sum {
1342        my $arg = shift;
1343        my $ref = ref $arg;
1344        my $md5 = Digest::MD5->new();
1345        if ($ref eq 'GLOB' || $ref eq 'IO::File' || $ref eq 'File::Temp') {
1346                $md5->addfile($arg) or croak $!;
1347        } elsif ($ref eq 'SCALAR') {
1348                $md5->add($$arg) or croak $!;
1349        } elsif (!$ref) {
1350                $md5->add($arg) or croak $!;
1351        } else {
1352                ::fatal "Can't provide MD5 hash for unknown ref type: '", $ref, "'";
1353        }
1354        return $md5->hexdigest();
1355}
1356
1357package Git::SVN;
1358use strict;
1359use warnings;
1360use Fcntl qw/:DEFAULT :seek/;
1361use constant rev_map_fmt => 'NH40';
1362use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
1363            $_repack $_repack_flags $_use_svm_props $_head
1364            $_use_svnsync_props $no_reuse_existing $_minimize_url
1365            $_use_log_author $_add_author_from/;
1366use Carp qw/croak/;
1367use File::Path qw/mkpath/;
1368use File::Copy qw/copy/;
1369use IPC::Open3;
1370
1371my ($_gc_nr, $_gc_period);
1372
1373# properties that we do not log:
1374my %SKIP_PROP;
1375BEGIN {
1376        %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
1377                                        svn:special svn:executable
1378                                        svn:entry:committed-rev
1379                                        svn:entry:last-author
1380                                        svn:entry:uuid
1381                                        svn:entry:committed-date/;
1382
1383        # some options are read globally, but can be overridden locally
1384        # per [svn-remote "..."] section.  Command-line options will *NOT*
1385        # override options set in an [svn-remote "..."] section
1386        no strict 'refs';
1387        for my $option (qw/follow_parent no_metadata use_svm_props
1388                           use_svnsync_props/) {
1389                my $key = $option;
1390                $key =~ tr/_//d;
1391                my $prop = "-$option";
1392                *$option = sub {
1393                        my ($self) = @_;
1394                        return $self->{$prop} if exists $self->{$prop};
1395                        my $k = "svn-remote.$self->{repo_id}.$key";
1396                        eval { command_oneline(qw/config --get/, $k) };
1397                        if ($@) {
1398                                $self->{$prop} = ${"Git::SVN::_$option"};
1399                        } else {
1400                                my $v = command_oneline(qw/config --bool/,$k);
1401                                $self->{$prop} = $v eq 'false' ? 0 : 1;
1402                        }
1403                        return $self->{$prop};
1404                }
1405        }
1406}
1407
1408
1409my (%LOCKFILES, %INDEX_FILES);
1410END {
1411        unlink keys %LOCKFILES if %LOCKFILES;
1412        unlink keys %INDEX_FILES if %INDEX_FILES;
1413}
1414
1415sub resolve_local_globs {
1416        my ($url, $fetch, $glob_spec) = @_;
1417        return unless defined $glob_spec;
1418        my $ref = $glob_spec->{ref};
1419        my $path = $glob_spec->{path};
1420        foreach (command(qw#for-each-ref --format=%(refname) refs/remotes#)) {
1421                next unless m#^refs/remotes/$ref->{regex}$#;
1422                my $p = $1;
1423                my $pathname = desanitize_refname($path->full_path($p));
1424                my $refname = desanitize_refname($ref->full_path($p));
1425                if (my $existing = $fetch->{$pathname}) {
1426                        if ($existing ne $refname) {
1427                                die "Refspec conflict:\n",
1428                                    "existing: refs/remotes/$existing\n",
1429                                    " globbed: refs/remotes/$refname\n";
1430                        }
1431                        my $u = (::cmt_metadata("refs/remotes/$refname"))[0];
1432                        $u =~ s!^\Q$url\E(/|$)!! or die
1433                          "refs/remotes/$refname: '$url' not found in '$u'\n";
1434                        if ($pathname ne $u) {
1435                                warn "W: Refspec glob conflict ",
1436                                     "(ref: refs/remotes/$refname):\n",
1437                                     "expected path: $pathname\n",
1438                                     "    real path: $u\n",
1439                                     "Continuing ahead with $u\n";
1440                                next;
1441                        }
1442                } else {
1443                        $fetch->{$pathname} = $refname;
1444                }
1445        }
1446}
1447
1448sub parse_revision_argument {
1449        my ($base, $head) = @_;
1450        if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
1451                return ($base, $head);
1452        }
1453        return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
1454        return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
1455        return ($head, $head) if ($::_revision eq 'HEAD');
1456        return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
1457        return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
1458        die "revision argument: $::_revision not understood by git-svn\n";
1459}
1460
1461sub fetch_all {
1462        my ($repo_id, $remotes) = @_;
1463        if (ref $repo_id) {
1464                my $gs = $repo_id;
1465                $repo_id = undef;
1466                $repo_id = $gs->{repo_id};
1467        }
1468        $remotes ||= read_all_remotes();
1469        my $remote = $remotes->{$repo_id} or
1470                     die "[svn-remote \"$repo_id\"] unknown\n";
1471        my $fetch = $remote->{fetch};
1472        my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
1473        my (@gs, @globs);
1474        my $ra = Git::SVN::Ra->new($url);
1475        my $uuid = $ra->get_uuid;
1476        my $head = $ra->get_latest_revnum;
1477        my $base = defined $fetch ? $head : 0;
1478
1479        # read the max revs for wildcard expansion (branches/*, tags/*)
1480        foreach my $t (qw/branches tags/) {
1481                defined $remote->{$t} or next;
1482                push @globs, $remote->{$t};
1483                my $max_rev = eval { tmp_config(qw/--int --get/,
1484                                         "svn-remote.$repo_id.${t}-maxRev") };
1485                if (defined $max_rev && ($max_rev < $base)) {
1486                        $base = $max_rev;
1487                } elsif (!defined $max_rev) {
1488                        $base = 0;
1489                }
1490        }
1491
1492        if ($fetch) {
1493                foreach my $p (sort keys %$fetch) {
1494                        my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
1495                        my $lr = $gs->rev_map_max;
1496                        if (defined $lr) {
1497                                $base = $lr if ($lr < $base);
1498                        }
1499                        push @gs, $gs;
1500                }
1501        }
1502
1503        ($base, $head) = parse_revision_argument($base, $head);
1504        $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
1505}
1506
1507sub read_all_remotes {
1508        my $r = {};
1509        my $use_svm_props = eval { command_oneline(qw/config --bool
1510            svn.useSvmProps/) };
1511        $use_svm_props = $use_svm_props eq 'true' if $use_svm_props;
1512        foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
1513                if (m!^(.+)\.fetch=\s*(.*)\s*:\s*(.+)\s*$!) {
1514                        my ($remote, $local_ref, $_remote_ref) = ($1, $2, $3);
1515                        die("svn-remote.$remote: remote ref '$_remote_ref' "
1516                            . "must start with 'refs/remotes/'\n")
1517                                unless $_remote_ref =~ m{^refs/remotes/(.+)};
1518                        my $remote_ref = $1;
1519                        $local_ref =~ s{^/}{};
1520                        $r->{$remote}->{fetch}->{$local_ref} = $remote_ref;
1521                        $r->{$remote}->{svm} = {} if $use_svm_props;
1522                } elsif (m!^(.+)\.usesvmprops=\s*(.*)\s*$!) {
1523                        $r->{$1}->{svm} = {};
1524                } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
1525                        $r->{$1}->{url} = $2;
1526                } elsif (m!^(.+)\.(branches|tags)=
1527                           (.*):refs/remotes/(.+)\s*$/!x) {
1528                        my ($p, $g) = ($3, $4);
1529                        my $rs = $r->{$1}->{$2} = {
1530                                          t => $2,
1531                                          remote => $1,
1532                                          path => Git::SVN::GlobSpec->new($p),
1533                                          ref => Git::SVN::GlobSpec->new($g) };
1534                        if (length($rs->{ref}->{right}) != 0) {
1535                                die "The '*' glob character must be the last ",
1536                                    "character of '$g'\n";
1537                        }
1538                }
1539        }
1540
1541        map {
1542                if (defined $r->{$_}->{svm}) {
1543                        my $svm;
1544                        eval {
1545                                my $section = "svn-remote.$_";
1546                                $svm = {
1547                                        source => tmp_config('--get',
1548                                            "$section.svm-source"),
1549                                        replace => tmp_config('--get',
1550                                            "$section.svm-replace"),
1551                                }
1552                        };
1553                        $r->{$_}->{svm} = $svm;
1554                }
1555        } keys %$r;
1556
1557        $r;
1558}
1559
1560sub init_vars {
1561        $_gc_nr = $_gc_period = 1000;
1562        if (defined $_repack || defined $_repack_flags) {
1563               warn "Repack options are obsolete; they have no effect.\n";
1564        }
1565}
1566
1567sub verify_remotes_sanity {
1568        return unless -d $ENV{GIT_DIR};
1569        my %seen;
1570        foreach (command(qw/config -l/)) {
1571                if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
1572                        if ($seen{$1}) {
1573                                die "Remote ref refs/remote/$1 is tracked by",
1574                                    "\n  \"$_\"\nand\n  \"$seen{$1}\"\n",
1575                                    "Please resolve this ambiguity in ",
1576                                    "your git configuration file before ",
1577                                    "continuing\n";
1578                        }
1579                        $seen{$1} = $_;
1580                }
1581        }
1582}
1583
1584sub find_existing_remote {
1585        my ($url, $remotes) = @_;
1586        return undef if $no_reuse_existing;
1587        my $existing;
1588        foreach my $repo_id (keys %$remotes) {
1589                my $u = $remotes->{$repo_id}->{url} or next;
1590                next if $u ne $url;
1591                $existing = $repo_id;
1592                last;
1593        }
1594        $existing;
1595}
1596
1597sub init_remote_config {
1598        my ($self, $url, $no_write) = @_;
1599        $url =~ s!/+$!!; # strip trailing slash
1600        my $r = read_all_remotes();
1601        my $existing = find_existing_remote($url, $r);
1602        if ($existing) {
1603                unless ($no_write) {
1604                        print STDERR "Using existing ",
1605                                     "[svn-remote \"$existing\"]\n";
1606                }
1607                $self->{repo_id} = $existing;
1608        } elsif ($_minimize_url) {
1609                my $min_url = Git::SVN::Ra->new($url)->minimize_url;
1610                $existing = find_existing_remote($min_url, $r);
1611                if ($existing) {
1612                        unless ($no_write) {
1613                                print STDERR "Using existing ",
1614                                             "[svn-remote \"$existing\"]\n";
1615                        }
1616                        $self->{repo_id} = $existing;
1617                }
1618                if ($min_url ne $url) {
1619                        unless ($no_write) {
1620                                print STDERR "Using higher level of URL: ",
1621                                             "$url => $min_url\n";
1622                        }
1623                        my $old_path = $self->{path};
1624                        $self->{path} = $url;
1625                        $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
1626                        if (length $old_path) {
1627                                $self->{path} .= "/$old_path";
1628                        }
1629                        $url = $min_url;
1630                }
1631        }
1632        my $orig_url;
1633        if (!$existing) {
1634                # verify that we aren't overwriting anything:
1635                $orig_url = eval {
1636                        command_oneline('config', '--get',
1637                                        "svn-remote.$self->{repo_id}.url")
1638                };
1639                if ($orig_url && ($orig_url ne $url)) {
1640                        die "svn-remote.$self->{repo_id}.url already set: ",
1641                            "$orig_url\nwanted to set to: $url\n";
1642                }
1643        }
1644        my ($xrepo_id, $xpath) = find_ref($self->refname);
1645        if (defined $xpath) {
1646                die "svn-remote.$xrepo_id.fetch already set to track ",
1647                    "$xpath:refs/remotes/", $self->refname, "\n";
1648        }
1649        unless ($no_write) {
1650                command_noisy('config',
1651                              "svn-remote.$self->{repo_id}.url", $url);
1652                $self->{path} =~ s{^/}{};
1653                command_noisy('config', '--add',
1654                              "svn-remote.$self->{repo_id}.fetch",
1655                              "$self->{path}:".$self->refname);
1656        }
1657        $self->{url} = $url;
1658}
1659
1660sub find_by_url { # repos_root and, path are optional
1661        my ($class, $full_url, $repos_root, $path) = @_;
1662
1663        return undef unless defined $full_url;
1664        remove_username($full_url);
1665        remove_username($repos_root) if defined $repos_root;
1666        my $remotes = read_all_remotes();
1667        if (defined $full_url && defined $repos_root && !defined $path) {
1668                $path = $full_url;
1669                $path =~ s#^\Q$repos_root\E(?:/|$)##;
1670        }
1671        foreach my $repo_id (keys %$remotes) {
1672                my $u = $remotes->{$repo_id}->{url} or next;
1673                remove_username($u);
1674                next if defined $repos_root && $repos_root ne $u;
1675
1676                my $fetch = $remotes->{$repo_id}->{fetch} || {};
1677                foreach (qw/branches tags/) {
1678                        resolve_local_globs($u, $fetch,
1679                                            $remotes->{$repo_id}->{$_});
1680                }
1681                my $p = $path;
1682                my $rwr = rewrite_root({repo_id => $repo_id});
1683                my $svm = $remotes->{$repo_id}->{svm}
1684                        if defined $remotes->{$repo_id}->{svm};
1685                unless (defined $p) {
1686                        $p = $full_url;
1687                        my $z = $u;
1688                        my $prefix = '';
1689                        if ($rwr) {
1690                                $z = $rwr;
1691                        } elsif (defined $svm) {
1692                                $z = $svm->{source};
1693                                $prefix = $svm->{replace};
1694                                $prefix =~ s#^\Q$u\E(?:/|$)##;
1695                                $prefix =~ s#/$##;
1696                        }
1697                        $p =~ s#^\Q$z\E(?:/|$)#$prefix# or next;
1698                }
1699                foreach my $f (keys %$fetch) {
1700                        next if $f ne $p;
1701                        return Git::SVN->new($fetch->{$f}, $repo_id, $f);
1702                }
1703        }
1704        undef;
1705}
1706
1707sub init {
1708        my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
1709        my $self = _new($class, $repo_id, $ref_id, $path);
1710        if (defined $url) {
1711                $self->init_remote_config($url, $no_write);
1712        }
1713        $self;
1714}
1715
1716sub find_ref {
1717        my ($ref_id) = @_;
1718        foreach (command(qw/config -l/)) {
1719                next unless m!^svn-remote\.(.+)\.fetch=
1720                              \s*(.*)\s*:\s*refs/remotes/(.+)\s*$!x;
1721                my ($repo_id, $path, $ref) = ($1, $2, $3);
1722                if ($ref eq $ref_id) {
1723                        $path = '' if ($path =~ m#^\./?#);
1724                        return ($repo_id, $path);
1725                }
1726        }
1727        (undef, undef, undef);
1728}
1729
1730sub new {
1731        my ($class, $ref_id, $repo_id, $path) = @_;
1732        if (defined $ref_id && !defined $repo_id && !defined $path) {
1733                ($repo_id, $path) = find_ref($ref_id);
1734                if (!defined $repo_id) {
1735                        die "Could not find a \"svn-remote.*.fetch\" key ",
1736                            "in the repository configuration matching: ",
1737                            "refs/remotes/$ref_id\n";
1738                }
1739        }
1740        my $self = _new($class, $repo_id, $ref_id, $path);
1741        if (!defined $self->{path} || !length $self->{path}) {
1742                my $fetch = command_oneline('config', '--get',
1743                                            "svn-remote.$repo_id.fetch",
1744                                            ":refs/remotes/$ref_id\$") or
1745                     die "Failed to read \"svn-remote.$repo_id.fetch\" ",
1746                         "\":refs/remotes/$ref_id\$\" in config\n";
1747                ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
1748        }
1749        $self->{url} = command_oneline('config', '--get',
1750                                       "svn-remote.$repo_id.url") or
1751                  die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
1752        $self->rebuild;
1753        $self;
1754}
1755
1756sub refname {
1757        my ($refname) = "refs/remotes/$_[0]->{ref_id}" ;
1758
1759        # It cannot end with a slash /, we'll throw up on this because
1760        # SVN can't have directories with a slash in their name, either:
1761        if ($refname =~ m{/$}) {
1762                die "ref: '$refname' ends with a trailing slash, this is ",
1763                    "not permitted by git nor Subversion\n";
1764        }
1765
1766        # It cannot have ASCII control character space, tilde ~, caret ^,
1767        # colon :, question-mark ?, asterisk *, space, or open bracket [
1768        # anywhere.
1769        #
1770        # Additionally, % must be escaped because it is used for escaping
1771        # and we want our escaped refname to be reversible
1772        $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
1773
1774        # no slash-separated component can begin with a dot .
1775        # /.* becomes /%2E*
1776        $refname =~ s{/\.}{/%2E}g;
1777
1778        # It cannot have two consecutive dots .. anywhere
1779        # .. becomes %2E%2E
1780        $refname =~ s{\.\.}{%2E%2E}g;
1781
1782        return $refname;
1783}
1784
1785sub desanitize_refname {
1786        my ($refname) = @_;
1787        $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
1788        return $refname;
1789}
1790
1791sub svm_uuid {
1792        my ($self) = @_;
1793        return $self->{svm}->{uuid} if $self->svm;
1794        $self->ra;
1795        unless ($self->{svm}) {
1796                die "SVM UUID not cached, and reading remotely failed\n";
1797        }
1798        $self->{svm}->{uuid};
1799}
1800
1801sub svm {
1802        my ($self) = @_;
1803        return $self->{svm} if $self->{svm};
1804        my $svm;
1805        # see if we have it in our config, first:
1806        eval {
1807                my $section = "svn-remote.$self->{repo_id}";
1808                $svm = {
1809                  source => tmp_config('--get', "$section.svm-source"),
1810                  uuid => tmp_config('--get', "$section.svm-uuid"),
1811                  replace => tmp_config('--get', "$section.svm-replace"),
1812                }
1813        };
1814        if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
1815                $self->{svm} = $svm;
1816        }
1817        $self->{svm};
1818}
1819
1820sub _set_svm_vars {
1821        my ($self, $ra) = @_;
1822        return $ra if $self->svm;
1823
1824        my @err = ( "useSvmProps set, but failed to read SVM properties\n",
1825                    "(svm:source, svm:uuid) ",
1826                    "from the following URLs:\n" );
1827        sub read_svm_props {
1828                my ($self, $ra, $path, $r) = @_;
1829                my $props = ($ra->get_dir($path, $r))[2];
1830                my $src = $props->{'svm:source'};
1831                my $uuid = $props->{'svm:uuid'};
1832                return undef if (!$src || !$uuid);
1833
1834                chomp($src, $uuid);
1835
1836                $uuid =~ m{^[0-9a-f\-]{30,}$}
1837                    or die "doesn't look right - svm:uuid is '$uuid'\n";
1838
1839                # the '!' is used to mark the repos_root!/relative/path
1840                $src =~ s{/?!/?}{/};
1841                $src =~ s{/+$}{}; # no trailing slashes please
1842                # username is of no interest
1843                $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
1844
1845                my $replace = $ra->{url};
1846                $replace .= "/$path" if length $path;
1847
1848                my $section = "svn-remote.$self->{repo_id}";
1849                tmp_config("$section.svm-source", $src);
1850                tmp_config("$section.svm-replace", $replace);
1851                tmp_config("$section.svm-uuid", $uuid);
1852                $self->{svm} = {
1853                        source => $src,
1854                        uuid => $uuid,
1855                        replace => $replace
1856                };
1857        }
1858
1859        my $r = $ra->get_latest_revnum;
1860        my $path = $self->{path};
1861        my %tried;
1862        while (length $path) {
1863                unless ($tried{"$self->{url}/$path"}) {
1864                        return $ra if $self->read_svm_props($ra, $path, $r);
1865                        $tried{"$self->{url}/$path"} = 1;
1866                }
1867                $path =~ s#/?[^/]+$##;
1868        }
1869        die "Path: '$path' should be ''\n" if $path ne '';
1870        return $ra if $self->read_svm_props($ra, $path, $r);
1871        $tried{"$self->{url}/$path"} = 1;
1872
1873        if ($ra->{repos_root} eq $self->{url}) {
1874                die @err, (map { "  $_\n" } keys %tried), "\n";
1875        }
1876
1877        # nope, make sure we're connected to the repository root:
1878        my $ok;
1879        my @tried_b;
1880        $path = $ra->{svn_path};
1881        $ra = Git::SVN::Ra->new($ra->{repos_root});
1882        while (length $path) {
1883                unless ($tried{"$ra->{url}/$path"}) {
1884                        $ok = $self->read_svm_props($ra, $path, $r);
1885                        last if $ok;
1886                        $tried{"$ra->{url}/$path"} = 1;
1887                }
1888                $path =~ s#/?[^/]+$##;
1889        }
1890        die "Path: '$path' should be ''\n" if $path ne '';
1891        $ok ||= $self->read_svm_props($ra, $path, $r);
1892        $tried{"$ra->{url}/$path"} = 1;
1893        if (!$ok) {
1894                die @err, (map { "  $_\n" } keys %tried), "\n";
1895        }
1896        Git::SVN::Ra->new($self->{url});
1897}
1898
1899sub svnsync {
1900        my ($self) = @_;
1901        return $self->{svnsync} if $self->{svnsync};
1902
1903        if ($self->no_metadata) {
1904                die "Can't have both 'noMetadata' and ",
1905                    "'useSvnsyncProps' options set!\n";
1906        }
1907        if ($self->rewrite_root) {
1908                die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
1909                    "options set!\n";
1910        }
1911
1912        my $svnsync;
1913        # see if we have it in our config, first:
1914        eval {
1915                my $section = "svn-remote.$self->{repo_id}";
1916
1917                my $url = tmp_config('--get', "$section.svnsync-url");
1918                ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
1919                   die "doesn't look right - svn:sync-from-url is '$url'\n";
1920
1921                my $uuid = tmp_config('--get', "$section.svnsync-uuid");
1922                ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}) or
1923                   die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
1924
1925                $svnsync = { url => $url, uuid => $uuid }
1926        };
1927        if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
1928                return $self->{svnsync} = $svnsync;
1929        }
1930
1931        my $err = "useSvnsyncProps set, but failed to read " .
1932                  "svnsync property: svn:sync-from-";
1933        my $rp = $self->ra->rev_proplist(0);
1934
1935        my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
1936        ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
1937                   die "doesn't look right - svn:sync-from-url is '$url'\n";
1938
1939        my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
1940        ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}) or
1941                   die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
1942
1943        my $section = "svn-remote.$self->{repo_id}";
1944        tmp_config('--add', "$section.svnsync-uuid", $uuid);
1945        tmp_config('--add', "$section.svnsync-url", $url);
1946        return $self->{svnsync} = { url => $url, uuid => $uuid };
1947}
1948
1949# this allows us to memoize our SVN::Ra UUID locally and avoid a
1950# remote lookup (useful for 'git svn log').
1951sub ra_uuid {
1952        my ($self) = @_;
1953        unless ($self->{ra_uuid}) {
1954                my $key = "svn-remote.$self->{repo_id}.uuid";
1955                my $uuid = eval { tmp_config('--get', $key) };
1956                if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/) {
1957                        $self->{ra_uuid} = $uuid;
1958                } else {
1959                        die "ra_uuid called without URL\n" unless $self->{url};
1960                        $self->{ra_uuid} = $self->ra->get_uuid;
1961                        tmp_config('--add', $key, $self->{ra_uuid});
1962                }
1963        }
1964        $self->{ra_uuid};
1965}
1966
1967sub _set_repos_root {
1968        my ($self, $repos_root) = @_;
1969        my $k = "svn-remote.$self->{repo_id}.reposRoot";
1970        $repos_root ||= $self->ra->{repos_root};
1971        tmp_config($k, $repos_root);
1972        $repos_root;
1973}
1974
1975sub repos_root {
1976        my ($self) = @_;
1977        my $k = "svn-remote.$self->{repo_id}.reposRoot";
1978        eval { tmp_config('--get', $k) } || $self->_set_repos_root;
1979}
1980
1981sub ra {
1982        my ($self) = shift;
1983        my $ra = Git::SVN::Ra->new($self->{url});
1984        $self->_set_repos_root($ra->{repos_root});
1985        if ($self->use_svm_props && !$self->{svm}) {
1986                if ($self->no_metadata) {
1987                        die "Can't have both 'noMetadata' and ",
1988                            "'useSvmProps' options set!\n";
1989                } elsif ($self->use_svnsync_props) {
1990                        die "Can't have both 'useSvnsyncProps' and ",
1991                            "'useSvmProps' options set!\n";
1992                }
1993                $ra = $self->_set_svm_vars($ra);
1994                $self->{-want_revprops} = 1;
1995        }
1996        $ra;
1997}
1998
1999sub rel_path {
2000        my ($self) = @_;
2001        my $repos_root = $self->ra->{repos_root};
2002        return $self->{path} if ($self->{url} eq $repos_root);
2003        my $url = $self->{url} .
2004                  (length $self->{path} ? "/$self->{path}" : $self->{path});
2005        $url =~ s!^\Q$repos_root\E(?:/+|$)!!g;
2006        $url;
2007}
2008
2009# prop_walk(PATH, REV, SUB)
2010# -------------------------
2011# Recursively traverse PATH at revision REV and invoke SUB for each
2012# directory that contains a SVN property.  SUB will be invoked as
2013# follows:  &SUB(gs, path, props);  where `gs' is this instance of
2014# Git::SVN, `path' the path to the directory where the properties
2015# `props' were found.  The `path' will be relative to point of checkout,
2016# that is, if url://repo/trunk is the current Git branch, and that
2017# directory contains a sub-directory `d', SUB will be invoked with `/d/'
2018# as `path' (note the trailing `/').
2019sub prop_walk {
2020        my ($self, $path, $rev, $sub) = @_;
2021
2022        $path =~ s#^/##;
2023        my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev);
2024        $path =~ s#^/*#/#g;
2025        my $p = $path;
2026        # Strip the irrelevant part of the path.
2027        $p =~ s#^/+\Q$self->{path}\E(/|$)#/#;
2028        # Ensure the path is terminated by a `/'.
2029        $p =~ s#/*$#/#;
2030
2031        # The properties contain all the internal SVN stuff nobody
2032        # (usually) cares about.
2033        my $interesting_props = 0;
2034        foreach (keys %{$props}) {
2035                # If it doesn't start with `svn:', it must be a
2036                # user-defined property.
2037                ++$interesting_props and next if $_ !~ /^svn:/;
2038                # FIXME: Fragile, if SVN adds new public properties,
2039                # this needs to be updated.
2040                ++$interesting_props if /^svn:(?:ignore|keywords|executable
2041                                                 |eol-style|mime-type
2042                                                 |externals|needs-lock)$/x;
2043        }
2044        &$sub($self, $p, $props) if $interesting_props;
2045
2046        foreach (sort keys %$dirent) {
2047                next if $dirent->{$_}->{kind} != $SVN::Node::dir;
2048                $self->prop_walk($self->{path} . $p . $_, $rev, $sub);
2049        }
2050}
2051
2052sub last_rev { ($_[0]->last_rev_commit)[0] }
2053sub last_commit { ($_[0]->last_rev_commit)[1] }
2054
2055# returns the newest SVN revision number and newest commit SHA1
2056sub last_rev_commit {
2057        my ($self) = @_;
2058        if (defined $self->{last_rev} && defined $self->{last_commit}) {
2059                return ($self->{last_rev}, $self->{last_commit});
2060        }
2061        my $c = ::verify_ref($self->refname.'^0');
2062        if ($c && !$self->use_svm_props && !$self->no_metadata) {
2063                my $rev = (::cmt_metadata($c))[1];
2064                if (defined $rev) {
2065                        ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
2066                        return ($rev, $c);
2067                }
2068        }
2069        my $map_path = $self->map_path;
2070        unless (-e $map_path) {
2071                ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
2072                return (undef, undef);
2073        }
2074        my ($rev, $commit) = $self->rev_map_max(1);
2075        ($self->{last_rev}, $self->{last_commit}) = ($rev, $commit);
2076        return ($rev, $commit);
2077}
2078
2079sub get_fetch_range {
2080        my ($self, $min, $max) = @_;
2081        $max ||= $self->ra->get_latest_revnum;
2082        $min ||= $self->rev_map_max;
2083        (++$min, $max);
2084}
2085
2086sub tmp_config {
2087        my (@args) = @_;
2088        my $old_def_config = "$ENV{GIT_DIR}/svn/config";
2089        my $config = "$ENV{GIT_DIR}/svn/.metadata";
2090        if (! -f $config && -f $old_def_config) {
2091                rename $old_def_config, $config or
2092                       die "Failed rename $old_def_config => $config: $!\n";
2093        }
2094        my $old_config = $ENV{GIT_CONFIG};
2095        $ENV{GIT_CONFIG} = $config;
2096        $@ = undef;
2097        my @ret = eval {
2098                unless (-f $config) {
2099                        mkfile($config);
2100                        open my $fh, '>', $config or
2101                            die "Can't open $config: $!\n";
2102                        print $fh "; This file is used internally by ",
2103                                  "git-svn\n" or die
2104                                  "Couldn't write to $config: $!\n";
2105                        print $fh "; You should not have to edit it\n" or
2106                              die "Couldn't write to $config: $!\n";
2107                        close $fh or die "Couldn't close $config: $!\n";
2108                }
2109                command('config', @args);
2110        };
2111        my $err = $@;
2112        if (defined $old_config) {
2113                $ENV{GIT_CONFIG} = $old_config;
2114        } else {
2115                delete $ENV{GIT_CONFIG};
2116        }
2117        die $err if $err;
2118        wantarray ? @ret : $ret[0];
2119}
2120
2121sub tmp_index_do {
2122        my ($self, $sub) = @_;
2123        my $old_index = $ENV{GIT_INDEX_FILE};
2124        $ENV{GIT_INDEX_FILE} = $self->{index};
2125        $@ = undef;
2126        my @ret = eval {
2127                my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
2128                mkpath([$dir]) unless -d $dir;
2129                &$sub;
2130        };
2131        my $err = $@;
2132        if (defined $old_index) {
2133                $ENV{GIT_INDEX_FILE} = $old_index;
2134        } else {
2135                delete $ENV{GIT_INDEX_FILE};
2136        }
2137        die $err if $err;
2138        wantarray ? @ret : $ret[0];
2139}
2140
2141sub assert_index_clean {
2142        my ($self, $treeish) = @_;
2143
2144        $self->tmp_index_do(sub {
2145                command_noisy('read-tree', $treeish) unless -e $self->{index};
2146                my $x = command_oneline('write-tree');
2147                my ($y) = (command(qw/cat-file commit/, $treeish) =~
2148                           /^tree ($::sha1)/mo);
2149                return if $y eq $x;
2150
2151                warn "Index mismatch: $y != $x\nrereading $treeish\n";
2152                unlink $self->{index} or die "unlink $self->{index}: $!\n";
2153                command_noisy('read-tree', $treeish);
2154                $x = command_oneline('write-tree');
2155                if ($y ne $x) {
2156                        ::fatal "trees ($treeish) $y != $x\n",
2157                                "Something is seriously wrong...";
2158                }
2159        });
2160}
2161
2162sub get_commit_parents {
2163        my ($self, $log_entry) = @_;
2164        my (%seen, @ret, @tmp);
2165        # legacy support for 'set-tree'; this is only used by set_tree_cb:
2166        if (my $ip = $self->{inject_parents}) {
2167                if (my $commit = delete $ip->{$log_entry->{revision}}) {
2168                        push @tmp, $commit;
2169                }
2170        }
2171        if (my $cur = ::verify_ref($self->refname.'^0')) {
2172                push @tmp, $cur;
2173        }
2174        if (my $ipd = $self->{inject_parents_dcommit}) {
2175                if (my $commit = delete $ipd->{$log_entry->{revision}}) {
2176                        push @tmp, @$commit;
2177                }
2178        }
2179        push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
2180        while (my $p = shift @tmp) {
2181                next if $seen{$p};
2182                $seen{$p} = 1;
2183                push @ret, $p;
2184                # MAXPARENT is defined to 16 in commit-tree.c:
2185                last if @ret >= 16;
2186        }
2187        if (@tmp) {
2188                die "r$log_entry->{revision}: No room for parents:\n\t",
2189                    join("\n\t", @tmp), "\n";
2190        }
2191        @ret;
2192}
2193
2194sub rewrite_root {
2195        my ($self) = @_;
2196        return $self->{-rewrite_root} if exists $self->{-rewrite_root};
2197        my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
2198        my $rwr = eval { command_oneline(qw/config --get/, $k) };
2199        if ($rwr) {
2200                $rwr =~ s#/+$##;
2201                if ($rwr !~ m#^[a-z\+]+://#) {
2202                        die "$rwr is not a valid URL (key: $k)\n";
2203                }
2204        }
2205        $self->{-rewrite_root} = $rwr;
2206}
2207
2208sub metadata_url {
2209        my ($self) = @_;
2210        ($self->rewrite_root || $self->{url}) .
2211           (length $self->{path} ? '/' . $self->{path} : '');
2212}
2213
2214sub full_url {
2215        my ($self) = @_;
2216        $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
2217}
2218
2219
2220sub set_commit_header_env {
2221        my ($log_entry) = @_;
2222        my %env;
2223        foreach my $ned (qw/NAME EMAIL DATE/) {
2224                foreach my $ac (qw/AUTHOR COMMITTER/) {
2225                        $env{"GIT_${ac}_${ned}"} = $ENV{"GIT_${ac}_${ned}"};
2226                }
2227        }
2228
2229        $ENV{GIT_AUTHOR_NAME} = $log_entry->{name};
2230        $ENV{GIT_AUTHOR_EMAIL} = $log_entry->{email};
2231        $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
2232
2233        $ENV{GIT_COMMITTER_NAME} = (defined $log_entry->{commit_name})
2234                                                ? $log_entry->{commit_name}
2235                                                : $log_entry->{name};
2236        $ENV{GIT_COMMITTER_EMAIL} = (defined $log_entry->{commit_email})
2237                                                ? $log_entry->{commit_email}
2238                                                : $log_entry->{email};
2239        \%env;
2240}
2241
2242sub restore_commit_header_env {
2243        my ($env) = @_;
2244        foreach my $ned (qw/NAME EMAIL DATE/) {
2245                foreach my $ac (qw/AUTHOR COMMITTER/) {
2246                        my $k = "GIT_${ac}_${ned}";
2247                        if (defined $env->{$k}) {
2248                                $ENV{$k} = $env->{$k};
2249                        } else {
2250                                delete $ENV{$k};
2251                        }
2252                }
2253        }
2254}
2255
2256sub gc {
2257        command_noisy('gc', '--auto');
2258};
2259
2260sub do_git_commit {
2261        my ($self, $log_entry) = @_;
2262        my $lr = $self->last_rev;
2263        if (defined $lr && $lr >= $log_entry->{revision}) {
2264                die "Last fetched revision of ", $self->refname,
2265                    " was r$lr, but we are about to fetch: ",
2266                    "r$log_entry->{revision}!\n";
2267        }
2268        if (my $c = $self->rev_map_get($log_entry->{revision})) {
2269                croak "$log_entry->{revision} = $c already exists! ",
2270                      "Why are we refetching it?\n";
2271        }
2272        my $old_env = set_commit_header_env($log_entry);
2273        my $tree = $log_entry->{tree};
2274        if (!defined $tree) {
2275                $tree = $self->tmp_index_do(sub {
2276                                            command_oneline('write-tree') });
2277        }
2278        die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
2279
2280        my @exec = ('git', 'commit-tree', $tree);
2281        foreach ($self->get_commit_parents($log_entry)) {
2282                push @exec, '-p', $_;
2283        }
2284        defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
2285                                                                   or croak $!;
2286        binmode $msg_fh;
2287
2288        # we always get UTF-8 from SVN, but we may want our commits in
2289        # a different encoding.
2290        if (my $enc = Git::config('i18n.commitencoding')) {
2291                require Encode;
2292                Encode::from_to($log_entry->{log}, 'UTF-8', $enc);
2293        }
2294        print $msg_fh $log_entry->{log} or croak $!;
2295        restore_commit_header_env($old_env);
2296        unless ($self->no_metadata) {
2297                print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
2298                              or croak $!;
2299        }
2300        $msg_fh->flush == 0 or croak $!;
2301        close $msg_fh or croak $!;
2302        chomp(my $commit = do { local $/; <$out_fh> });
2303        close $out_fh or croak $!;
2304        waitpid $pid, 0;
2305        croak $? if $?;
2306        if ($commit !~ /^$::sha1$/o) {
2307                die "Failed to commit, invalid sha1: $commit\n";
2308        }
2309
2310        $self->rev_map_set($log_entry->{revision}, $commit, 1);
2311
2312        $self->{last_rev} = $log_entry->{revision};
2313        $self->{last_commit} = $commit;
2314        print "r$log_entry->{revision}";
2315        if (defined $log_entry->{svm_revision}) {
2316                 print " (\@$log_entry->{svm_revision})";
2317                 $self->rev_map_set($log_entry->{svm_revision}, $commit,
2318                                   0, $self->svm_uuid);
2319        }
2320        print " = $commit ($self->{ref_id})\n";
2321        if (--$_gc_nr == 0) {
2322                $_gc_nr = $_gc_period;
2323                gc();
2324        }
2325        return $commit;
2326}
2327
2328sub match_paths {
2329        my ($self, $paths, $r) = @_;
2330        return 1 if $self->{path} eq '';
2331        if (my $path = $paths->{"/$self->{path}"}) {
2332                return ($path->{action} eq 'D') ? 0 : 1;
2333        }
2334        $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
2335        if (grep /$self->{path_regex}/, keys %$paths) {
2336                return 1;
2337        }
2338        my $c = '';
2339        foreach (split m#/#, $self->{path}) {
2340                $c .= "/$_";
2341                next unless ($paths->{$c} &&
2342                             ($paths->{$c}->{action} =~ /^[AR]$/));
2343                if ($self->ra->check_path($self->{path}, $r) ==
2344                    $SVN::Node::dir) {
2345                        return 1;
2346                }
2347        }
2348        return 0;
2349}
2350
2351sub find_parent_branch {
2352        my ($self, $paths, $rev) = @_;
2353        return undef unless $self->follow_parent;
2354        unless (defined $paths) {
2355                my $err_handler = $SVN::Error::handler;
2356                $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
2357                $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1, sub {
2358                                   $paths =
2359                                      Git::SVN::Ra::dup_changed_paths($_[0]) });
2360                $SVN::Error::handler = $err_handler;
2361        }
2362        return undef unless defined $paths;
2363
2364        # look for a parent from another branch:
2365        my @b_path_components = split m#/#, $self->rel_path;
2366        my @a_path_components;
2367        my $i;
2368        while (@b_path_components) {
2369                $i = $paths->{'/'.join('/', @b_path_components)};
2370                last if $i && defined $i->{copyfrom_path};
2371                unshift(@a_path_components, pop(@b_path_components));
2372        }
2373        return undef unless defined $i && defined $i->{copyfrom_path};
2374        my $branch_from = $i->{copyfrom_path};
2375        if (@a_path_components) {
2376                print STDERR "branch_from: $branch_from => ";
2377                $branch_from .= '/'.join('/', @a_path_components);
2378                print STDERR $branch_from, "\n";
2379        }
2380        my $r = $i->{copyfrom_rev};
2381        my $repos_root = $self->ra->{repos_root};
2382        my $url = $self->ra->{url};
2383        my $new_url = $repos_root . $branch_from;
2384        print STDERR  "Found possible branch point: ",
2385                      "$new_url => ", $self->full_url, ", $r\n";
2386        $branch_from =~ s#^/##;
2387        my $gs = Git::SVN->find_by_url($new_url, $repos_root, $branch_from);
2388        unless ($gs) {
2389                my $ref_id = $self->{ref_id};
2390                $ref_id =~ s/\@\d+$//;
2391                $ref_id .= "\@$r";
2392                # just grow a tail if we're not unique enough :x
2393                $ref_id .= '-' while find_ref($ref_id);
2394                print STDERR "Initializing parent: $ref_id\n";
2395                my ($u, $p, $repo_id) = ($new_url, '', $ref_id);
2396                if ($u =~ s#^\Q$url\E(/|$)##) {
2397                        $p = $u;
2398                        $u = $url;
2399                        $repo_id = $self->{repo_id};
2400                }
2401                $gs = Git::SVN->init($u, $p, $repo_id, $ref_id, 1);
2402        }
2403        my ($r0, $parent) = $gs->find_rev_before($r, 1);
2404        if (!defined $r0 || !defined $parent) {
2405                my ($base, $head) = parse_revision_argument(0, $r);
2406                if ($base <= $r) {
2407                        $gs->fetch($base, $r);
2408                }
2409                ($r0, $parent) = $gs->last_rev_commit;
2410        }
2411        if (defined $r0 && defined $parent) {
2412                print STDERR "Found branch parent: ($self->{ref_id}) $parent\n";
2413                my $ed;
2414                if ($self->ra->can_do_switch) {
2415                        $self->assert_index_clean($parent);
2416                        print STDERR "Following parent with do_switch\n";
2417                        # do_switch works with svn/trunk >= r22312, but that
2418                        # is not included with SVN 1.4.3 (the latest version
2419                        # at the moment), so we can't rely on it
2420                        $self->{last_commit} = $parent;
2421                        $ed = SVN::Git::Fetcher->new($self);
2422                        $gs->ra->gs_do_switch($r0, $rev, $gs,
2423                                              $self->full_url, $ed)
2424                          or die "SVN connection failed somewhere...\n";
2425                } elsif ($self->ra->trees_match($new_url, $r0,
2426                                                $self->full_url, $rev)) {
2427                        print STDERR "Trees match:\n",
2428                                     "  $new_url\@$r0\n",
2429                                     "  ${\$self->full_url}\@$rev\n",
2430                                     "Following parent with no changes\n";
2431                        $self->tmp_index_do(sub {
2432                            command_noisy('read-tree', $parent);
2433                        });
2434                        $self->{last_commit} = $parent;
2435                } else {
2436                        print STDERR "Following parent with do_update\n";
2437                        $ed = SVN::Git::Fetcher->new($self);
2438                        $self->ra->gs_do_update($rev, $rev, $self, $ed)
2439                          or die "SVN connection failed somewhere...\n";
2440                }
2441                print STDERR "Successfully followed parent\n";
2442                return $self->make_log_entry($rev, [$parent], $ed);
2443        }
2444        return undef;
2445}
2446
2447sub do_fetch {
2448        my ($self, $paths, $rev) = @_;
2449        my $ed;
2450        my ($last_rev, @parents);
2451        if (my $lc = $self->last_commit) {
2452                # we can have a branch that was deleted, then re-added
2453                # under the same name but copied from another path, in
2454                # which case we'll have multiple parents (we don't
2455                # want to break the original ref, nor lose copypath info):
2456                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2457                        push @{$log_entry->{parents}}, $lc;
2458                        return $log_entry;
2459                }
2460                $ed = SVN::Git::Fetcher->new($self);
2461                $last_rev = $self->{last_rev};
2462                $ed->{c} = $lc;
2463                @parents = ($lc);
2464        } else {
2465                $last_rev = $rev;
2466                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2467                        return $log_entry;
2468                }
2469                $ed = SVN::Git::Fetcher->new($self);
2470        }
2471        unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
2472                die "SVN connection failed somewhere...\n";
2473        }
2474        $self->make_log_entry($rev, \@parents, $ed);
2475}
2476
2477sub get_untracked {
2478        my ($self, $ed) = @_;
2479        my @out;
2480        my $h = $ed->{empty};
2481        foreach (sort keys %$h) {
2482                my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
2483                push @out, "  $act: " . uri_encode($_);
2484                warn "W: $act: $_\n";
2485        }
2486        foreach my $t (qw/dir_prop file_prop/) {
2487                $h = $ed->{$t} or next;
2488                foreach my $path (sort keys %$h) {
2489                        my $ppath = $path eq '' ? '.' : $path;
2490                        foreach my $prop (sort keys %{$h->{$path}}) {
2491                                next if $SKIP_PROP{$prop};
2492                                my $v = $h->{$path}->{$prop};
2493                                my $t_ppath_prop = "$t: " .
2494                                                    uri_encode($ppath) . ' ' .
2495                                                    uri_encode($prop);
2496                                if (defined $v) {
2497                                        push @out, "  +$t_ppath_prop " .
2498                                                   uri_encode($v);
2499                                } else {
2500                                        push @out, "  -$t_ppath_prop";
2501                                }
2502                        }
2503                }
2504        }
2505        foreach my $t (qw/absent_file absent_directory/) {
2506                $h = $ed->{$t} or next;
2507                foreach my $parent (sort keys %$h) {
2508                        foreach my $path (sort @{$h->{$parent}}) {
2509                                push @out, "  $t: " .
2510                                           uri_encode("$parent/$path");
2511                                warn "W: $t: $parent/$path ",
2512                                     "Insufficient permissions?\n";
2513                        }
2514                }
2515        }
2516        \@out;
2517}
2518
2519sub parse_svn_date {
2520        my $date = shift || return '+0000 1970-01-01 00:00:00';
2521        my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2522                                            (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x) or
2523                                         croak "Unable to parse date: $date\n";
2524        "+0000 $Y-$m-$d $H:$M:$S";
2525}
2526
2527sub check_author {
2528        my ($author) = @_;
2529        if (!defined $author || length $author == 0) {
2530                $author = '(no author)';
2531        } elsif (defined $::_authors && ! defined $::users{$author}) {
2532                die "Author: $author not defined in $::_authors file\n";
2533        }
2534        $author;
2535}
2536
2537sub make_log_entry {
2538        my ($self, $rev, $parents, $ed) = @_;
2539        my $untracked = $self->get_untracked($ed);
2540
2541        open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
2542        print $un "r$rev\n" or croak $!;
2543        print $un $_, "\n" foreach @$untracked;
2544        my %log_entry = ( parents => $parents || [], revision => $rev,
2545                          log => '');
2546
2547        my $headrev;
2548        my $logged = delete $self->{logged_rev_props};
2549        if (!$logged || $self->{-want_revprops}) {
2550                my $rp = $self->ra->rev_proplist($rev);
2551                foreach (sort keys %$rp) {
2552                        my $v = $rp->{$_};
2553                        if (/^svn:(author|date|log)$/) {
2554                                $log_entry{$1} = $v;
2555                        } elsif ($_ eq 'svm:headrev') {
2556                                $headrev = $v;
2557                        } else {
2558                                print $un "  rev_prop: ", uri_encode($_), ' ',
2559                                          uri_encode($v), "\n";
2560                        }
2561                }
2562        } else {
2563                map { $log_entry{$_} = $logged->{$_} } keys %$logged;
2564        }
2565        close $un or croak $!;
2566
2567        $log_entry{date} = parse_svn_date($log_entry{date});
2568        $log_entry{log} .= "\n";
2569        my $author = $log_entry{author} = check_author($log_entry{author});
2570        my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
2571                                                       : ($author, undef);
2572
2573        my ($commit_name, $commit_email) = ($name, $email);
2574        if ($_use_log_author) {
2575                my $name_field;
2576                if ($log_entry{log} =~ /From:\s+(.*\S)\s*\n/i) {
2577                        $name_field = $1;
2578                } elsif ($log_entry{log} =~ /Signed-off-by:\s+(.*\S)\s*\n/i) {
2579                        $name_field = $1;
2580                }
2581                if (!defined $name_field) {
2582                        if (!defined $email) {
2583                                $email = $name;
2584                        }
2585                } elsif ($name_field =~ /(.*?)\s+<(.*)>/) {
2586                        ($name, $email) = ($1, $2);
2587                } elsif ($name_field =~ /(.*)@/) {
2588                        ($name, $email) = ($1, $name_field);
2589                } else {
2590                        ($name, $email) = ($name_field, $name_field);
2591                }
2592        }
2593        if (defined $headrev && $self->use_svm_props) {
2594                if ($self->rewrite_root) {
2595                        die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
2596                            "options set!\n";
2597                }
2598                my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$};
2599                # we don't want "SVM: initializing mirror for junk" ...
2600                return undef if $r == 0;
2601                my $svm = $self->svm;
2602                if ($uuid ne $svm->{uuid}) {
2603                        die "UUID mismatch on SVM path:\n",
2604                            "expected: $svm->{uuid}\n",
2605                            "     got: $uuid\n";
2606                }
2607                my $full_url = $self->full_url;
2608                $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
2609                             die "Failed to replace '$svm->{replace}' with ",
2610                                 "'$svm->{source}' in $full_url\n";
2611                # throw away username for storing in records
2612                remove_username($full_url);
2613                $log_entry{metadata} = "$full_url\@$r $uuid";
2614                $log_entry{svm_revision} = $r;
2615                $email ||= "$author\@$uuid";
2616                $commit_email ||= "$author\@$uuid";
2617        } elsif ($self->use_svnsync_props) {
2618                my $full_url = $self->svnsync->{url};
2619                $full_url .= "/$self->{path}" if length $self->{path};
2620                remove_username($full_url);
2621                my $uuid = $self->svnsync->{uuid};
2622                $log_entry{metadata} = "$full_url\@$rev $uuid";
2623                $email ||= "$author\@$uuid";
2624                $commit_email ||= "$author\@$uuid";
2625        } else {
2626                my $url = $self->metadata_url;
2627                remove_username($url);
2628                $log_entry{metadata} = "$url\@$rev " .
2629                                       $self->ra->get_uuid;
2630                $email ||= "$author\@" . $self->ra->get_uuid;
2631                $commit_email ||= "$author\@" . $self->ra->get_uuid;
2632        }
2633        $log_entry{name} = $name;
2634        $log_entry{email} = $email;
2635        $log_entry{commit_name} = $commit_name;
2636        $log_entry{commit_email} = $commit_email;
2637        \%log_entry;
2638}
2639
2640sub fetch {
2641        my ($self, $min_rev, $max_rev, @parents) = @_;
2642        my ($last_rev, $last_commit) = $self->last_rev_commit;
2643        my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
2644        $self->ra->gs_fetch_loop_common($base, $head, [$self]);
2645}
2646
2647sub set_tree_cb {
2648        my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
2649        $self->{inject_parents} = { $rev => $tree };
2650        $self->fetch(undef, undef);
2651}
2652
2653sub set_tree {
2654        my ($self, $tree) = (shift, shift);
2655        my $log_entry = ::get_commit_entry($tree);
2656        unless ($self->{last_rev}) {
2657                ::fatal("Must have an existing revision to commit");
2658        }
2659        my %ed_opts = ( r => $self->{last_rev},
2660                        log => $log_entry->{log},
2661                        ra => $self->ra,
2662                        tree_a => $self->{last_commit},
2663                        tree_b => $tree,
2664                        editor_cb => sub {
2665                               $self->set_tree_cb($log_entry, $tree, @_) },
2666                        svn_path => $self->{path} );
2667        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
2668                print "No changes\nr$self->{last_rev} = $tree\n";
2669        }
2670}
2671
2672sub rebuild_from_rev_db {
2673        my ($self, $path) = @_;
2674        my $r = -1;
2675        open my $fh, '<', $path or croak "open: $!";
2676        binmode $fh or croak "binmode: $!";
2677        while (<$fh>) {
2678                length($_) == 41 or croak "inconsistent size in ($_) != 41";
2679                chomp($_);
2680                ++$r;
2681                next if $_ eq ('0' x 40);
2682                $self->rev_map_set($r, $_);
2683                print "r$r = $_\n";
2684        }
2685        close $fh or croak "close: $!";
2686        unlink $path or croak "unlink: $!";
2687}
2688
2689sub rebuild {
2690        my ($self) = @_;
2691        my $map_path = $self->map_path;
2692        my $partial = (-e $map_path && ! -z $map_path);
2693        return unless ::verify_ref($self->refname.'^0');
2694        if (!$partial && ($self->use_svm_props || $self->no_metadata)) {
2695                my $rev_db = $self->rev_db_path;
2696                $self->rebuild_from_rev_db($rev_db);
2697                if ($self->use_svm_props) {
2698                        my $svm_rev_db = $self->rev_db_path($self->svm_uuid);
2699                        $self->rebuild_from_rev_db($svm_rev_db);
2700                }
2701                $self->unlink_rev_db_symlink;
2702                return;
2703        }
2704        print "Rebuilding $map_path ...\n" if (!$partial);
2705        my ($base_rev, $head) = ($partial ? $self->rev_map_max_norebuild(1) :
2706                (undef, undef));
2707        my ($log, $ctx) =
2708            command_output_pipe(qw/rev-list --pretty=raw --no-color --reverse/,
2709                                ($head ? "$head.." : "") . $self->refname,
2710                                '--');
2711        my $metadata_url = $self->metadata_url;
2712        remove_username($metadata_url);
2713        my $svn_uuid = $self->ra_uuid;
2714        my $c;
2715        while (<$log>) {
2716                if ( m{^commit ($::sha1)$} ) {
2717                        $c = $1;
2718                        next;
2719                }
2720                next unless s{^\s*(git-svn-id:)}{$1};
2721                my ($url, $rev, $uuid) = ::extract_metadata($_);
2722                remove_username($url);
2723
2724                # ignore merges (from set-tree)
2725                next if (!defined $rev || !$uuid);
2726
2727                # if we merged or otherwise started elsewhere, this is
2728                # how we break out of it
2729                if (($uuid ne $svn_uuid) ||
2730                    ($metadata_url && $url && ($url ne $metadata_url))) {
2731                        next;
2732                }
2733                if ($partial && $head) {
2734                        print "Partial-rebuilding $map_path ...\n";
2735                        print "Currently at $base_rev = $head\n";
2736                        $head = undef;
2737                }
2738
2739                $self->rev_map_set($rev, $c);
2740                print "r$rev = $c\n";
2741        }
2742        command_close_pipe($log, $ctx);
2743        print "Done rebuilding $map_path\n" if (!$partial || !$head);
2744        my $rev_db_path = $self->rev_db_path;
2745        if (-f $self->rev_db_path) {
2746                unlink $self->rev_db_path or croak "unlink: $!";
2747        }
2748        $self->unlink_rev_db_symlink;
2749}
2750
2751# rev_map:
2752# Tie::File seems to be prone to offset errors if revisions get sparse,
2753# it's not that fast, either.  Tie::File is also not in Perl 5.6.  So
2754# one of my favorite modules is out :<  Next up would be one of the DBM
2755# modules, but I'm not sure which is most portable...
2756#
2757# This is the replacement for the rev_db format, which was too big
2758# and inefficient for large repositories with a lot of sparse history
2759# (mainly tags)
2760#
2761# The format is this:
2762#   - 24 bytes for every record,
2763#     * 4 bytes for the integer representing an SVN revision number
2764#     * 20 bytes representing the sha1 of a git commit
2765#   - No empty padding records like the old format
2766#     (except the last record, which can be overwritten)
2767#   - new records are written append-only since SVN revision numbers
2768#     increase monotonically
2769#   - lookups on SVN revision number are done via a binary search
2770#   - Piping the file to xxd -c24 is a good way of dumping it for
2771#     viewing or editing (piped back through xxd -r), should the need
2772#     ever arise.
2773#   - The last record can be padding revision with an all-zero sha1
2774#     This is used to optimize fetch performance when using multiple
2775#     "fetch" directives in .git/config
2776#
2777# These files are disposable unless noMetadata or useSvmProps is set
2778
2779sub _rev_map_set {
2780        my ($fh, $rev, $commit) = @_;
2781
2782        binmode $fh or croak "binmode: $!";
2783        my $size = (stat($fh))[7];
2784        ($size % 24) == 0 or croak "inconsistent size: $size";
2785
2786        my $wr_offset = 0;
2787        if ($size > 0) {
2788                sysseek($fh, -24, SEEK_END) or croak "seek: $!";
2789                my $read = sysread($fh, my $buf, 24) or croak "read: $!";
2790                $read == 24 or croak "read only $read bytes (!= 24)";
2791                my ($last_rev, $last_commit) = unpack(rev_map_fmt, $buf);
2792                if ($last_commit eq ('0' x40)) {
2793                        if ($size >= 48) {
2794                                sysseek($fh, -48, SEEK_END) or croak "seek: $!";
2795                                $read = sysread($fh, $buf, 24) or
2796                                    croak "read: $!";
2797                                $read == 24 or
2798                                    croak "read only $read bytes (!= 24)";
2799                                ($last_rev, $last_commit) =
2800                                    unpack(rev_map_fmt, $buf);
2801                                if ($last_commit eq ('0' x40)) {
2802                                        croak "inconsistent .rev_map\n";
2803                                }
2804                        }
2805                        if ($last_rev >= $rev) {
2806                                croak "last_rev is higher!: $last_rev >= $rev";
2807                        }
2808                        $wr_offset = -24;
2809                }
2810        }
2811        sysseek($fh, $wr_offset, SEEK_END) or croak "seek: $!";
2812        syswrite($fh, pack(rev_map_fmt, $rev, $commit), 24) == 24 or
2813          croak "write: $!";
2814}
2815
2816sub mkfile {
2817        my ($path) = @_;
2818        unless (-e $path) {
2819                my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
2820                mkpath([$dir]) unless -d $dir;
2821                open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
2822                close $fh or die "Couldn't close (create) $path: $!\n";
2823        }
2824}
2825
2826sub rev_map_set {
2827        my ($self, $rev, $commit, $update_ref, $uuid) = @_;
2828        length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
2829        my $db = $self->map_path($uuid);
2830        my $db_lock = "$db.lock";
2831        my $sig;
2832        if ($update_ref) {
2833                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
2834                            $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
2835        }
2836        mkfile($db);
2837
2838        $LOCKFILES{$db_lock} = 1;
2839        my $sync;
2840        # both of these options make our .rev_db file very, very important
2841        # and we can't afford to lose it because rebuild() won't work
2842        if ($self->use_svm_props || $self->no_metadata) {
2843                $sync = 1;
2844                copy($db, $db_lock) or die "rev_map_set(@_): ",
2845                                           "Failed to copy: ",
2846                                           "$db => $db_lock ($!)\n";
2847        } else {
2848                rename $db, $db_lock or die "rev_map_set(@_): ",
2849                                            "Failed to rename: ",
2850                                            "$db => $db_lock ($!)\n";
2851        }
2852
2853        sysopen(my $fh, $db_lock, O_RDWR | O_CREAT)
2854             or croak "Couldn't open $db_lock: $!\n";
2855        _rev_map_set($fh, $rev, $commit);
2856        if ($sync) {
2857                $fh->flush or die "Couldn't flush $db_lock: $!\n";
2858                $fh->sync or die "Couldn't sync $db_lock: $!\n";
2859        }
2860        close $fh or croak $!;
2861        if ($update_ref) {
2862                $_head = $self;
2863                command_noisy('update-ref', '-m', "r$rev",
2864                              $self->refname, $commit);
2865        }
2866        rename $db_lock, $db or die "rev_map_set(@_): ", "Failed to rename: ",
2867                                    "$db_lock => $db ($!)\n";
2868        delete $LOCKFILES{$db_lock};
2869        if ($update_ref) {
2870                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
2871                            $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
2872                kill $sig, $$ if defined $sig;
2873        }
2874}
2875
2876# If want_commit, this will return an array of (rev, commit) where
2877# commit _must_ be a valid commit in the archive.
2878# Otherwise, it'll return the max revision (whether or not the
2879# commit is valid or just a 0x40 placeholder).
2880sub rev_map_max {
2881        my ($self, $want_commit) = @_;
2882        $self->rebuild;
2883        my ($r, $c) = $self->rev_map_max_norebuild($want_commit);
2884        $want_commit ? ($r, $c) : $r;
2885}
2886
2887sub rev_map_max_norebuild {
2888        my ($self, $want_commit) = @_;
2889        my $map_path = $self->map_path;
2890        stat $map_path or return $want_commit ? (0, undef) : 0;
2891        sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
2892        binmode $fh or croak "binmode: $!";
2893        my $size = (stat($fh))[7];
2894        ($size % 24) == 0 or croak "inconsistent size: $size";
2895
2896        if ($size == 0) {
2897                close $fh or croak "close: $!";
2898                return $want_commit ? (0, undef) : 0;
2899        }
2900
2901        sysseek($fh, -24, SEEK_END) or croak "seek: $!";
2902        sysread($fh, my $buf, 24) == 24 or croak "read: $!";
2903        my ($r, $c) = unpack(rev_map_fmt, $buf);
2904        if ($want_commit && $c eq ('0' x40)) {
2905                if ($size < 48) {
2906                        return $want_commit ? (0, undef) : 0;
2907                }
2908                sysseek($fh, -48, SEEK_END) or croak "seek: $!";
2909                sysread($fh, $buf, 24) == 24 or croak "read: $!";
2910                ($r, $c) = unpack(rev_map_fmt, $buf);
2911                if ($c eq ('0'x40)) {
2912                        croak "Penultimate record is all-zeroes in $map_path";
2913                }
2914        }
2915        close $fh or croak "close: $!";
2916        $want_commit ? ($r, $c) : $r;
2917}
2918
2919sub rev_map_get {
2920        my ($self, $rev, $uuid) = @_;
2921        my $map_path = $self->map_path($uuid);
2922        return undef unless -e $map_path;
2923
2924        sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
2925        binmode $fh or croak "binmode: $!";
2926        my $size = (stat($fh))[7];
2927        ($size % 24) == 0 or croak "inconsistent size: $size";
2928
2929        if ($size == 0) {
2930                close $fh or croak "close: $fh";
2931                return undef;
2932        }
2933
2934        my ($l, $u) = (0, $size - 24);
2935        my ($r, $c, $buf);
2936
2937        while ($l <= $u) {
2938                my $i = int(($l/24 + $u/24) / 2) * 24;
2939                sysseek($fh, $i, SEEK_SET) or croak "seek: $!";
2940                sysread($fh, my $buf, 24) == 24 or croak "read: $!";
2941                my ($r, $c) = unpack('NH40', $buf);
2942
2943                if ($r < $rev) {
2944                        $l = $i + 24;
2945                } elsif ($r > $rev) {
2946                        $u = $i - 24;
2947                } else { # $r == $rev
2948                        close($fh) or croak "close: $!";
2949                        return $c eq ('0' x 40) ? undef : $c;
2950                }
2951        }
2952        close($fh) or croak "close: $!";
2953        undef;
2954}
2955
2956# Finds the first svn revision that exists on (if $eq_ok is true) or
2957# before $rev for the current branch.  It will not search any lower
2958# than $min_rev.  Returns the git commit hash and svn revision number
2959# if found, else (undef, undef).
2960sub find_rev_before {
2961        my ($self, $rev, $eq_ok, $min_rev) = @_;
2962        --$rev unless $eq_ok;
2963        $min_rev ||= 1;
2964        while ($rev >= $min_rev) {
2965                if (my $c = $self->rev_map_get($rev)) {
2966                        return ($rev, $c);
2967                }
2968                --$rev;
2969        }
2970        return (undef, undef);
2971}
2972
2973# Finds the first svn revision that exists on (if $eq_ok is true) or
2974# after $rev for the current branch.  It will not search any higher
2975# than $max_rev.  Returns the git commit hash and svn revision number
2976# if found, else (undef, undef).
2977sub find_rev_after {
2978        my ($self, $rev, $eq_ok, $max_rev) = @_;
2979        ++$rev unless $eq_ok;
2980        $max_rev ||= $self->rev_map_max;
2981        while ($rev <= $max_rev) {
2982                if (my $c = $self->rev_map_get($rev)) {
2983                        return ($rev, $c);
2984                }
2985                ++$rev;
2986        }
2987        return (undef, undef);
2988}
2989
2990sub _new {
2991        my ($class, $repo_id, $ref_id, $path) = @_;
2992        unless (defined $repo_id && length $repo_id) {
2993                $repo_id = $Git::SVN::default_repo_id;
2994        }
2995        unless (defined $ref_id && length $ref_id) {
2996                $_[2] = $ref_id = $Git::SVN::default_ref_id;
2997        }
2998        $_[1] = $repo_id;
2999        my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
3000        $_[3] = $path = '' unless (defined $path);
3001        mkpath(["$ENV{GIT_DIR}/svn"]);
3002        bless {
3003                ref_id => $ref_id, dir => $dir, index => "$dir/index",
3004                path => $path, config => "$ENV{GIT_DIR}/svn/config",
3005                map_root => "$dir/.rev_map", repo_id => $repo_id }, $class;
3006}
3007
3008# for read-only access of old .rev_db formats
3009sub unlink_rev_db_symlink {
3010        my ($self) = @_;
3011        my $link = $self->rev_db_path;
3012        $link =~ s/\.[\w-]+$// or croak "missing UUID at the end of $link";
3013        if (-l $link) {
3014                unlink $link or croak "unlink: $link failed!";
3015        }
3016}
3017
3018sub rev_db_path {
3019        my ($self, $uuid) = @_;
3020        my $db_path = $self->map_path($uuid);
3021        $db_path =~ s{/\.rev_map\.}{/\.rev_db\.}
3022            or croak "map_path: $db_path does not contain '/.rev_map.' !";
3023        $db_path;
3024}
3025
3026# the new replacement for .rev_db
3027sub map_path {
3028        my ($self, $uuid) = @_;
3029        $uuid ||= $self->ra_uuid;
3030        "$self->{map_root}.$uuid";
3031}
3032
3033sub uri_encode {
3034        my ($f) = @_;
3035        $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
3036        $f
3037}
3038
3039sub remove_username {
3040        $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
3041}
3042
3043package Git::SVN::Prompt;
3044use strict;
3045use warnings;
3046require SVN::Core;
3047use vars qw/$_no_auth_cache $_username/;
3048
3049sub simple {
3050        my ($cred, $realm, $default_username, $may_save, $pool) = @_;
3051        $may_save = undef if $_no_auth_cache;
3052        $default_username = $_username if defined $_username;
3053        if (defined $default_username && length $default_username) {
3054                if (defined $realm && length $realm) {
3055                        print STDERR "Authentication realm: $realm\n";
3056                        STDERR->flush;
3057                }
3058                $cred->username($default_username);
3059        } else {
3060                username($cred, $realm, $may_save, $pool);
3061        }
3062        $cred->password(_read_password("Password for '" .
3063                                       $cred->username . "': ", $realm));
3064        $cred->may_save($may_save);
3065        $SVN::_Core::SVN_NO_ERROR;
3066}
3067
3068sub ssl_server_trust {
3069        my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
3070        $may_save = undef if $_no_auth_cache;
3071        print STDERR "Error validating server certificate for '$realm':\n";
3072        {
3073                no warnings 'once';
3074                # All variables SVN::Auth::SSL::* are used only once,
3075                # so we're shutting up Perl warnings about this.
3076                if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
3077                        print STDERR " - The certificate is not issued ",
3078                            "by a trusted authority. Use the\n",
3079                            "   fingerprint to validate ",
3080                            "the certificate manually!\n";
3081                }
3082                if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
3083                        print STDERR " - The certificate hostname ",
3084                            "does not match.\n";
3085                }
3086                if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
3087                        print STDERR " - The certificate is not yet valid.\n";
3088                }
3089                if ($failures & $SVN::Auth::SSL::EXPIRED) {
3090                        print STDERR " - The certificate has expired.\n";
3091                }
3092                if ($failures & $SVN::Auth::SSL::OTHER) {
3093                        print STDERR " - The certificate has ",
3094                            "an unknown error.\n";
3095                }
3096        } # no warnings 'once'
3097        printf STDERR
3098                "Certificate information:\n".
3099                " - Hostname: %s\n".
3100                " - Valid: from %s until %s\n".
3101                " - Issuer: %s\n".
3102                " - Fingerprint: %s\n",
3103                map $cert_info->$_, qw(hostname valid_from valid_until
3104                                       issuer_dname fingerprint);
3105        my $choice;
3106prompt:
3107        print STDERR $may_save ?
3108              "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
3109              "(R)eject or accept (t)emporarily? ";
3110        STDERR->flush;
3111        $choice = lc(substr(<STDIN> || 'R', 0, 1));
3112        if ($choice =~ /^t$/i) {
3113                $cred->may_save(undef);
3114        } elsif ($choice =~ /^r$/i) {
3115                return -1;
3116        } elsif ($may_save && $choice =~ /^p$/i) {
3117                $cred->may_save($may_save);
3118        } else {
3119                goto prompt;
3120        }
3121        $cred->accepted_failures($failures);
3122        $SVN::_Core::SVN_NO_ERROR;
3123}
3124
3125sub ssl_client_cert {
3126        my ($cred, $realm, $may_save, $pool) = @_;
3127        $may_save = undef if $_no_auth_cache;
3128        print STDERR "Client certificate filename: ";
3129        STDERR->flush;
3130        chomp(my $filename = <STDIN>);
3131        $cred->cert_file($filename);
3132        $cred->may_save($may_save);
3133        $SVN::_Core::SVN_NO_ERROR;
3134}
3135
3136sub ssl_client_cert_pw {
3137        my ($cred, $realm, $may_save, $pool) = @_;
3138        $may_save = undef if $_no_auth_cache;
3139        $cred->password(_read_password("Password: ", $realm));
3140        $cred->may_save($may_save);
3141        $SVN::_Core::SVN_NO_ERROR;
3142}
3143
3144sub username {
3145        my ($cred, $realm, $may_save, $pool) = @_;
3146        $may_save = undef if $_no_auth_cache;
3147        if (defined $realm && length $realm) {
3148                print STDERR "Authentication realm: $realm\n";
3149        }
3150        my $username;
3151        if (defined $_username) {
3152                $username = $_username;
3153        } else {
3154                print STDERR "Username: ";
3155                STDERR->flush;
3156                chomp($username = <STDIN>);
3157        }
3158        $cred->username($username);
3159        $cred->may_save($may_save);
3160        $SVN::_Core::SVN_NO_ERROR;
3161}
3162
3163sub _read_password {
3164        my ($prompt, $realm) = @_;
3165        print STDERR $prompt;
3166        STDERR->flush;
3167        require Term::ReadKey;
3168        Term::ReadKey::ReadMode('noecho');
3169        my $password = '';
3170        while (defined(my $key = Term::ReadKey::ReadKey(0))) {
3171                last if $key =~ /[\012\015]/; # \n\r
3172                $password .= $key;
3173        }
3174        Term::ReadKey::ReadMode('restore');
3175        print STDERR "\n";
3176        STDERR->flush;
3177        $password;
3178}
3179
3180package SVN::Git::Fetcher;
3181use vars qw/@ISA/;
3182use strict;
3183use warnings;
3184use Carp qw/croak/;
3185use File::Temp qw/tempfile/;
3186use IO::File qw//;
3187
3188# file baton members: path, mode_a, mode_b, pool, fh, blob, base
3189sub new {
3190        my ($class, $git_svn) = @_;
3191        my $self = SVN::Delta::Editor->new;
3192        bless $self, $class;
3193        $self->{c} = $git_svn->{last_commit} if exists $git_svn->{last_commit};
3194        $self->{empty} = {};
3195        $self->{dir_prop} = {};
3196        $self->{file_prop} = {};
3197        $self->{absent_dir} = {};
3198        $self->{absent_file} = {};
3199        $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
3200        $self;
3201}
3202
3203sub set_path_strip {
3204        my ($self, $path) = @_;
3205        $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
3206}
3207
3208sub open_root {
3209        { path => '' };
3210}
3211
3212sub open_directory {
3213        my ($self, $path, $pb, $rev) = @_;
3214        { path => $path };
3215}
3216
3217sub git_path {
3218        my ($self, $path) = @_;
3219        if ($self->{path_strip}) {
3220                $path =~ s!$self->{path_strip}!! or
3221                  die "Failed to strip path '$path' ($self->{path_strip})\n";
3222        }
3223        $path;
3224}
3225
3226sub delete_entry {
3227        my ($self, $path, $rev, $pb) = @_;
3228
3229        my $gpath = $self->git_path($path);
3230        return undef if ($gpath eq '');
3231
3232        # remove entire directories.
3233        if (command('ls-tree', $self->{c}, '--', $gpath) =~ /^040000 tree/) {
3234                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3235                                                     -r --name-only -z/,
3236                                                     $self->{c}, '--', $gpath);
3237                local $/ = "\0";
3238                while (<$ls>) {
3239                        chomp;
3240                        $self->{gii}->remove($_);
3241                        print "\tD\t$_\n" unless $::_q;
3242                }
3243                print "\tD\t$gpath/\n" unless $::_q;
3244                command_close_pipe($ls, $ctx);
3245                $self->{empty}->{$path} = 0
3246        } else {
3247                $self->{gii}->remove($gpath);
3248                print "\tD\t$gpath\n" unless $::_q;
3249        }
3250        undef;
3251}
3252
3253sub open_file {
3254        my ($self, $path, $pb, $rev) = @_;
3255        my $gpath = $self->git_path($path);
3256        my ($mode, $blob) = (command('ls-tree', $self->{c}, '--', $gpath)
3257                             =~ /^(\d{6}) blob ([a-f\d]{40})\t/);
3258        unless (defined $mode && defined $blob) {
3259                die "$path was not found in commit $self->{c} (r$rev)\n";
3260        }
3261        { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
3262          pool => SVN::Pool->new, action => 'M' };
3263}
3264
3265sub add_file {
3266        my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
3267        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3268        delete $self->{empty}->{$dir};
3269        { path => $path, mode_a => 100644, mode_b => 100644,
3270          pool => SVN::Pool->new, action => 'A' };
3271}
3272
3273sub add_directory {
3274        my ($self, $path, $cp_path, $cp_rev) = @_;
3275        my $gpath = $self->git_path($path);
3276        if ($gpath eq '') {
3277                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3278                                                     -r --name-only -z/,
3279                                                     $self->{c});
3280                local $/ = "\0";
3281                while (<$ls>) {
3282                        chomp;
3283                        $self->{gii}->remove($_);
3284                        print "\tD\t$_\n" unless $::_q;
3285                }
3286                command_close_pipe($ls, $ctx);
3287                $self->{empty}->{$path} = 0;
3288        }
3289        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3290        delete $self->{empty}->{$dir};
3291        $self->{empty}->{$path} = 1;
3292        { path => $path };
3293}
3294
3295sub change_dir_prop {
3296        my ($self, $db, $prop, $value) = @_;
3297        $self->{dir_prop}->{$db->{path}} ||= {};
3298        $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
3299        undef;
3300}
3301
3302sub absent_directory {
3303        my ($self, $path, $pb) = @_;
3304        $self->{absent_dir}->{$pb->{path}} ||= [];
3305        push @{$self->{absent_dir}->{$pb->{path}}}, $path;
3306        undef;
3307}
3308
3309sub absent_file {
3310        my ($self, $path, $pb) = @_;
3311        $self->{absent_file}->{$pb->{path}} ||= [];
3312        push @{$self->{absent_file}->{$pb->{path}}}, $path;
3313        undef;
3314}
3315
3316sub change_file_prop {
3317        my ($self, $fb, $prop, $value) = @_;
3318        if ($prop eq 'svn:executable') {
3319                if ($fb->{mode_b} != 120000) {
3320                        $fb->{mode_b} = defined $value ? 100755 : 100644;
3321                }
3322        } elsif ($prop eq 'svn:special') {
3323                $fb->{mode_b} = defined $value ? 120000 : 100644;
3324        } else {
3325                $self->{file_prop}->{$fb->{path}} ||= {};
3326                $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
3327        }
3328        undef;
3329}
3330
3331sub apply_textdelta {
3332        my ($self, $fb, $exp) = @_;
3333        my $fh = Git::temp_acquire('svn_delta');
3334        # $fh gets auto-closed() by SVN::TxDelta::apply(),
3335        # (but $base does not,) so dup() it for reading in close_file
3336        open my $dup, '<&', $fh or croak $!;
3337        my $base = Git::temp_acquire('git_blob');
3338        if ($fb->{blob}) {
3339                print $base 'link ' if ($fb->{mode_a} == 120000);
3340                my $size = $::_repository->cat_blob($fb->{blob}, $base);
3341                die "Failed to read object $fb->{blob}" if ($size < 0);
3342
3343                if (defined $exp) {
3344                        seek $base, 0, 0 or croak $!;
3345                        my $got = ::md5sum($base);
3346                        die "Checksum mismatch: $fb->{path} $fb->{blob}\n",
3347                            "expected: $exp\n",
3348                            "     got: $got\n" if ($got ne $exp);
3349                }
3350        }
3351        seek $base, 0, 0 or croak $!;
3352        $fb->{fh} = $fh;
3353        $fb->{base} = $base;
3354        [ SVN::TxDelta::apply($base, $dup, undef, $fb->{path}, $fb->{pool}) ];
3355}
3356
3357sub close_file {
3358        my ($self, $fb, $exp) = @_;
3359        my $hash;
3360        my $path = $self->git_path($fb->{path});
3361        if (my $fh = $fb->{fh}) {
3362                if (defined $exp) {
3363                        seek($fh, 0, 0) or croak $!;
3364                        my $got = ::md5sum($fh);
3365                        if ($got ne $exp) {
3366                                die "Checksum mismatch: $path\n",
3367                                    "expected: $exp\n    got: $got\n";
3368                        }
3369                }
3370                if ($fb->{mode_b} == 120000) {
3371                        sysseek($fh, 0, 0) or croak $!;
3372                        sysread($fh, my $buf, 5) == 5 or croak $!;
3373
3374                        unless ($buf eq 'link ') {
3375                                warn "$path has mode 120000",
3376                                                " but is not a link\n";
3377                        } else {
3378                                my $tmp_fh = Git::temp_acquire('svn_hash');
3379                                my $res;
3380                                while ($res = sysread($fh, my $str, 1024)) {
3381                                        my $out = syswrite($tmp_fh, $str, $res);
3382                                        defined($out) && $out == $res
3383                                                or croak("write ",
3384                                                        Git::temp_path($tmp_fh),
3385                                                        ": $!\n");
3386                                }
3387                                defined $res or croak $!;
3388
3389                                ($fh, $tmp_fh) = ($tmp_fh, $fh);
3390                                Git::temp_release($tmp_fh, 1);
3391                        }
3392                }
3393
3394                $hash = $::_repository->hash_and_insert_object(
3395                                Git::temp_path($fh));
3396                $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
3397
3398                Git::temp_release($fb->{base}, 1);
3399                Git::temp_release($fh, 1);
3400        } else {
3401                $hash = $fb->{blob} or die "no blob information\n";
3402        }
3403        $fb->{pool}->clear;
3404        $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
3405        print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
3406        undef;
3407}
3408
3409sub abort_edit {
3410        my $self = shift;
3411        $self->{nr} = $self->{gii}->{nr};
3412        delete $self->{gii};
3413        $self->SUPER::abort_edit(@_);
3414}
3415
3416sub close_edit {
3417        my $self = shift;
3418        $self->{git_commit_ok} = 1;
3419        $self->{nr} = $self->{gii}->{nr};
3420        delete $self->{gii};
3421        $self->SUPER::close_edit(@_);
3422}
3423
3424package SVN::Git::Editor;
3425use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
3426use strict;
3427use warnings;
3428use Carp qw/croak/;
3429use IO::File;
3430
3431sub new {
3432        my ($class, $opts) = @_;
3433        foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
3434                die "$_ required!\n" unless (defined $opts->{$_});
3435        }
3436
3437        my $pool = SVN::Pool->new;
3438        my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
3439        my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
3440                                     $opts->{r}, $mods);
3441
3442        # $opts->{ra} functions should not be used after this:
3443        my @ce  = $opts->{ra}->get_commit_editor($opts->{log},
3444                                                $opts->{editor_cb}, $pool);
3445        my $self = SVN::Delta::Editor->new(@ce, $pool);
3446        bless $self, $class;
3447        foreach (qw/svn_path r tree_a tree_b/) {
3448                $self->{$_} = $opts->{$_};
3449        }
3450        $self->{url} = $opts->{ra}->{url};
3451        $self->{mods} = $mods;
3452        $self->{types} = $types;
3453        $self->{pool} = $pool;
3454        $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
3455        $self->{rm} = { };
3456        $self->{path_prefix} = length $self->{svn_path} ?
3457                               "$self->{svn_path}/" : '';
3458        $self->{config} = $opts->{config};
3459        return $self;
3460}
3461
3462sub generate_diff {
3463        my ($tree_a, $tree_b) = @_;
3464        my @diff_tree = qw(diff-tree -z -r);
3465        if ($_cp_similarity) {
3466                push @diff_tree, "-C$_cp_similarity";
3467        } else {
3468                push @diff_tree, '-C';
3469        }
3470        push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
3471        push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
3472        push @diff_tree, $tree_a, $tree_b;
3473        my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
3474        local $/ = "\0";
3475        my $state = 'meta';
3476        my @mods;
3477        while (<$diff_fh>) {
3478                chomp $_; # this gets rid of the trailing "\0"
3479                if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
3480                                        ($::sha1)\s($::sha1)\s
3481                                        ([MTCRAD])\d*$/xo) {
3482                        push @mods, {   mode_a => $1, mode_b => $2,
3483                                        sha1_a => $3, sha1_b => $4,
3484                                        chg => $5 };
3485                        if ($5 =~ /^(?:C|R)$/) {
3486                                $state = 'file_a';
3487                        } else {
3488                                $state = 'file_b';
3489                        }
3490                } elsif ($state eq 'file_a') {
3491                        my $x = $mods[$#mods] or croak "Empty array\n";
3492                        if ($x->{chg} !~ /^(?:C|R)$/) {
3493                                croak "Error parsing $_, $x->{chg}\n";
3494                        }
3495                        $x->{file_a} = $_;
3496                        $state = 'file_b';
3497                } elsif ($state eq 'file_b') {
3498                        my $x = $mods[$#mods] or croak "Empty array\n";
3499                        if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
3500                                croak "Error parsing $_, $x->{chg}\n";
3501                        }
3502                        if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
3503                                croak "Error parsing $_, $x->{chg}\n";
3504                        }
3505                        $x->{file_b} = $_;
3506                        $state = 'meta';
3507                } else {
3508                        croak "Error parsing $_\n";
3509                }
3510        }
3511        command_close_pipe($diff_fh, $ctx);
3512        \@mods;
3513}
3514
3515sub check_diff_paths {
3516        my ($ra, $pfx, $rev, $mods) = @_;
3517        my %types;
3518        $pfx .= '/' if length $pfx;
3519
3520        sub type_diff_paths {
3521                my ($ra, $types, $path, $rev) = @_;
3522                my @p = split m#/+#, $path;
3523                my $c = shift @p;
3524                unless (defined $types->{$c}) {
3525                        $types->{$c} = $ra->check_path($c, $rev);
3526                }
3527                while (@p) {
3528                        $c .= '/' . shift @p;
3529                        next if defined $types->{$c};
3530                        $types->{$c} = $ra->check_path($c, $rev);
3531                }
3532        }
3533
3534        foreach my $m (@$mods) {
3535                foreach my $f (qw/file_a file_b/) {
3536                        next unless defined $m->{$f};
3537                        my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
3538                        if (length $pfx.$dir && ! defined $types{$dir}) {
3539                                type_diff_paths($ra, \%types, $pfx.$dir, $rev);
3540                        }
3541                }
3542        }
3543        \%types;
3544}
3545
3546sub split_path {
3547        return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
3548}
3549
3550sub repo_path {
3551        my ($self, $path) = @_;
3552        $self->{path_prefix}.(defined $path ? $path : '');
3553}
3554
3555sub url_path {
3556        my ($self, $path) = @_;
3557        if ($self->{url} =~ m#^https?://#) {
3558                $path =~ s/([^~a-zA-Z0-9_.-])/uc sprintf("%%%02x",ord($1))/eg;
3559        }
3560        $self->{url} . '/' . $self->repo_path($path);
3561}
3562
3563sub rmdirs {
3564        my ($self) = @_;
3565        my $rm = $self->{rm};
3566        delete $rm->{''}; # we never delete the url we're tracking
3567        return unless %$rm;
3568
3569        foreach (keys %$rm) {
3570                my @d = split m#/#, $_;
3571                my $c = shift @d;
3572                $rm->{$c} = 1;
3573                while (@d) {
3574                        $c .= '/' . shift @d;
3575                        $rm->{$c} = 1;
3576                }
3577        }
3578        delete $rm->{$self->{svn_path}};
3579        delete $rm->{''}; # we never delete the url we're tracking
3580        return unless %$rm;
3581
3582        my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
3583                                             $self->{tree_b});
3584        local $/ = "\0";
3585        while (<$fh>) {
3586                chomp;
3587                my @dn = split m#/#, $_;
3588                while (pop @dn) {
3589                        delete $rm->{join '/', @dn};
3590                }
3591                unless (%$rm) {
3592                        close $fh;
3593                        return;
3594                }
3595        }
3596        command_close_pipe($fh, $ctx);
3597
3598        my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
3599        foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
3600                $self->close_directory($bat->{$d}, $p);
3601                my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
3602                print "\tD+\t$d/\n" unless $::_q;
3603                $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
3604                delete $bat->{$d};
3605        }
3606}
3607
3608sub open_or_add_dir {
3609        my ($self, $full_path, $baton) = @_;
3610        my $t = $self->{types}->{$full_path};
3611        if (!defined $t) {
3612                die "$full_path not known in r$self->{r} or we have a bug!\n";
3613        }
3614        {
3615                no warnings 'once';
3616                # SVN::Node::none and SVN::Node::file are used only once,
3617                # so we're shutting up Perl's warnings about them.
3618                if ($t == $SVN::Node::none) {
3619                        return $self->add_directory($full_path, $baton,
3620                            undef, -1, $self->{pool});
3621                } elsif ($t == $SVN::Node::dir) {
3622                        return $self->open_directory($full_path, $baton,
3623                            $self->{r}, $self->{pool});
3624                } # no warnings 'once'
3625                print STDERR "$full_path already exists in repository at ",
3626                    "r$self->{r} and it is not a directory (",
3627                    ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
3628        } # no warnings 'once'
3629        exit 1;
3630}
3631
3632sub ensure_path {
3633        my ($self, $path) = @_;
3634        my $bat = $self->{bat};
3635        my $repo_path = $self->repo_path($path);
3636        return $bat->{''} unless (length $repo_path);
3637        my @p = split m#/+#, $repo_path;
3638        my $c = shift @p;
3639        $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
3640        while (@p) {
3641                my $c0 = $c;
3642                $c .= '/' . shift @p;
3643                $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
3644        }
3645        return $bat->{$c};
3646}
3647
3648# Subroutine to convert a globbing pattern to a regular expression.
3649# From perl cookbook.
3650sub glob2pat {
3651        my $globstr = shift;
3652        my %patmap = ('*' => '.*', '?' => '.', '[' => '[', ']' => ']');
3653        $globstr =~ s{(.)} { $patmap{$1} || "\Q$1" }ge;
3654        return '^' . $globstr . '$';
3655}
3656
3657sub check_autoprop {
3658        my ($self, $pattern, $properties, $file, $fbat) = @_;
3659        # Convert the globbing pattern to a regular expression.
3660        my $regex = glob2pat($pattern);
3661        # Check if the pattern matches the file name.
3662        if($file =~ m/($regex)/) {
3663                # Parse the list of properties to set.
3664                my @props = split(/;/, $properties);
3665                foreach my $prop (@props) {
3666                        # Parse 'name=value' syntax and set the property.
3667                        if ($prop =~ /([^=]+)=(.*)/) {
3668                                my ($n,$v) = ($1,$2);
3669                                for ($n, $v) {
3670                                        s/^\s+//; s/\s+$//;
3671                                }
3672                                $self->change_file_prop($fbat, $n, $v);
3673                        }
3674                }
3675        }
3676}
3677
3678sub apply_autoprops {
3679        my ($self, $file, $fbat) = @_;
3680        my $conf_t = ${$self->{config}}{'config'};
3681        no warnings 'once';
3682        # Check [miscellany]/enable-auto-props in svn configuration.
3683        if (SVN::_Core::svn_config_get_bool(
3684                $conf_t,
3685                $SVN::_Core::SVN_CONFIG_SECTION_MISCELLANY,
3686                $SVN::_Core::SVN_CONFIG_OPTION_ENABLE_AUTO_PROPS,
3687                0)) {
3688                # Auto-props are enabled.  Enumerate them to look for matches.
3689                my $callback = sub {
3690                        $self->check_autoprop($_[0], $_[1], $file, $fbat);
3691                };
3692                SVN::_Core::svn_config_enumerate(
3693                        $conf_t,
3694                        $SVN::_Core::SVN_CONFIG_SECTION_AUTO_PROPS,
3695                        $callback);
3696        }
3697}
3698
3699sub A {
3700        my ($self, $m) = @_;
3701        my ($dir, $file) = split_path($m->{file_b});
3702        my $pbat = $self->ensure_path($dir);
3703        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3704                                        undef, -1);
3705        print "\tA\t$m->{file_b}\n" unless $::_q;
3706        $self->apply_autoprops($file, $fbat);
3707        $self->chg_file($fbat, $m);
3708        $self->close_file($fbat,undef,$self->{pool});
3709}
3710
3711sub C {
3712        my ($self, $m) = @_;
3713        my ($dir, $file) = split_path($m->{file_b});
3714        my $pbat = $self->ensure_path($dir);
3715        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3716                                $self->url_path($m->{file_a}), $self->{r});
3717        print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
3718        $self->chg_file($fbat, $m);
3719        $self->close_file($fbat,undef,$self->{pool});
3720}
3721
3722sub delete_entry {
3723        my ($self, $path, $pbat) = @_;
3724        my $rpath = $self->repo_path($path);
3725        my ($dir, $file) = split_path($rpath);
3726        $self->{rm}->{$dir} = 1;
3727        $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
3728}
3729
3730sub R {
3731        my ($self, $m) = @_;
3732        my ($dir, $file) = split_path($m->{file_b});
3733        my $pbat = $self->ensure_path($dir);
3734        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3735                                $self->url_path($m->{file_a}), $self->{r});
3736        print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
3737        $self->apply_autoprops($file, $fbat);
3738        $self->chg_file($fbat, $m);
3739        $self->close_file($fbat,undef,$self->{pool});
3740
3741        ($dir, $file) = split_path($m->{file_a});
3742        $pbat = $self->ensure_path($dir);
3743        $self->delete_entry($m->{file_a}, $pbat);
3744}
3745
3746sub M {
3747        my ($self, $m) = @_;
3748        my ($dir, $file) = split_path($m->{file_b});
3749        my $pbat = $self->ensure_path($dir);
3750        my $fbat = $self->open_file($self->repo_path($m->{file_b}),
3751                                $pbat,$self->{r},$self->{pool});
3752        print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
3753        $self->chg_file($fbat, $m);
3754        $self->close_file($fbat,undef,$self->{pool});
3755}
3756
3757sub T { shift->M(@_) }
3758
3759sub change_file_prop {
3760        my ($self, $fbat, $pname, $pval) = @_;
3761        $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
3762}
3763
3764sub _chg_file_get_blob ($$$$) {
3765        my ($self, $fbat, $m, $which) = @_;
3766        my $fh = Git::temp_acquire("git_blob_$which");
3767        if ($m->{"mode_$which"} =~ /^120/) {
3768                print $fh 'link ' or croak $!;
3769                $self->change_file_prop($fbat,'svn:special','*');
3770        } elsif ($m->{mode_a} =~ /^120/ && $m->{"mode_$which"} !~ /^120/) {
3771                $self->change_file_prop($fbat,'svn:special',undef);
3772        }
3773        my $blob = $m->{"sha1_$which"};
3774        return ($fh,) if ($blob =~ /^0{40}$/);
3775        my $size = $::_repository->cat_blob($blob, $fh);
3776        croak "Failed to read object $blob" if ($size < 0);
3777        $fh->flush == 0 or croak $!;
3778        seek $fh, 0, 0 or croak $!;
3779
3780        my $exp = ::md5sum($fh);
3781        seek $fh, 0, 0 or croak $!;
3782        return ($fh, $exp);
3783}
3784
3785sub chg_file {
3786        my ($self, $fbat, $m) = @_;
3787        if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
3788                $self->change_file_prop($fbat,'svn:executable','*');
3789        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
3790                $self->change_file_prop($fbat,'svn:executable',undef);
3791        }
3792        my ($fh_a, $exp_a) = _chg_file_get_blob $self, $fbat, $m, 'a';
3793        my ($fh_b, $exp_b) = _chg_file_get_blob $self, $fbat, $m, 'b';
3794        my $pool = SVN::Pool->new;
3795        my $atd = $self->apply_textdelta($fbat, $exp_a, $pool);
3796        if (-s $fh_a) {
3797                my $txstream = SVN::TxDelta::new ($fh_a, $fh_b, $pool);
3798                my $res = SVN::TxDelta::send_txstream($txstream, @$atd, $pool);
3799                if (defined $res) {
3800                        die "Unexpected result from send_txstream: $res\n",
3801                            "(SVN::Core::VERSION: $SVN::Core::VERSION)\n";
3802                }
3803        } else {
3804                my $got = SVN::TxDelta::send_stream($fh_b, @$atd, $pool);
3805                die "Checksum mismatch\nexpected: $exp_b\ngot: $got\n"
3806                    if ($got ne $exp_b);
3807        }
3808        Git::temp_release($fh_b, 1);
3809        Git::temp_release($fh_a, 1);
3810        $pool->clear;
3811}
3812
3813sub D {
3814        my ($self, $m) = @_;
3815        my ($dir, $file) = split_path($m->{file_b});
3816        my $pbat = $self->ensure_path($dir);
3817        print "\tD\t$m->{file_b}\n" unless $::_q;
3818        $self->delete_entry($m->{file_b}, $pbat);
3819}
3820
3821sub close_edit {
3822        my ($self) = @_;
3823        my ($p,$bat) = ($self->{pool}, $self->{bat});
3824        foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
3825                next if $_ eq '';
3826                $self->close_directory($bat->{$_}, $p);
3827        }
3828        $self->close_directory($bat->{''}, $p);
3829        $self->SUPER::close_edit($p);
3830        $p->clear;
3831}
3832
3833sub abort_edit {
3834        my ($self) = @_;
3835        $self->SUPER::abort_edit($self->{pool});
3836}
3837
3838sub DESTROY {
3839        my $self = shift;
3840        $self->SUPER::DESTROY(@_);
3841        $self->{pool}->clear;
3842}
3843
3844# this drives the editor
3845sub apply_diff {
3846        my ($self) = @_;
3847        my $mods = $self->{mods};
3848        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
3849        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
3850                my $f = $m->{chg};
3851                if (defined $o{$f}) {
3852                        $self->$f($m);
3853                } else {
3854                        fatal("Invalid change type: $f");
3855                }
3856        }
3857        $self->rmdirs if $_rmdir;
3858        if (@$mods == 0) {
3859                $self->abort_edit;
3860        } else {
3861                $self->close_edit;
3862        }
3863        return scalar @$mods;
3864}
3865
3866package Git::SVN::Ra;
3867use vars qw/@ISA $config_dir $_log_window_size/;
3868use strict;
3869use warnings;
3870my ($ra_invalid, $can_do_switch, %ignored_err, $RA);
3871
3872BEGIN {
3873        # enforce temporary pool usage for some simple functions
3874        no strict 'refs';
3875        for my $f (qw/rev_proplist get_latest_revnum get_uuid get_repos_root/) {
3876                my $SUPER = "SUPER::$f";
3877                *$f = sub {
3878                        my $self = shift;
3879                        my $pool = SVN::Pool->new;
3880                        my @ret = $self->$SUPER(@_,$pool);
3881                        $pool->clear;
3882                        wantarray ? @ret : $ret[0];
3883                };
3884        }
3885}
3886
3887sub _auth_providers () {
3888        [
3889          SVN::Client::get_simple_provider(),
3890          SVN::Client::get_ssl_server_trust_file_provider(),
3891          SVN::Client::get_simple_prompt_provider(
3892            \&Git::SVN::Prompt::simple, 2),
3893          SVN::Client::get_ssl_client_cert_file_provider(),
3894          SVN::Client::get_ssl_client_cert_prompt_provider(
3895            \&Git::SVN::Prompt::ssl_client_cert, 2),
3896          SVN::Client::get_ssl_client_cert_pw_file_provider(),
3897          SVN::Client::get_ssl_client_cert_pw_prompt_provider(
3898            \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
3899          SVN::Client::get_username_provider(),
3900          SVN::Client::get_ssl_server_trust_prompt_provider(
3901            \&Git::SVN::Prompt::ssl_server_trust),
3902          SVN::Client::get_username_prompt_provider(
3903            \&Git::SVN::Prompt::username, 2)
3904        ]
3905}
3906
3907sub escape_uri_only {
3908        my ($uri) = @_;
3909        my @tmp;
3910        foreach (split m{/}, $uri) {
3911                s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
3912                push @tmp, $_;
3913        }
3914        join('/', @tmp);
3915}
3916
3917sub escape_url {
3918        my ($url) = @_;
3919        if ($url =~ m#^(https?)://([^/]+)(.*)$#) {
3920                my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
3921                $url = "$scheme://$domain$uri";
3922        }
3923        $url;
3924}
3925
3926sub new {
3927        my ($class, $url) = @_;
3928        $url =~ s!/+$!!;
3929        return $RA if ($RA && $RA->{url} eq $url);
3930
3931        SVN::_Core::svn_config_ensure($config_dir, undef);
3932        my ($baton, $callbacks) = SVN::Core::auth_open_helper(_auth_providers);
3933        my $config = SVN::Core::config_get_config($config_dir);
3934        $RA = undef;
3935        my $dont_store_passwords = 1;
3936        my $conf_t = ${$config}{'config'};
3937        {
3938                no warnings 'once';
3939                # The usage of $SVN::_Core::SVN_CONFIG_* variables
3940                # produces warnings that variables are used only once.
3941                # I had not found the better way to shut them up, so
3942                # the warnings of type 'once' are disabled in this block.
3943                if (SVN::_Core::svn_config_get_bool($conf_t,
3944                    $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
3945                    $SVN::_Core::SVN_CONFIG_OPTION_STORE_PASSWORDS,
3946                    1) == 0) {
3947                        SVN::_Core::svn_auth_set_parameter($baton,
3948                            $SVN::_Core::SVN_AUTH_PARAM_DONT_STORE_PASSWORDS,
3949                            bless (\$dont_store_passwords, "_p_void"));
3950                }
3951                if (SVN::_Core::svn_config_get_bool($conf_t,
3952                    $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
3953                    $SVN::_Core::SVN_CONFIG_OPTION_STORE_AUTH_CREDS,
3954                    1) == 0) {
3955                        $Git::SVN::Prompt::_no_auth_cache = 1;
3956                }
3957        } # no warnings 'once'
3958        my $self = SVN::Ra->new(url => escape_url($url), auth => $baton,
3959                              config => $config,
3960                              pool => SVN::Pool->new,
3961                              auth_provider_callbacks => $callbacks);
3962        $self->{url} = $url;
3963        $self->{svn_path} = $url;
3964        $self->{repos_root} = $self->get_repos_root;
3965        $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
3966        $self->{cache} = { check_path => { r => 0, data => {} },
3967                           get_dir => { r => 0, data => {} } };
3968        $RA = bless $self, $class;
3969}
3970
3971sub check_path {
3972        my ($self, $path, $r) = @_;
3973        my $cache = $self->{cache}->{check_path};
3974        if ($r == $cache->{r} && exists $cache->{data}->{$path}) {
3975                return $cache->{data}->{$path};
3976        }
3977        my $pool = SVN::Pool->new;
3978        my $t = $self->SUPER::check_path($path, $r, $pool);
3979        $pool->clear;
3980        if ($r != $cache->{r}) {
3981                %{$cache->{data}} = ();
3982                $cache->{r} = $r;
3983        }
3984        $cache->{data}->{$path} = $t;
3985}
3986
3987sub get_dir {
3988        my ($self, $dir, $r) = @_;
3989        my $cache = $self->{cache}->{get_dir};
3990        if ($r == $cache->{r}) {
3991                if (my $x = $cache->{data}->{$dir}) {
3992                        return wantarray ? @$x : $x->[0];
3993                }
3994        }
3995        my $pool = SVN::Pool->new;
3996        my ($d, undef, $props) = $self->SUPER::get_dir($dir, $r, $pool);
3997        my %dirents = map { $_ => { kind => $d->{$_}->kind } } keys %$d;
3998        $pool->clear;
3999        if ($r != $cache->{r}) {
4000                %{$cache->{data}} = ();
4001                $cache->{r} = $r;
4002        }
4003        $cache->{data}->{$dir} = [ \%dirents, $r, $props ];
4004        wantarray ? (\%dirents, $r, $props) : \%dirents;
4005}
4006
4007sub DESTROY {
4008        # do not call the real DESTROY since we store ourselves in $RA
4009}
4010
4011sub get_log {
4012        my ($self, @args) = @_;
4013        my $pool = SVN::Pool->new;
4014        splice(@args, 3, 1) if ($SVN::Core::VERSION le '1.2.0');
4015        my $ret = $self->SUPER::get_log(@args, $pool);
4016        $pool->clear;
4017        $ret;
4018}
4019
4020sub trees_match {
4021        my ($self, $url1, $rev1, $url2, $rev2) = @_;
4022        my $ctx = SVN::Client->new(auth => _auth_providers);
4023        my $out = IO::File->new_tmpfile;
4024
4025        # older SVN (1.1.x) doesn't take $pool as the last parameter for
4026        # $ctx->diff(), so we'll create a default one
4027        my $pool = SVN::Pool->new_default_sub;
4028
4029        $ra_invalid = 1; # this will open a new SVN::Ra connection to $url1
4030        $ctx->diff([], $url1, $rev1, $url2, $rev2, 1, 1, 0, $out, $out);
4031        $out->flush;
4032        my $ret = (($out->stat)[7] == 0);
4033        close $out or croak $!;
4034
4035        $ret;
4036}
4037
4038sub get_commit_editor {
4039        my ($self, $log, $cb, $pool) = @_;
4040        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
4041        $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
4042}
4043
4044sub gs_do_update {
4045        my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
4046        my $new = ($rev_a == $rev_b);
4047        my $path = $gs->{path};
4048
4049        if ($new && -e $gs->{index}) {
4050                unlink $gs->{index} or die
4051                  "Couldn't unlink index: $gs->{index}: $!\n";
4052        }
4053        my $pool = SVN::Pool->new;
4054        $editor->set_path_strip($path);
4055        my (@pc) = split m#/#, $path;
4056        my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
4057                                        1, $editor, $pool);
4058        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
4059
4060        # Since we can't rely on svn_ra_reparent being available, we'll
4061        # just have to do some magic with set_path to make it so
4062        # we only want a partial path.
4063        my $sp = '';
4064        my $final = join('/', @pc);
4065        while (@pc) {
4066                $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
4067                $sp .= '/' if length $sp;
4068                $sp .= shift @pc;
4069        }
4070        die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
4071
4072        $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
4073
4074        $reporter->finish_report($pool);
4075        $pool->clear;
4076        $editor->{git_commit_ok};
4077}
4078
4079# this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
4080# svn_ra_reparent didn't work before 1.4)
4081sub gs_do_switch {
4082        my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
4083        my $path = $gs->{path};
4084        my $pool = SVN::Pool->new;
4085
4086        my $full_url = $self->{url};
4087        my $old_url = $full_url;
4088        $full_url .= '/' . escape_uri_only($path) if length $path;
4089        my ($ra, $reparented);
4090
4091        if ($old_url =~ m#^svn(\+ssh)?://#) {
4092                $_[0] = undef;
4093                $self = undef;
4094                $RA = undef;
4095                $ra = Git::SVN::Ra->new($full_url);
4096                $ra_invalid = 1;
4097        } elsif ($old_url ne $full_url) {
4098                SVN::_Ra::svn_ra_reparent($self->{session}, $full_url, $pool);
4099                $self->{url} = $full_url;
4100                $reparented = 1;
4101        }
4102
4103        $ra ||= $self;
4104        $url_b = escape_url($url_b);
4105        my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
4106        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
4107        $reporter->set_path('', $rev_a, 0, @lock, $pool);
4108        $reporter->finish_report($pool);
4109
4110        if ($reparented) {
4111                SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
4112                $self->{url} = $old_url;
4113        }
4114
4115        $pool->clear;
4116        $editor->{git_commit_ok};
4117}
4118
4119sub longest_common_path {
4120        my ($gsv, $globs) = @_;
4121        my %common;
4122        my $common_max = scalar @$gsv;
4123
4124        foreach my $gs (@$gsv) {
4125                my @tmp = split m#/#, $gs->{path};
4126                my $p = '';
4127                foreach (@tmp) {
4128                        $p .= length($p) ? "/$_" : $_;
4129                        $common{$p} ||= 0;
4130                        $common{$p}++;
4131                }
4132        }
4133        $globs ||= [];
4134        $common_max += scalar @$globs;
4135        foreach my $glob (@$globs) {
4136                my @tmp = split m#/#, $glob->{path}->{left};
4137                my $p = '';
4138                foreach (@tmp) {
4139                        $p .= length($p) ? "/$_" : $_;
4140                        $common{$p} ||= 0;
4141                        $common{$p}++;
4142                }
4143        }
4144
4145        my $longest_path = '';
4146        foreach (sort {length $b <=> length $a} keys %common) {
4147                if ($common{$_} == $common_max) {
4148                        $longest_path = $_;
4149                        last;
4150                }
4151        }
4152        $longest_path;
4153}
4154
4155sub gs_fetch_loop_common {
4156        my ($self, $base, $head, $gsv, $globs) = @_;
4157        return if ($base > $head);
4158        my $inc = $_log_window_size;
4159        my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
4160        my $longest_path = longest_common_path($gsv, $globs);
4161        my $ra_url = $self->{url};
4162        while (1) {
4163                my %revs;
4164                my $err;
4165                my $err_handler = $SVN::Error::handler;
4166                $SVN::Error::handler = sub {
4167                        ($err) = @_;
4168                        skip_unknown_revs($err);
4169                };
4170                sub _cb {
4171                        my ($paths, $r, $author, $date, $log) = @_;
4172                        [ dup_changed_paths($paths),
4173                          { author => $author, date => $date, log => $log } ];
4174                }
4175                $self->get_log([$longest_path], $min, $max, 0, 1, 1,
4176                               sub { $revs{$_[1]} = _cb(@_) });
4177                if ($err && $max >= $head) {
4178                        print STDERR "Path '$longest_path' ",
4179                                     "was probably deleted:\n",
4180                                     $err->expanded_message,
4181                                     "\nWill attempt to follow ",
4182                                     "revisions r$min .. r$max ",
4183                                     "committed before the deletion\n";
4184                        my $hi = $max;
4185                        while (--$hi >= $min) {
4186                                my $ok;
4187                                $self->get_log([$longest_path], $min, $hi,
4188                                               0, 1, 1, sub {
4189                                               $ok ||= $_[1];
4190                                               $revs{$_[1]} = _cb(@_) });
4191                                if ($ok) {
4192                                        print STDERR "r$min .. r$ok OK\n";
4193                                        last;
4194                                }
4195                        }
4196                }
4197                $SVN::Error::handler = $err_handler;
4198
4199                my %exists = map { $_->{path} => $_ } @$gsv;
4200                foreach my $r (sort {$a <=> $b} keys %revs) {
4201                        my ($paths, $logged) = @{$revs{$r}};
4202
4203                        foreach my $gs ($self->match_globs(\%exists, $paths,
4204                                                           $globs, $r)) {
4205                                if ($gs->rev_map_max >= $r) {
4206                                        next;
4207                                }
4208                                next unless $gs->match_paths($paths, $r);
4209                                $gs->{logged_rev_props} = $logged;
4210                                if (my $last_commit = $gs->last_commit) {
4211                                        $gs->assert_index_clean($last_commit);
4212                                }
4213                                my $log_entry = $gs->do_fetch($paths, $r);
4214                                if ($log_entry) {
4215                                        $gs->do_git_commit($log_entry);
4216                                }
4217                                $INDEX_FILES{$gs->{index}} = 1;
4218                        }
4219                        foreach my $g (@$globs) {
4220                                my $k = "svn-remote.$g->{remote}." .
4221                                        "$g->{t}-maxRev";
4222                                Git::SVN::tmp_config($k, $r);
4223                        }
4224                        if ($ra_invalid) {
4225                                $_[0] = undef;
4226                                $self = undef;
4227                                $RA = undef;
4228                                $self = Git::SVN::Ra->new($ra_url);
4229                                $ra_invalid = undef;
4230                        }
4231                }
4232                # pre-fill the .rev_db since it'll eventually get filled in
4233                # with '0' x40 if something new gets committed
4234                foreach my $gs (@$gsv) {
4235                        next if $gs->rev_map_max >= $max;
4236                        next if defined $gs->rev_map_get($max);
4237                        $gs->rev_map_set($max, 0 x40);
4238                }
4239                foreach my $g (@$globs) {
4240                        my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
4241                        Git::SVN::tmp_config($k, $max);
4242                }
4243                last if $max >= $head;
4244                $min = $max + 1;
4245                $max += $inc;
4246                $max = $head if ($max > $head);
4247        }
4248        Git::SVN::gc();
4249}
4250
4251sub get_dir_globbed {
4252        my ($self, $left, $depth, $r) = @_;
4253
4254        my @x = eval { $self->get_dir($left, $r) };
4255        return unless scalar @x == 3;
4256        my $dirents = $x[0];
4257        my @finalents;
4258        foreach my $de (keys %$dirents) {
4259                next if $dirents->{$de}->{kind} != $SVN::Node::dir;
4260                if ($depth > 1) {
4261                        my @args = ("$left/$de", $depth - 1, $r);
4262                        foreach my $dir ($self->get_dir_globbed(@args)) {
4263                                push @finalents, "$de/$dir";
4264                        }
4265                } else {
4266                        push @finalents, $de;
4267                }
4268        }
4269        @finalents;
4270}
4271
4272sub match_globs {
4273        my ($self, $exists, $paths, $globs, $r) = @_;
4274
4275        sub get_dir_check {
4276                my ($self, $exists, $g, $r) = @_;
4277
4278                my @dirs = $self->get_dir_globbed($g->{path}->{left},
4279                                                  $g->{path}->{depth},
4280                                                  $r);
4281
4282                foreach my $de (@dirs) {
4283                        my $p = $g->{path}->full_path($de);
4284                        next if $exists->{$p};
4285                        next if (length $g->{path}->{right} &&
4286                                 ($self->check_path($p, $r) !=
4287                                  $SVN::Node::dir));
4288                        $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
4289                                         $g->{ref}->full_path($de), 1);
4290                }
4291        }
4292        foreach my $g (@$globs) {
4293                if (my $path = $paths->{"/$g->{path}->{left}"}) {
4294                        if ($path->{action} =~ /^[AR]$/) {
4295                                get_dir_check($self, $exists, $g, $r);
4296                        }
4297                }
4298                foreach (keys %$paths) {
4299                        if (/$g->{path}->{left_regex}/ &&
4300                            !/$g->{path}->{regex}/) {
4301                                next if $paths->{$_}->{action} !~ /^[AR]$/;
4302                                get_dir_check($self, $exists, $g, $r);
4303                        }
4304                        next unless /$g->{path}->{regex}/;
4305                        my $p = $1;
4306                        my $pathname = $g->{path}->full_path($p);
4307                        next if $exists->{$pathname};
4308                        next if ($self->check_path($pathname, $r) !=
4309                                 $SVN::Node::dir);
4310                        $exists->{$pathname} = Git::SVN->init(
4311                                              $self->{url}, $pathname, undef,
4312                                              $g->{ref}->full_path($p), 1);
4313                }
4314                my $c = '';
4315                foreach (split m#/#, $g->{path}->{left}) {
4316                        $c .= "/$_";
4317                        next unless ($paths->{$c} &&
4318                                     ($paths->{$c}->{action} =~ /^[AR]$/));
4319                        get_dir_check($self, $exists, $g, $r);
4320                }
4321        }
4322        values %$exists;
4323}
4324
4325sub minimize_url {
4326        my ($self) = @_;
4327        return $self->{url} if ($self->{url} eq $self->{repos_root});
4328        my $url = $self->{repos_root};
4329        my @components = split(m!/!, $self->{svn_path});
4330        my $c = '';
4331        do {
4332                $url .= "/$c" if length $c;
4333                eval { (ref $self)->new($url)->get_latest_revnum };
4334        } while ($@ && ($c = shift @components));
4335        $url;
4336}
4337
4338sub can_do_switch {
4339        my $self = shift;
4340        unless (defined $can_do_switch) {
4341                my $pool = SVN::Pool->new;
4342                my $rep = eval {
4343                        $self->do_switch(1, '', 0, $self->{url},
4344                                         SVN::Delta::Editor->new, $pool);
4345                };
4346                if ($@) {
4347                        $can_do_switch = 0;
4348                } else {
4349                        $rep->abort_report($pool);
4350                        $can_do_switch = 1;
4351                }
4352                $pool->clear;
4353        }
4354        $can_do_switch;
4355}
4356
4357sub skip_unknown_revs {
4358        my ($err) = @_;
4359        my $errno = $err->apr_err();
4360        # Maybe the branch we're tracking didn't
4361        # exist when the repo started, so it's
4362        # not an error if it doesn't, just continue
4363        #
4364        # Wonderfully consistent library, eh?
4365        # 160013 - svn:// and file://
4366        # 175002 - http(s)://
4367        # 175007 - http(s):// (this repo required authorization, too...)
4368        #   More codes may be discovered later...
4369        if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
4370                my $err_key = $err->expanded_message;
4371                # revision numbers change every time, filter them out
4372                $err_key =~ s/\d+/\0/g;
4373                $err_key = "$errno\0$err_key";
4374                unless ($ignored_err{$err_key}) {
4375                        warn "W: Ignoring error from SVN, path probably ",
4376                             "does not exist: ($errno): ",
4377                             $err->expanded_message,"\n";
4378                        warn "W: Do not be alarmed at the above message ",
4379                             "git-svn is just searching aggressively for ",
4380                             "old history.\n",
4381                             "This may take a while on large repositories\n";
4382                        $ignored_err{$err_key} = 1;
4383                }
4384                return;
4385        }
4386        die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
4387}
4388
4389# svn_log_changed_path_t objects passed to get_log are likely to be
4390# overwritten even if only the refs are copied to an external variable,
4391# so we should dup the structures in their entirety.  Using an externally
4392# passed pool (instead of our temporary and quickly cleared pool in
4393# Git::SVN::Ra) does not help matters at all...
4394sub dup_changed_paths {
4395        my ($paths) = @_;
4396        return undef unless $paths;
4397        my %ret;
4398        foreach my $p (keys %$paths) {
4399                my $i = $paths->{$p};
4400                my %s = map { $_ => $i->$_ }
4401                              qw/copyfrom_path copyfrom_rev action/;
4402                $ret{$p} = \%s;
4403        }
4404        \%ret;
4405}
4406
4407package Git::SVN::Log;
4408use strict;
4409use warnings;
4410use POSIX qw/strftime/;
4411use constant commit_log_separator => ('-' x 72) . "\n";
4412use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
4413            %rusers $show_commit $incremental/;
4414my $l_fmt;
4415
4416sub cmt_showable {
4417        my ($c) = @_;
4418        return 1 if defined $c->{r};
4419
4420        # big commit message got truncated by the 16k pretty buffer in rev-list
4421        if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
4422                                $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
4423                @{$c->{l}} = ();
4424                my @log = command(qw/cat-file commit/, $c->{c});
4425
4426                # shift off the headers
4427                shift @log while ($log[0] ne '');
4428                shift @log;
4429
4430                # TODO: make $c->{l} not have a trailing newline in the future
4431                @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
4432
4433                (undef, $c->{r}, undef) = ::extract_metadata(
4434                                (grep(/^git-svn-id: /, @log))[-1]);
4435        }
4436        return defined $c->{r};
4437}
4438
4439sub log_use_color {
4440        return $color || Git->repository->get_colorbool('color.diff');
4441}
4442
4443sub git_svn_log_cmd {
4444        my ($r_min, $r_max, @args) = @_;
4445        my $head = 'HEAD';
4446        my (@files, @log_opts);
4447        foreach my $x (@args) {
4448                if ($x eq '--' || @files) {
4449                        push @files, $x;
4450                } else {
4451                        if (::verify_ref("$x^0")) {
4452                                $head = $x;
4453                        } else {
4454                                push @log_opts, $x;
4455                        }
4456                }
4457        }
4458
4459        my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
4460        $gs ||= Git::SVN->_new;
4461        my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
4462                   $gs->refname);
4463        push @cmd, '-r' unless $non_recursive;
4464        push @cmd, qw/--raw --name-status/ if $verbose;
4465        push @cmd, '--color' if log_use_color();
4466        push @cmd, @log_opts;
4467        if (defined $r_max && $r_max == $r_min) {
4468                push @cmd, '--max-count=1';
4469                if (my $c = $gs->rev_map_get($r_max)) {
4470                        push @cmd, $c;
4471                }
4472        } elsif (defined $r_max) {
4473                if ($r_max < $r_min) {
4474                        ($r_min, $r_max) = ($r_max, $r_min);
4475                }
4476                my (undef, $c_max) = $gs->find_rev_before($r_max, 1, $r_min);
4477                my (undef, $c_min) = $gs->find_rev_after($r_min, 1, $r_max);
4478                # If there are no commits in the range, both $c_max and $c_min
4479                # will be undefined.  If there is at least 1 commit in the
4480                # range, both will be defined.
4481                return () if !defined $c_min || !defined $c_max;
4482                if ($c_min eq $c_max) {
4483                        push @cmd, '--max-count=1', $c_min;
4484                } else {
4485                        push @cmd, '--boundary', "$c_min..$c_max";
4486                }
4487        }
4488        return (@cmd, @files);
4489}
4490
4491# adapted from pager.c
4492sub config_pager {
4493        $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
4494        if (!defined $pager) {
4495                $pager = 'less';
4496        } elsif (length $pager == 0 || $pager eq 'cat') {
4497                $pager = undef;
4498        }
4499        $ENV{GIT_PAGER_IN_USE} = defined($pager);
4500}
4501
4502sub run_pager {
4503        return unless -t *STDOUT && defined $pager;
4504        pipe my ($rfd, $wfd) or return;
4505        defined(my $pid = fork) or ::fatal "Can't fork: $!";
4506        if (!$pid) {
4507                open STDOUT, '>&', $wfd or
4508                                     ::fatal "Can't redirect to stdout: $!";
4509                return;
4510        }
4511        open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!";
4512        $ENV{LESS} ||= 'FRSX';
4513        exec $pager or ::fatal "Can't run pager: $! ($pager)";
4514}
4515
4516sub format_svn_date {
4517        return strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)", localtime(shift));
4518}
4519
4520sub parse_git_date {
4521        my ($t, $tz) = @_;
4522        # Date::Parse isn't in the standard Perl distro :(
4523        if ($tz =~ s/^\+//) {
4524                $t += tz_to_s_offset($tz);
4525        } elsif ($tz =~ s/^\-//) {
4526                $t -= tz_to_s_offset($tz);
4527        }
4528        return $t;
4529}
4530
4531sub set_local_timezone {
4532        if (defined $TZ) {
4533                $ENV{TZ} = $TZ;
4534        } else {
4535                delete $ENV{TZ};
4536        }
4537}
4538
4539sub tz_to_s_offset {
4540        my ($tz) = @_;
4541        $tz =~ s/(\d\d)$//;
4542        return ($1 * 60) + ($tz * 3600);
4543}
4544
4545sub get_author_info {
4546        my ($dest, $author, $t, $tz) = @_;
4547        $author =~ s/(?:^\s*|\s*$)//g;
4548        $dest->{a_raw} = $author;
4549        my $au;
4550        if ($::_authors) {
4551                $au = $rusers{$author} || undef;
4552        }
4553        if (!$au) {
4554                ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
4555        }
4556        $dest->{t} = $t;
4557        $dest->{tz} = $tz;
4558        $dest->{a} = $au;
4559        $dest->{t_utc} = parse_git_date($t, $tz);
4560}
4561
4562sub process_commit {
4563        my ($c, $r_min, $r_max, $defer) = @_;
4564        if (defined $r_min && defined $r_max) {
4565                if ($r_min == $c->{r} && $r_min == $r_max) {
4566                        show_commit($c);
4567                        return 0;
4568                }
4569                return 1 if $r_min == $r_max;
4570                if ($r_min < $r_max) {
4571                        # we need to reverse the print order
4572                        return 0 if (defined $limit && --$limit < 0);
4573                        push @$defer, $c;
4574                        return 1;
4575                }
4576                if ($r_min != $r_max) {
4577                        return 1 if ($r_min < $c->{r});
4578                        return 1 if ($r_max > $c->{r});
4579                }
4580        }
4581        return 0 if (defined $limit && --$limit < 0);
4582        show_commit($c);
4583        return 1;
4584}
4585
4586sub show_commit {
4587        my $c = shift;
4588        if ($oneline) {
4589                my $x = "\n";
4590                if (my $l = $c->{l}) {
4591                        while ($l->[0] =~ /^\s*$/) { shift @$l }
4592                        $x = $l->[0];
4593                }
4594                $l_fmt ||= 'A' . length($c->{r});
4595                print 'r',pack($l_fmt, $c->{r}),' | ';
4596                print "$c->{c} | " if $show_commit;
4597                print $x;
4598        } else {
4599                show_commit_normal($c);
4600        }
4601}
4602
4603sub show_commit_changed_paths {
4604        my ($c) = @_;
4605        return unless $c->{changed};
4606        print "Changed paths:\n", @{$c->{changed}};
4607}
4608
4609sub show_commit_normal {
4610        my ($c) = @_;
4611        print commit_log_separator, "r$c->{r} | ";
4612        print "$c->{c} | " if $show_commit;
4613        print "$c->{a} | ", format_svn_date($c->{t_utc}), ' | ';
4614        my $nr_line = 0;
4615
4616        if (my $l = $c->{l}) {
4617                while ($l->[$#$l] eq "\n" && $#$l > 0
4618                                          && $l->[($#$l - 1)] eq "\n") {
4619                        pop @$l;
4620                }
4621                $nr_line = scalar @$l;
4622                if (!$nr_line) {
4623                        print "1 line\n\n\n";
4624                } else {
4625                        if ($nr_line == 1) {
4626                                $nr_line = '1 line';
4627                        } else {
4628                                $nr_line .= ' lines';
4629                        }
4630                        print $nr_line, "\n";
4631                        show_commit_changed_paths($c);
4632                        print "\n";
4633                        print $_ foreach @$l;
4634                }
4635        } else {
4636                print "1 line\n";
4637                show_commit_changed_paths($c);
4638                print "\n";
4639
4640        }
4641        foreach my $x (qw/raw stat diff/) {
4642                if ($c->{$x}) {
4643                        print "\n";
4644                        print $_ foreach @{$c->{$x}}
4645                }
4646        }
4647}
4648
4649sub cmd_show_log {
4650        my (@args) = @_;
4651        my ($r_min, $r_max);
4652        my $r_last = -1; # prevent dupes
4653        set_local_timezone();
4654        if (defined $::_revision) {
4655                if ($::_revision =~ /^(\d+):(\d+)$/) {
4656                        ($r_min, $r_max) = ($1, $2);
4657                } elsif ($::_revision =~ /^\d+$/) {
4658                        $r_min = $r_max = $::_revision;
4659                } else {
4660                        ::fatal "-r$::_revision is not supported, use ",
4661                                "standard 'git log' arguments instead";
4662                }
4663        }
4664
4665        config_pager();
4666        @args = git_svn_log_cmd($r_min, $r_max, @args);
4667        if (!@args) {
4668                print commit_log_separator unless $incremental || $oneline;
4669                return;
4670        }
4671        my $log = command_output_pipe(@args);
4672        run_pager();
4673        my (@k, $c, $d, $stat);
4674        my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
4675        while (<$log>) {
4676                if (/^${esc_color}commit -?($::sha1_short)/o) {
4677                        my $cmt = $1;
4678                        if ($c && cmt_showable($c) && $c->{r} != $r_last) {
4679                                $r_last = $c->{r};
4680                                process_commit($c, $r_min, $r_max, \@k) or
4681                                                                goto out;
4682                        }
4683                        $d = undef;
4684                        $c = { c => $cmt };
4685                } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
4686                        get_author_info($c, $1, $2, $3);
4687                } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
4688                        # ignore
4689                } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
4690                        push @{$c->{raw}}, $_;
4691                } elsif (/^${esc_color}[ACRMDT]\t/) {
4692                        # we could add $SVN->{svn_path} here, but that requires
4693                        # remote access at the moment (repo_path_split)...
4694                        s#^(${esc_color})([ACRMDT])\t#$1   $2 #o;
4695                        push @{$c->{changed}}, $_;
4696                } elsif (/^${esc_color}diff /o) {
4697                        $d = 1;
4698                        push @{$c->{diff}}, $_;
4699                } elsif ($d) {
4700                        push @{$c->{diff}}, $_;
4701                } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
4702                          $esc_color*[\+\-]*$esc_color$/x) {
4703                        $stat = 1;
4704                        push @{$c->{stat}}, $_;
4705                } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
4706                        push @{$c->{stat}}, $_;
4707                        $stat = undef;
4708                } elsif (/^${esc_color}    (git-svn-id:.+)$/o) {
4709                        ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
4710                } elsif (s/^${esc_color}    //o) {
4711                        push @{$c->{l}}, $_;
4712                }
4713        }
4714        if ($c && defined $c->{r} && $c->{r} != $r_last) {
4715                $r_last = $c->{r};
4716                process_commit($c, $r_min, $r_max, \@k);
4717        }
4718        if (@k) {
4719                ($r_min, $r_max) = ($r_max, $r_min);
4720                process_commit($_, $r_min, $r_max) foreach reverse @k;
4721        }
4722out:
4723        close $log;
4724        print commit_log_separator unless $incremental || $oneline;
4725}
4726
4727sub cmd_blame {
4728        my $path = pop;
4729
4730        config_pager();
4731        run_pager();
4732
4733        my ($fh, $ctx, $rev);
4734
4735        if ($_git_format) {
4736                ($fh, $ctx) = command_output_pipe('blame', @_, $path);
4737                while (my $line = <$fh>) {
4738                        if ($line =~ /^\^?([[:xdigit:]]+)\s/) {
4739                                # Uncommitted edits show up as a rev ID of
4740                                # all zeros, which we can't look up with
4741                                # cmt_metadata
4742                                if ($1 !~ /^0+$/) {
4743                                        (undef, $rev, undef) =
4744                                                ::cmt_metadata($1);
4745                                        $rev = '0' if (!$rev);
4746                                } else {
4747                                        $rev = '0';
4748                                }
4749                                $rev = sprintf('%-10s', $rev);
4750                                $line =~ s/^\^?[[:xdigit:]]+(\s)/$rev$1/;
4751                        }
4752                        print $line;
4753                }
4754        } else {
4755                ($fh, $ctx) = command_output_pipe('blame', '-p', @_, 'HEAD',
4756                                                  '--', $path);
4757                my ($sha1);
4758                my %authors;
4759                while (my $line = <$fh>) {
4760                        if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
4761                                $sha1 = $1;
4762                                (undef, $rev, undef) = ::cmt_metadata($1);
4763                                $rev = '0' if (!$rev);
4764                        }
4765                        elsif ($line =~ /^author (.*)/) {
4766                                $authors{$rev} = $1;
4767                                $authors{$rev} =~ s/\s/_/g;
4768                        }
4769                        elsif ($line =~ /^\t(.*)$/) {
4770                                printf("%6s %10s %s\n", $rev, $authors{$rev}, $1);
4771                        }
4772                }
4773        }
4774        command_close_pipe($fh, $ctx);
4775}
4776
4777package Git::SVN::Migration;
4778# these version numbers do NOT correspond to actual version numbers
4779# of git nor git-svn.  They are just relative.
4780#
4781# v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
4782#
4783# v1 layout: .git/$id/info/url, refs/remotes/$id
4784#
4785# v2 layout: .git/svn/$id/info/url, refs/remotes/$id
4786#
4787# v3 layout: .git/svn/$id, refs/remotes/$id
4788#            - info/url may remain for backwards compatibility
4789#            - this is what we migrate up to this layout automatically,
4790#            - this will be used by git svn init on single branches
4791# v3.1 layout (auto migrated):
4792#            - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
4793#              for backwards compatibility
4794#
4795# v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
4796#            - this is only created for newly multi-init-ed
4797#              repositories.  Similar in spirit to the
4798#              --use-separate-remotes option in git-clone (now default)
4799#            - we do not automatically migrate to this (following
4800#              the example set by core git)
4801#
4802# v5 layout: .rev_db.$UUID => .rev_map.$UUID
4803#            - newer, more-efficient format that uses 24-bytes per record
4804#              with no filler space.
4805#            - use xxd -c24 < .rev_map.$UUID to view and debug
4806#            - This is a one-way migration, repositories updated to the
4807#              new format will not be able to use old git-svn without
4808#              rebuilding the .rev_db.  Rebuilding the rev_db is not
4809#              possible if noMetadata or useSvmProps are set; but should
4810#              be no problem for users that use the (sensible) defaults.
4811use strict;
4812use warnings;
4813use Carp qw/croak/;
4814use File::Path qw/mkpath/;
4815use File::Basename qw/dirname basename/;
4816use vars qw/$_minimize/;
4817
4818sub migrate_from_v0 {
4819        my $git_dir = $ENV{GIT_DIR};
4820        return undef unless -d $git_dir;
4821        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
4822        my $migrated = 0;
4823        while (<$fh>) {
4824                chomp;
4825                my ($id, $orig_ref) = ($_, $_);
4826                next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
4827                next unless -f "$git_dir/$id/info/url";
4828                my $new_ref = "refs/remotes/$id";
4829                if (::verify_ref("$new_ref^0")) {
4830                        print STDERR "W: $orig_ref is probably an old ",
4831                                     "branch used by an ancient version of ",
4832                                     "git-svn.\n",
4833                                     "However, $new_ref also exists.\n",
4834                                     "We will not be able ",
4835                                     "to use this branch until this ",
4836                                     "ambiguity is resolved.\n";
4837                        next;
4838                }
4839                print STDERR "Migrating from v0 layout...\n" if !$migrated;
4840                print STDERR "Renaming ref: $orig_ref => $new_ref\n";
4841                command_noisy('update-ref', $new_ref, $orig_ref);
4842                command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
4843                $migrated++;
4844        }
4845        command_close_pipe($fh, $ctx);
4846        print STDERR "Done migrating from v0 layout...\n" if $migrated;
4847        $migrated;
4848}
4849
4850sub migrate_from_v1 {
4851        my $git_dir = $ENV{GIT_DIR};
4852        my $migrated = 0;
4853        return $migrated unless -d $git_dir;
4854        my $svn_dir = "$git_dir/svn";
4855
4856        # just in case somebody used 'svn' as their $id at some point...
4857        return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
4858
4859        print STDERR "Migrating from a git-svn v1 layout...\n";
4860        mkpath([$svn_dir]);
4861        print STDERR "Data from a previous version of git-svn exists, but\n\t",
4862                     "$svn_dir\n\t(required for this version ",
4863                     "($::VERSION) of git-svn) does not exist.\n";
4864        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
4865        while (<$fh>) {
4866                my $x = $_;
4867                next unless $x =~ s#^refs/remotes/##;
4868                chomp $x;
4869                next unless -f "$git_dir/$x/info/url";
4870                my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
4871                next unless $u;
4872                my $dn = dirname("$git_dir/svn/$x");
4873                mkpath([$dn]) unless -d $dn;
4874                if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
4875                        mkpath(["$git_dir/svn/svn"]);
4876                        print STDERR " - $git_dir/$x/info => ",
4877                                        "$git_dir/svn/$x/info\n";
4878                        rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
4879                               croak "$!: $x";
4880                        # don't worry too much about these, they probably
4881                        # don't exist with repos this old (save for index,
4882                        # and we can easily regenerate that)
4883                        foreach my $f (qw/unhandled.log index .rev_db/) {
4884                                rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
4885                        }
4886                } else {
4887                        print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
4888                        rename "$git_dir/$x", "$git_dir/svn/$x" or
4889                               croak "$!: $x";
4890                }
4891                $migrated++;
4892        }
4893        command_close_pipe($fh, $ctx);
4894        print STDERR "Done migrating from a git-svn v1 layout\n";
4895        $migrated;
4896}
4897
4898sub read_old_urls {
4899        my ($l_map, $pfx, $path) = @_;
4900        my @dir;
4901        foreach (<$path/*>) {
4902                if (-r "$_/info/url") {
4903                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
4904                        my $ref_id = $pfx . basename $_;
4905                        my $url = ::file_to_s("$_/info/url");
4906                        $l_map->{$ref_id} = $url;
4907                } elsif (-d $_) {
4908                        push @dir, $_;
4909                }
4910        }
4911        foreach (@dir) {
4912                my $x = $_;
4913                $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
4914                read_old_urls($l_map, $x, $_);
4915        }
4916}
4917
4918sub migrate_from_v2 {
4919        my @cfg = command(qw/config -l/);
4920        return if grep /^svn-remote\..+\.url=/, @cfg;
4921        my %l_map;
4922        read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
4923        my $migrated = 0;
4924
4925        foreach my $ref_id (sort keys %l_map) {
4926                eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
4927                if ($@) {
4928                        Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
4929                }
4930                $migrated++;
4931        }
4932        $migrated;
4933}
4934
4935sub minimize_connections {
4936        my $r = Git::SVN::read_all_remotes();
4937        my $new_urls = {};
4938        my $root_repos = {};
4939        foreach my $repo_id (keys %$r) {
4940                my $url = $r->{$repo_id}->{url} or next;
4941                my $fetch = $r->{$repo_id}->{fetch} or next;
4942                my $ra = Git::SVN::Ra->new($url);
4943
4944                # skip existing cases where we already connect to the root
4945                if (($ra->{url} eq $ra->{repos_root}) ||
4946                    ($ra->{repos_root} eq $repo_id)) {
4947                        $root_repos->{$ra->{url}} = $repo_id;
4948                        next;
4949                }
4950
4951                my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
4952                my $root_path = $ra->{url};
4953                $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
4954                foreach my $path (keys %$fetch) {
4955                        my $ref_id = $fetch->{$path};
4956                        my $gs = Git::SVN->new($ref_id, $repo_id, $path);
4957
4958                        # make sure we can read when connecting to
4959                        # a higher level of a repository
4960                        my ($last_rev, undef) = $gs->last_rev_commit;
4961                        if (!defined $last_rev) {
4962                                $last_rev = eval {
4963                                        $root_ra->get_latest_revnum;
4964                                };
4965                                next if $@;
4966                        }
4967                        my $new = $root_path;
4968                        $new .= length $path ? "/$path" : '';
4969                        eval {
4970                                $root_ra->get_log([$new], $last_rev, $last_rev,
4971                                                  0, 0, 1, sub { });
4972                        };
4973                        next if $@;
4974                        $new_urls->{$ra->{repos_root}}->{$new} =
4975                                { ref_id => $ref_id,
4976                                  old_repo_id => $repo_id,
4977                                  old_path => $path };
4978                }
4979        }
4980
4981        my @emptied;
4982        foreach my $url (keys %$new_urls) {
4983                # see if we can re-use an existing [svn-remote "repo_id"]
4984                # instead of creating a(n ugly) new section:
4985                my $repo_id = $root_repos->{$url} || $url;
4986
4987                my $fetch = $new_urls->{$url};
4988                foreach my $path (keys %$fetch) {
4989                        my $x = $fetch->{$path};
4990                        Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
4991                        my $pfx = "svn-remote.$x->{old_repo_id}";
4992
4993                        my $old_fetch = quotemeta("$x->{old_path}:".
4994                                                  "refs/remotes/$x->{ref_id}");
4995                        command_noisy(qw/config --unset/,
4996                                      "$pfx.fetch", '^'. $old_fetch . '$');
4997                        delete $r->{$x->{old_repo_id}}->
4998                               {fetch}->{$x->{old_path}};
4999                        if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
5000                                command_noisy(qw/config --unset/,
5001                                              "$pfx.url");
5002                                push @emptied, $x->{old_repo_id}
5003                        }
5004                }
5005        }
5006        if (@emptied) {
5007                my $file = $ENV{GIT_CONFIG} || $ENV{GIT_CONFIG_LOCAL} ||
5008                           "$ENV{GIT_DIR}/config";
5009                print STDERR <<EOF;
5010The following [svn-remote] sections in your config file ($file) are empty
5011and can be safely removed:
5012EOF
5013                print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
5014        }
5015}
5016
5017sub migration_check {
5018        migrate_from_v0();
5019        migrate_from_v1();
5020        migrate_from_v2();
5021        minimize_connections() if $_minimize;
5022}
5023
5024package Git::IndexInfo;
5025use strict;
5026use warnings;
5027use Git qw/command_input_pipe command_close_pipe/;
5028
5029sub new {
5030        my ($class) = @_;
5031        my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
5032        bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
5033}
5034
5035sub remove {
5036        my ($self, $path) = @_;
5037        if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
5038                return ++$self->{nr};
5039        }
5040        undef;
5041}
5042
5043sub update {
5044        my ($self, $mode, $hash, $path) = @_;
5045        if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
5046                return ++$self->{nr};
5047        }
5048        undef;
5049}
5050
5051sub DESTROY {
5052        my ($self) = @_;
5053        command_close_pipe($self->{gui}, $self->{ctx});
5054}
5055
5056package Git::SVN::GlobSpec;
5057use strict;
5058use warnings;
5059
5060sub new {
5061        my ($class, $glob) = @_;
5062        my $re = $glob;
5063        $re =~ s!/+$!!g; # no need for trailing slashes
5064        $re =~ m!^([^*]*)(\*(?:/\*)*)([^*]*)$!;
5065        my $temp = $re;
5066        my ($left, $right) = ($1, $3);
5067        $re = $2;
5068        my $depth = $re =~ tr/*/*/;
5069        if ($depth != $temp =~ tr/*/*/) {
5070                die "Only one set of wildcard directories " .
5071                        "(e.g. '*' or '*/*/*') is supported: '$glob'\n";
5072        }
5073        if ($depth == 0) {
5074                die "One '*' is needed for glob: '$glob'\n";
5075        }
5076        $re =~ s!\*!\[^/\]*!g;
5077        $re = quotemeta($left) . "($re)" . quotemeta($right);
5078        if (length $left && !($left =~ s!/+$!!g)) {
5079                die "Missing trailing '/' on left side of: '$glob' ($left)\n";
5080        }
5081        if (length $right && !($right =~ s!^/+!!g)) {
5082                die "Missing leading '/' on right side of: '$glob' ($right)\n";
5083        }
5084        my $left_re = qr/^\/\Q$left\E(\/|$)/;
5085        bless { left => $left, right => $right, left_regex => $left_re,
5086                regex => qr/$re/, glob => $glob, depth => $depth }, $class;
5087}
5088
5089sub full_path {
5090        my ($self, $path) = @_;
5091        return (length $self->{left} ? "$self->{left}/" : '') .
5092               $path . (length $self->{right} ? "/$self->{right}" : '');
5093}
5094
5095__END__
5096
5097Data structures:
5098
5099
5100$remotes = { # returned by read_all_remotes()
5101        'svn' => {
5102                # svn-remote.svn.url=https://svn.musicpd.org
5103                url => 'https://svn.musicpd.org',
5104                # svn-remote.svn.fetch=mpd/trunk:trunk
5105                fetch => {
5106                        'mpd/trunk' => 'trunk',
5107                },
5108                # svn-remote.svn.tags=mpd/tags/*:tags/*
5109                tags => {
5110                        path => {
5111                                left => 'mpd/tags',
5112                                right => '',
5113                                regex => qr!mpd/tags/([^/]+)$!,
5114                                glob => 'tags/*',
5115                        },
5116                        ref => {
5117                                left => 'tags',
5118                                right => '',
5119                                regex => qr!tags/([^/]+)$!,
5120                                glob => 'tags/*',
5121                        },
5122                }
5123        }
5124};
5125
5126$log_entry hashref as returned by libsvn_log_entry()
5127{
5128        log => 'whitespace-formatted log entry
5129',                                              # trailing newline is preserved
5130        revision => '8',                        # integer
5131        date => '2004-02-24T17:01:44.108345Z',  # commit date
5132        author => 'committer name'
5133};
5134
5135
5136# this is generated by generate_diff();
5137@mods = array of diff-index line hashes, each element represents one line
5138        of diff-index output
5139
5140diff-index line ($m hash)
5141{
5142        mode_a => first column of diff-index output, no leading ':',
5143        mode_b => second column of diff-index output,
5144        sha1_b => sha1sum of the final blob,
5145        chg => change type [MCRADT],
5146        file_a => original file name of a file (iff chg is 'C' or 'R')
5147        file_b => new/current file name of a file (any chg)
5148}
5149;
5150
5151# retval of read_url_paths{,_all}();
5152$l_map = {
5153        # repository root url
5154        'https://svn.musicpd.org' => {
5155                # repository path               # GIT_SVN_ID
5156                'mpd/trunk'             =>      'trunk',
5157                'mpd/tags/0.11.5'       =>      'tags/0.11.5',
5158        },
5159}
5160
5161Notes:
5162        I don't trust the each() function on unless I created %hash myself
5163        because the internal iterator may not have started at base.