fad29605db14d8d52d1f2bb8e93e0c94427f99d0
   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 $_authors_prog %users/;
   9$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$Git::SVN::_minimize_url = 'unset';
  23
  24$Git::SVN::Log::TZ = $ENV{TZ};
  25$ENV{TZ} = 'UTC';
  26$| = 1; # unbuffer STDOUT
  27
  28sub fatal (@) { print STDERR "@_\n"; exit 1 }
  29require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
  30require SVN::Ra;
  31require SVN::Delta;
  32if ($SVN::Core::VERSION lt '1.1.0') {
  33        fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)";
  34}
  35my $can_compress = eval { require Compress::Zlib; 1};
  36push @Git::SVN::Ra::ISA, 'SVN::Ra';
  37push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
  38push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
  39use Carp qw/croak/;
  40use Digest::MD5;
  41use IO::File qw//;
  42use File::Basename qw/dirname basename/;
  43use File::Path qw/mkpath/;
  44use File::Spec;
  45use File::Find;
  46use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev/;
  47use IPC::Open3;
  48use Git;
  49
  50BEGIN {
  51        # import functions from Git into our packages, en masse
  52        no strict 'refs';
  53        foreach (qw/command command_oneline command_noisy command_output_pipe
  54                    command_input_pipe command_close_pipe
  55                    command_bidi_pipe command_close_bidi_pipe/) {
  56                for my $package ( qw(SVN::Git::Editor SVN::Git::Fetcher
  57                        Git::SVN::Migration Git::SVN::Log Git::SVN),
  58                        __PACKAGE__) {
  59                        *{"${package}::$_"} = \&{"Git::$_"};
  60                }
  61        }
  62}
  63
  64my ($SVN);
  65
  66$sha1 = qr/[a-f\d]{40}/;
  67$sha1_short = qr/[a-f\d]{4,40}/;
  68my ($_stdin, $_help, $_edit,
  69        $_message, $_file, $_branch_dest,
  70        $_template, $_shared,
  71        $_version, $_fetch_all, $_no_rebase, $_fetch_parent,
  72        $_merge, $_strategy, $_dry_run, $_local,
  73        $_prefix, $_no_checkout, $_url, $_verbose,
  74        $_git_format, $_commit_url, $_tag);
  75$Git::SVN::_follow_parent = 1;
  76$_q ||= 0;
  77my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
  78                    'config-dir=s' => \$Git::SVN::Ra::config_dir,
  79                    'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache,
  80                    'ignore-paths=s' => \$SVN::Git::Fetcher::_ignore_regex );
  81my %fc_opts = ( 'follow-parent|follow!' => \$Git::SVN::_follow_parent,
  82                'authors-file|A=s' => \$_authors,
  83                'authors-prog=s' => \$_authors_prog,
  84                'repack:i' => \$Git::SVN::_repack,
  85                'noMetadata' => \$Git::SVN::_no_metadata,
  86                'useSvmProps' => \$Git::SVN::_use_svm_props,
  87                'useSvnsyncProps' => \$Git::SVN::_use_svnsync_props,
  88                'log-window-size=i' => \$Git::SVN::Ra::_log_window_size,
  89                'no-checkout' => \$_no_checkout,
  90                'quiet|q+' => \$_q,
  91                'repack-flags|repack-args|repack-opts=s' =>
  92                   \$Git::SVN::_repack_flags,
  93                'use-log-author' => \$Git::SVN::_use_log_author,
  94                'add-author-from' => \$Git::SVN::_add_author_from,
  95                'localtime' => \$Git::SVN::_localtime,
  96                %remote_opts );
  97
  98my ($_trunk, @_tags, @_branches, $_stdlayout);
  99my %icv;
 100my %init_opts = ( 'template=s' => \$_template, 'shared:s' => \$_shared,
 101                  'trunk|T=s' => \$_trunk, 'tags|t=s@' => \@_tags,
 102                  'branches|b=s@' => \@_branches, 'prefix=s' => \$_prefix,
 103                  'stdlayout|s' => \$_stdlayout,
 104                  'minimize-url|m!' => \$Git::SVN::_minimize_url,
 105                  'no-metadata' => sub { $icv{noMetadata} = 1 },
 106                  'use-svm-props' => sub { $icv{useSvmProps} = 1 },
 107                  'use-svnsync-props' => sub { $icv{useSvnsyncProps} = 1 },
 108                  'rewrite-root=s' => sub { $icv{rewriteRoot} = $_[1] },
 109                  %remote_opts );
 110my %cmt_opts = ( 'edit|e' => \$_edit,
 111                'rmdir' => \$SVN::Git::Editor::_rmdir,
 112                'find-copies-harder' => \$SVN::Git::Editor::_find_copies_harder,
 113                'l=i' => \$SVN::Git::Editor::_rename_limit,
 114                'copy-similarity|C=i'=> \$SVN::Git::Editor::_cp_similarity
 115);
 116
 117my %cmd = (
 118        fetch => [ \&cmd_fetch, "Download new revisions from SVN",
 119                        { 'revision|r=s' => \$_revision,
 120                          'fetch-all|all' => \$_fetch_all,
 121                          'parent|p' => \$_fetch_parent,
 122                           %fc_opts } ],
 123        clone => [ \&cmd_clone, "Initialize and fetch revisions",
 124                        { 'revision|r=s' => \$_revision,
 125                           %fc_opts, %init_opts } ],
 126        init => [ \&cmd_init, "Initialize a repo for tracking" .
 127                          " (requires URL argument)",
 128                          \%init_opts ],
 129        'multi-init' => [ \&cmd_multi_init,
 130                          "Deprecated alias for ".
 131                          "'$0 init -T<trunk> -b<branches> -t<tags>'",
 132                          \%init_opts ],
 133        dcommit => [ \&cmd_dcommit,
 134                     'Commit several diffs to merge with upstream',
 135                        { 'merge|m|M' => \$_merge,
 136                          'strategy|s=s' => \$_strategy,
 137                          'verbose|v' => \$_verbose,
 138                          'dry-run|n' => \$_dry_run,
 139                          'fetch-all|all' => \$_fetch_all,
 140                          'commit-url=s' => \$_commit_url,
 141                          'revision|r=i' => \$_revision,
 142                          'no-rebase' => \$_no_rebase,
 143                        %cmt_opts, %fc_opts } ],
 144        branch => [ \&cmd_branch,
 145                    'Create a branch in the SVN repository',
 146                    { 'message|m=s' => \$_message,
 147                      'destination|d=s' => \$_branch_dest,
 148                      'dry-run|n' => \$_dry_run,
 149                      'tag|t' => \$_tag } ],
 150        tag => [ sub { $_tag = 1; cmd_branch(@_) },
 151                 'Create a tag in the SVN repository',
 152                 { 'message|m=s' => \$_message,
 153                   'destination|d=s' => \$_branch_dest,
 154                   'dry-run|n' => \$_dry_run } ],
 155        'set-tree' => [ \&cmd_set_tree,
 156                        "Set an SVN repository to a git tree-ish",
 157                        { 'stdin' => \$_stdin, %cmt_opts, %fc_opts, } ],
 158        'create-ignore' => [ \&cmd_create_ignore,
 159                             'Create a .gitignore per svn:ignore',
 160                             { 'revision|r=i' => \$_revision
 161                             } ],
 162        'propget' => [ \&cmd_propget,
 163                       'Print the value of a property on a file or directory',
 164                       { 'revision|r=i' => \$_revision } ],
 165        'proplist' => [ \&cmd_proplist,
 166                       'List all properties of a file or directory',
 167                       { 'revision|r=i' => \$_revision } ],
 168        'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
 169                        { 'revision|r=i' => \$_revision
 170                        } ],
 171        'show-externals' => [ \&cmd_show_externals, "Show svn:externals listings",
 172                        { 'revision|r=i' => \$_revision
 173                        } ],
 174        'multi-fetch' => [ \&cmd_multi_fetch,
 175                           "Deprecated alias for $0 fetch --all",
 176                           { 'revision|r=s' => \$_revision, %fc_opts } ],
 177        'migrate' => [ sub { },
 178                       # no-op, we automatically run this anyways,
 179                       'Migrate configuration/metadata/layout from
 180                        previous versions of git-svn',
 181                       { 'minimize' => \$Git::SVN::Migration::_minimize,
 182                         %remote_opts } ],
 183        'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
 184                        { 'limit=i' => \$Git::SVN::Log::limit,
 185                          'revision|r=s' => \$_revision,
 186                          'verbose|v' => \$Git::SVN::Log::verbose,
 187                          'incremental' => \$Git::SVN::Log::incremental,
 188                          'oneline' => \$Git::SVN::Log::oneline,
 189                          'show-commit' => \$Git::SVN::Log::show_commit,
 190                          'non-recursive' => \$Git::SVN::Log::non_recursive,
 191                          'authors-file|A=s' => \$_authors,
 192                          'color' => \$Git::SVN::Log::color,
 193                          'pager=s' => \$Git::SVN::Log::pager
 194                        } ],
 195        'find-rev' => [ \&cmd_find_rev,
 196                        "Translate between SVN revision numbers and tree-ish",
 197                        {} ],
 198        'rebase' => [ \&cmd_rebase, "Fetch and rebase your working directory",
 199                        { 'merge|m|M' => \$_merge,
 200                          'verbose|v' => \$_verbose,
 201                          'strategy|s=s' => \$_strategy,
 202                          'local|l' => \$_local,
 203                          'fetch-all|all' => \$_fetch_all,
 204                          'dry-run|n' => \$_dry_run,
 205                          %fc_opts } ],
 206        'commit-diff' => [ \&cmd_commit_diff,
 207                           'Commit a diff between two trees',
 208                        { 'message|m=s' => \$_message,
 209                          'file|F=s' => \$_file,
 210                          'revision|r=s' => \$_revision,
 211                        %cmt_opts } ],
 212        'info' => [ \&cmd_info,
 213                    "Show info about the latest SVN revision
 214                     on the current branch",
 215                    { 'url' => \$_url, } ],
 216        'blame' => [ \&Git::SVN::Log::cmd_blame,
 217                    "Show what revision and author last modified each line of a file",
 218                    { 'git-format' => \$_git_format } ],
 219        'reset' => [ \&cmd_reset,
 220                     "Undo fetches back to the specified SVN revision",
 221                     { 'revision|r=s' => \$_revision,
 222                       'parent|p' => \$_fetch_parent } ],
 223        'gc' => [ \&cmd_gc,
 224                  "Compress unhandled.log files in .git/svn and remove " .
 225                  "index files in .git/svn",
 226                {} ],
 227);
 228
 229my $cmd;
 230for (my $i = 0; $i < @ARGV; $i++) {
 231        if (defined $cmd{$ARGV[$i]}) {
 232                $cmd = $ARGV[$i];
 233                splice @ARGV, $i, 1;
 234                last;
 235        } elsif ($ARGV[$i] eq 'help') {
 236                $cmd = $ARGV[$i+1];
 237                usage(0);
 238        }
 239};
 240
 241# make sure we're always running at the top-level working directory
 242unless ($cmd && $cmd =~ /(?:clone|init|multi-init)$/) {
 243        unless (-d $ENV{GIT_DIR}) {
 244                if ($git_dir_user_set) {
 245                        die "GIT_DIR=$ENV{GIT_DIR} explicitly set, ",
 246                            "but it is not a directory\n";
 247                }
 248                my $git_dir = delete $ENV{GIT_DIR};
 249                my $cdup = undef;
 250                git_cmd_try {
 251                        $cdup = command_oneline(qw/rev-parse --show-cdup/);
 252                        $git_dir = '.' unless ($cdup);
 253                        chomp $cdup if ($cdup);
 254                        $cdup = "." unless ($cdup && length $cdup);
 255                } "Already at toplevel, but $git_dir not found\n";
 256                chdir $cdup or die "Unable to chdir up to '$cdup'\n";
 257                unless (-d $git_dir) {
 258                        die "$git_dir still not found after going to ",
 259                            "'$cdup'\n";
 260                }
 261                $ENV{GIT_DIR} = $git_dir;
 262        }
 263        $_repository = Git->repository(Repository => $ENV{GIT_DIR});
 264}
 265
 266my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
 267
 268read_repo_config(\%opts);
 269if ($cmd && ($cmd eq 'log' || $cmd eq 'blame')) {
 270        Getopt::Long::Configure('pass_through');
 271}
 272my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
 273                    'minimize-connections' => \$Git::SVN::Migration::_minimize,
 274                    'id|i=s' => \$Git::SVN::default_ref_id,
 275                    'svn-remote|remote|R=s' => sub {
 276                       $Git::SVN::no_reuse_existing = 1;
 277                       $Git::SVN::default_repo_id = $_[1] });
 278exit 1 if (!$rv && $cmd && $cmd ne 'log');
 279
 280usage(0) if $_help;
 281version() if $_version;
 282usage(1) unless defined $cmd;
 283load_authors() if $_authors;
 284if (defined $_authors_prog) {
 285        $_authors_prog = "'" . File::Spec->rel2abs($_authors_prog) . "'";
 286}
 287
 288unless ($cmd =~ /^(?:clone|init|multi-init|commit-diff)$/) {
 289        Git::SVN::Migration::migration_check();
 290}
 291Git::SVN::init_vars();
 292eval {
 293        Git::SVN::verify_remotes_sanity();
 294        $cmd{$cmd}->[0]->(@ARGV);
 295};
 296fatal $@ if $@;
 297post_fetch_checkout();
 298exit 0;
 299
 300####################### primary functions ######################
 301sub usage {
 302        my $exit = shift || 0;
 303        my $fd = $exit ? \*STDERR : \*STDOUT;
 304        print $fd <<"";
 305git-svn - bidirectional operations between a single Subversion tree and git
 306Usage: git svn <command> [options] [arguments]\n
 307
 308        print $fd "Available commands:\n" unless $cmd;
 309
 310        foreach (sort keys %cmd) {
 311                next if $cmd && $cmd ne $_;
 312                next if /^multi-/; # don't show deprecated commands
 313                print $fd '  ',pack('A17',$_),$cmd{$_}->[1],"\n";
 314                foreach (sort keys %{$cmd{$_}->[2]}) {
 315                        # mixed-case options are for .git/config only
 316                        next if /[A-Z]/ && /^[a-z]+$/i;
 317                        # prints out arguments as they should be passed:
 318                        my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
 319                        print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
 320                                                        "--$_" : "-$_" }
 321                                                split /\|/,$_)," $x\n";
 322                }
 323        }
 324        print $fd <<"";
 325\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
 326arbitrary identifier if you're tracking multiple SVN branches/repositories in
 327one git repository and want to keep them separate.  See git-svn(1) for more
 328information.
 329
 330        exit $exit;
 331}
 332
 333sub version {
 334        print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
 335        exit 0;
 336}
 337
 338sub do_git_init_db {
 339        unless (-d $ENV{GIT_DIR}) {
 340                my @init_db = ('init');
 341                push @init_db, "--template=$_template" if defined $_template;
 342                if (defined $_shared) {
 343                        if ($_shared =~ /[a-z]/) {
 344                                push @init_db, "--shared=$_shared";
 345                        } else {
 346                                push @init_db, "--shared";
 347                        }
 348                }
 349                command_noisy(@init_db);
 350                $_repository = Git->repository(Repository => ".git");
 351        }
 352        command_noisy('config', 'core.autocrlf', 'false');
 353        my $set;
 354        my $pfx = "svn-remote.$Git::SVN::default_repo_id";
 355        foreach my $i (keys %icv) {
 356                die "'$set' and '$i' cannot both be set\n" if $set;
 357                next unless defined $icv{$i};
 358                command_noisy('config', "$pfx.$i", $icv{$i});
 359                $set = $i;
 360        }
 361        my $ignore_regex = \$SVN::Git::Fetcher::_ignore_regex;
 362        command_noisy('config', "$pfx.ignore-paths", $$ignore_regex)
 363                if defined $$ignore_regex;
 364}
 365
 366sub init_subdir {
 367        my $repo_path = shift or return;
 368        mkpath([$repo_path]) unless -d $repo_path;
 369        chdir $repo_path or die "Couldn't chdir to $repo_path: $!\n";
 370        $ENV{GIT_DIR} = '.git';
 371        $_repository = Git->repository(Repository => $ENV{GIT_DIR});
 372}
 373
 374sub cmd_clone {
 375        my ($url, $path) = @_;
 376        if (!defined $path &&
 377            (defined $_trunk || @_branches || @_tags ||
 378             defined $_stdlayout) &&
 379            $url !~ m#^[a-z\+]+://#) {
 380                $path = $url;
 381        }
 382        $path = basename($url) if !defined $path || !length $path;
 383        cmd_init($url, $path);
 384        Git::SVN::fetch_all($Git::SVN::default_repo_id);
 385        command_oneline('config', 'svn.authorsfile', $_authors) if $_authors;
 386}
 387
 388sub cmd_init {
 389        if (defined $_stdlayout) {
 390                $_trunk = 'trunk' if (!defined $_trunk);
 391                @_tags = 'tags' if (! @_tags);
 392                @_branches = 'branches' if (! @_branches);
 393        }
 394        if (defined $_trunk || @_branches || @_tags) {
 395                return cmd_multi_init(@_);
 396        }
 397        my $url = shift or die "SVN repository location required ",
 398                               "as a command-line argument\n";
 399        $url = canonicalize_url($url);
 400        init_subdir(@_);
 401        do_git_init_db();
 402
 403        if ($Git::SVN::_minimize_url eq 'unset') {
 404                $Git::SVN::_minimize_url = 0;
 405        }
 406
 407        Git::SVN->init($url);
 408}
 409
 410sub cmd_fetch {
 411        if (grep /^\d+=./, @_) {
 412                die "'<rev>=<commit>' fetch arguments are ",
 413                    "no longer supported.\n";
 414        }
 415        my ($remote) = @_;
 416        if (@_ > 1) {
 417                die "Usage: $0 fetch [--all] [--parent] [svn-remote]\n";
 418        }
 419        if ($_fetch_parent) {
 420                my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 421                unless ($gs) {
 422                        die "Unable to determine upstream SVN information from ",
 423                            "working tree history\n";
 424                }
 425                # just fetch, don't checkout.
 426                $_no_checkout = 'true';
 427                $_fetch_all ? $gs->fetch_all : $gs->fetch;
 428        } elsif ($_fetch_all) {
 429                cmd_multi_fetch();
 430        } else {
 431                $remote ||= $Git::SVN::default_repo_id;
 432                Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
 433        }
 434}
 435
 436sub cmd_set_tree {
 437        my (@commits) = @_;
 438        if ($_stdin || !@commits) {
 439                print "Reading from stdin...\n";
 440                @commits = ();
 441                while (<STDIN>) {
 442                        if (/\b($sha1_short)\b/o) {
 443                                unshift @commits, $1;
 444                        }
 445                }
 446        }
 447        my @revs;
 448        foreach my $c (@commits) {
 449                my @tmp = command('rev-parse',$c);
 450                if (scalar @tmp == 1) {
 451                        push @revs, $tmp[0];
 452                } elsif (scalar @tmp > 1) {
 453                        push @revs, reverse(command('rev-list',@tmp));
 454                } else {
 455                        fatal "Failed to rev-parse $c";
 456                }
 457        }
 458        my $gs = Git::SVN->new;
 459        my ($r_last, $cmt_last) = $gs->last_rev_commit;
 460        $gs->fetch;
 461        if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
 462                fatal "There are new revisions that were fetched ",
 463                      "and need to be merged (or acknowledged) ",
 464                      "before committing.\nlast rev: $r_last\n",
 465                      " current: $gs->{last_rev}";
 466        }
 467        $gs->set_tree($_) foreach @revs;
 468        print "Done committing ",scalar @revs," revisions to SVN\n";
 469        unlink $gs->{index};
 470}
 471
 472sub cmd_dcommit {
 473        my $head = shift;
 474        git_cmd_try { command_oneline(qw/diff-index --quiet HEAD/) }
 475                'Cannot dcommit with a dirty index.  Commit your changes first, '
 476                . "or stash them with `git stash'.\n";
 477        $head ||= 'HEAD';
 478
 479        my $old_head;
 480        if ($head ne 'HEAD') {
 481                $old_head = eval {
 482                        command_oneline([qw/symbolic-ref -q HEAD/])
 483                };
 484                if ($old_head) {
 485                        $old_head =~ s{^refs/heads/}{};
 486                } else {
 487                        $old_head = eval { command_oneline(qw/rev-parse HEAD/) };
 488                }
 489                command(['checkout', $head], STDERR => 0);
 490        }
 491
 492        my @refs;
 493        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD', \@refs);
 494        unless ($gs) {
 495                die "Unable to determine upstream SVN information from ",
 496                    "$head history.\nPerhaps the repository is empty.";
 497        }
 498
 499        if (defined $_commit_url) {
 500                $url = $_commit_url;
 501        } else {
 502                $url = eval { command_oneline('config', '--get',
 503                              "svn-remote.$gs->{repo_id}.commiturl") };
 504                if (!$url) {
 505                        $url = $gs->full_url
 506                }
 507        }
 508
 509        my $last_rev = $_revision if defined $_revision;
 510        if ($url) {
 511                print "Committing to $url ...\n";
 512        }
 513        my ($linear_refs, $parents) = linearize_history($gs, \@refs);
 514        if ($_no_rebase && scalar(@$linear_refs) > 1) {
 515                warn "Attempting to commit more than one change while ",
 516                     "--no-rebase is enabled.\n",
 517                     "If these changes depend on each other, re-running ",
 518                     "without --no-rebase may be required."
 519        }
 520        my $expect_url = $url;
 521        Git::SVN::remove_username($expect_url);
 522        while (1) {
 523                my $d = shift @$linear_refs or last;
 524                unless (defined $last_rev) {
 525                        (undef, $last_rev, undef) = cmt_metadata("$d~1");
 526                        unless (defined $last_rev) {
 527                                fatal "Unable to extract revision information ",
 528                                      "from commit $d~1";
 529                        }
 530                }
 531                if ($_dry_run) {
 532                        print "diff-tree $d~1 $d\n";
 533                } else {
 534                        my $cmt_rev;
 535                        my %ed_opts = ( r => $last_rev,
 536                                        log => get_commit_entry($d)->{log},
 537                                        ra => Git::SVN::Ra->new($url),
 538                                        config => SVN::Core::config_get_config(
 539                                                $Git::SVN::Ra::config_dir
 540                                        ),
 541                                        tree_a => "$d~1",
 542                                        tree_b => $d,
 543                                        editor_cb => sub {
 544                                               print "Committed r$_[0]\n";
 545                                               $cmt_rev = $_[0];
 546                                        },
 547                                        svn_path => '');
 548                        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
 549                                print "No changes\n$d~1 == $d\n";
 550                        } elsif ($parents->{$d} && @{$parents->{$d}}) {
 551                                $gs->{inject_parents_dcommit}->{$cmt_rev} =
 552                                                               $parents->{$d};
 553                        }
 554                        $_fetch_all ? $gs->fetch_all : $gs->fetch;
 555                        $last_rev = $cmt_rev;
 556                        next if $_no_rebase;
 557
 558                        # we always want to rebase against the current HEAD,
 559                        # not any head that was passed to us
 560                        my @diff = command('diff-tree', $d,
 561                                           $gs->refname, '--');
 562                        my @finish;
 563                        if (@diff) {
 564                                @finish = rebase_cmd();
 565                                print STDERR "W: $d and ", $gs->refname,
 566                                             " differ, using @finish:\n",
 567                                             join("\n", @diff), "\n";
 568                        } else {
 569                                print "No changes between current HEAD and ",
 570                                      $gs->refname,
 571                                      "\nResetting to the latest ",
 572                                      $gs->refname, "\n";
 573                                @finish = qw/reset --mixed/;
 574                        }
 575                        command_noisy(@finish, $gs->refname);
 576                        if (@diff) {
 577                                @refs = ();
 578                                my ($url_, $rev_, $uuid_, $gs_) =
 579                                              working_head_info('HEAD', \@refs);
 580                                my ($linear_refs_, $parents_) =
 581                                              linearize_history($gs_, \@refs);
 582                                if (scalar(@$linear_refs) !=
 583                                    scalar(@$linear_refs_)) {
 584                                        fatal "# of revisions changed ",
 585                                          "\nbefore:\n",
 586                                          join("\n", @$linear_refs),
 587                                          "\n\nafter:\n",
 588                                          join("\n", @$linear_refs_), "\n",
 589                                          'If you are attempting to commit ',
 590                                          "merges, try running:\n\t",
 591                                          'git rebase --interactive',
 592                                          '--preserve-merges ',
 593                                          $gs->refname,
 594                                          "\nBefore dcommitting";
 595                                }
 596                                if ($url_ ne $expect_url) {
 597                                        fatal "URL mismatch after rebase: ",
 598                                              "$url_ != $expect_url";
 599                                }
 600                                if ($uuid_ ne $uuid) {
 601                                        fatal "uuid mismatch after rebase: ",
 602                                              "$uuid_ != $uuid";
 603                                }
 604                                # remap parents
 605                                my (%p, @l, $i);
 606                                for ($i = 0; $i < scalar @$linear_refs; $i++) {
 607                                        my $new = $linear_refs_->[$i] or next;
 608                                        $p{$new} =
 609                                                $parents->{$linear_refs->[$i]};
 610                                        push @l, $new;
 611                                }
 612                                $parents = \%p;
 613                                $linear_refs = \@l;
 614                        }
 615                }
 616        }
 617
 618        if ($old_head) {
 619                my $new_head = command_oneline(qw/rev-parse HEAD/);
 620                my $new_is_symbolic = eval {
 621                        command_oneline(qw/symbolic-ref -q HEAD/);
 622                };
 623                if ($new_is_symbolic) {
 624                        print "dcommitted the branch ", $head, "\n";
 625                } else {
 626                        print "dcommitted on a detached HEAD because you gave ",
 627                              "a revision argument.\n",
 628                              "The rewritten commit is: ", $new_head, "\n";
 629                }
 630                command(['checkout', $old_head], STDERR => 0);
 631        }
 632
 633        unlink $gs->{index};
 634}
 635
 636sub cmd_branch {
 637        my ($branch_name, $head) = @_;
 638
 639        unless (defined $branch_name && length $branch_name) {
 640                die(($_tag ? "tag" : "branch") . " name required\n");
 641        }
 642        $head ||= 'HEAD';
 643
 644        my ($src, $rev, undef, $gs) = working_head_info($head);
 645
 646        my $remote = Git::SVN::read_all_remotes()->{$gs->{repo_id}};
 647        my $allglobs = $remote->{ $_tag ? 'tags' : 'branches' };
 648        my $glob;
 649        if ($#{$allglobs} == 0) {
 650                $glob = $allglobs->[0];
 651        } else {
 652                unless(defined $_branch_dest) {
 653                        die "Multiple ",
 654                            $_tag ? "tag" : "branch",
 655                            " paths defined for Subversion repository.\n",
 656                            "You must specify where you want to create the ",
 657                            $_tag ? "tag" : "branch",
 658                            " with the --destination argument.\n";
 659                }
 660                foreach my $g (@{$allglobs}) {
 661                        # SVN::Git::Editor could probably be moved to Git.pm..
 662                        my $re = SVN::Git::Editor::glob2pat($g->{path}->{left});
 663                        if ($_branch_dest =~ /$re/) {
 664                                $glob = $g;
 665                                last;
 666                        }
 667                }
 668                unless (defined $glob) {
 669                        my $dest_re = qr/\b\Q$_branch_dest\E\b/;
 670                        foreach my $g (@{$allglobs}) {
 671                                $g->{path}->{left} =~ /$dest_re/ or next;
 672                                if (defined $glob) {
 673                                        die "Ambiguous destination: ",
 674                                            $_branch_dest, "\nmatches both '",
 675                                            $glob->{path}->{left}, "' and '",
 676                                            $g->{path}->{left}, "'\n";
 677                                }
 678                                $glob = $g;
 679                        }
 680                        unless (defined $glob) {
 681                                die "Unknown ",
 682                                    $_tag ? "tag" : "branch",
 683                                    " destination $_branch_dest\n";
 684                        }
 685                }
 686        }
 687        my ($lft, $rgt) = @{ $glob->{path} }{qw/left right/};
 688        my $dst = join '/', $remote->{url}, $lft, $branch_name, ($rgt || ());
 689
 690        my $ctx = SVN::Client->new(
 691                auth    => Git::SVN::Ra::_auth_providers(),
 692                log_msg => sub {
 693                        ${ $_[0] } = defined $_message
 694                                ? $_message
 695                                : 'Create ' . ($_tag ? 'tag ' : 'branch ' )
 696                                . $branch_name;
 697                },
 698        );
 699
 700        eval {
 701                $ctx->ls($dst, 'HEAD', 0);
 702        } and die "branch ${branch_name} already exists\n";
 703
 704        print "Copying ${src} at r${rev} to ${dst}...\n";
 705        $ctx->copy($src, $rev, $dst)
 706                unless $_dry_run;
 707
 708        $gs->fetch_all;
 709}
 710
 711sub cmd_find_rev {
 712        my $revision_or_hash = shift or die "SVN or git revision required ",
 713                                            "as a command-line argument\n";
 714        my $result;
 715        if ($revision_or_hash =~ /^r\d+$/) {
 716                my $head = shift;
 717                $head ||= 'HEAD';
 718                my @refs;
 719                my (undef, undef, $uuid, $gs) = working_head_info($head, \@refs);
 720                unless ($gs) {
 721                        die "Unable to determine upstream SVN information from ",
 722                            "$head history\n";
 723                }
 724                my $desired_revision = substr($revision_or_hash, 1);
 725                $result = $gs->rev_map_get($desired_revision, $uuid);
 726        } else {
 727                my (undef, $rev, undef) = cmt_metadata($revision_or_hash);
 728                $result = $rev;
 729        }
 730        print "$result\n" if $result;
 731}
 732
 733sub cmd_rebase {
 734        command_noisy(qw/update-index --refresh/);
 735        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 736        unless ($gs) {
 737                die "Unable to determine upstream SVN information from ",
 738                    "working tree history\n";
 739        }
 740        if ($_dry_run) {
 741                print "Remote Branch: " . $gs->refname . "\n";
 742                print "SVN URL: " . $url . "\n";
 743                return;
 744        }
 745        if (command(qw/diff-index HEAD --/)) {
 746                print STDERR "Cannot rebase with uncommited changes:\n";
 747                command_noisy('status');
 748                exit 1;
 749        }
 750        unless ($_local) {
 751                # rebase will checkout for us, so no need to do it explicitly
 752                $_no_checkout = 'true';
 753                $_fetch_all ? $gs->fetch_all : $gs->fetch;
 754        }
 755        command_noisy(rebase_cmd(), $gs->refname);
 756}
 757
 758sub cmd_show_ignore {
 759        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 760        $gs ||= Git::SVN->new;
 761        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 762        $gs->prop_walk($gs->{path}, $r, sub {
 763                my ($gs, $path, $props) = @_;
 764                print STDOUT "\n# $path\n";
 765                my $s = $props->{'svn:ignore'} or return;
 766                $s =~ s/[\r\n]+/\n/g;
 767                $s =~ s/^\n+//;
 768                chomp $s;
 769                $s =~ s#^#$path#gm;
 770                print STDOUT "$s\n";
 771        });
 772}
 773
 774sub cmd_show_externals {
 775        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 776        $gs ||= Git::SVN->new;
 777        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 778        $gs->prop_walk($gs->{path}, $r, sub {
 779                my ($gs, $path, $props) = @_;
 780                print STDOUT "\n# $path\n";
 781                my $s = $props->{'svn:externals'} or return;
 782                $s =~ s/[\r\n]+/\n/g;
 783                chomp $s;
 784                $s =~ s#^#$path#gm;
 785                print STDOUT "$s\n";
 786        });
 787}
 788
 789sub cmd_create_ignore {
 790        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 791        $gs ||= Git::SVN->new;
 792        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 793        $gs->prop_walk($gs->{path}, $r, sub {
 794                my ($gs, $path, $props) = @_;
 795                # $path is of the form /path/to/dir/
 796                $path = '.' . $path;
 797                # SVN can have attributes on empty directories,
 798                # which git won't track
 799                mkpath([$path]) unless -d $path;
 800                my $ignore = $path . '.gitignore';
 801                my $s = $props->{'svn:ignore'} or return;
 802                open(GITIGNORE, '>', $ignore)
 803                  or fatal("Failed to open `$ignore' for writing: $!");
 804                $s =~ s/[\r\n]+/\n/g;
 805                $s =~ s/^\n+//;
 806                chomp $s;
 807                # Prefix all patterns so that the ignore doesn't apply
 808                # to sub-directories.
 809                $s =~ s#^#/#gm;
 810                print GITIGNORE "$s\n";
 811                close(GITIGNORE)
 812                  or fatal("Failed to close `$ignore': $!");
 813                command_noisy('add', '-f', $ignore);
 814        });
 815}
 816
 817sub canonicalize_path {
 818        my ($path) = @_;
 819        my $dot_slash_added = 0;
 820        if (substr($path, 0, 1) ne "/") {
 821                $path = "./" . $path;
 822                $dot_slash_added = 1;
 823        }
 824        # File::Spec->canonpath doesn't collapse x/../y into y (for a
 825        # good reason), so let's do this manually.
 826        $path =~ s#/+#/#g;
 827        $path =~ s#/\.(?:/|$)#/#g;
 828        $path =~ s#/[^/]+/\.\.##g;
 829        $path =~ s#/$##g;
 830        $path =~ s#^\./## if $dot_slash_added;
 831        $path =~ s#^/##;
 832        $path =~ s#^\.$##;
 833        return $path;
 834}
 835
 836sub canonicalize_url {
 837        my ($url) = @_;
 838        $url =~ s#^([^:]+://[^/]*/)(.*)$#$1 . canonicalize_path($2)#e;
 839        return $url;
 840}
 841
 842# get_svnprops(PATH)
 843# ------------------
 844# Helper for cmd_propget and cmd_proplist below.
 845sub get_svnprops {
 846        my $path = shift;
 847        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 848        $gs ||= Git::SVN->new;
 849
 850        # prefix THE PATH by the sub-directory from which the user
 851        # invoked us.
 852        $path = $cmd_dir_prefix . $path;
 853        fatal("No such file or directory: $path") unless -e $path;
 854        my $is_dir = -d $path ? 1 : 0;
 855        $path = $gs->{path} . '/' . $path;
 856
 857        # canonicalize the path (otherwise libsvn will abort or fail to
 858        # find the file)
 859        $path = canonicalize_path($path);
 860
 861        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 862        my $props;
 863        if ($is_dir) {
 864                (undef, undef, $props) = $gs->ra->get_dir($path, $r);
 865        }
 866        else {
 867                (undef, $props) = $gs->ra->get_file($path, $r, undef);
 868        }
 869        return $props;
 870}
 871
 872# cmd_propget (PROP, PATH)
 873# ------------------------
 874# Print the SVN property PROP for PATH.
 875sub cmd_propget {
 876        my ($prop, $path) = @_;
 877        $path = '.' if not defined $path;
 878        usage(1) if not defined $prop;
 879        my $props = get_svnprops($path);
 880        if (not defined $props->{$prop}) {
 881                fatal("`$path' does not have a `$prop' SVN property.");
 882        }
 883        print $props->{$prop} . "\n";
 884}
 885
 886# cmd_proplist (PATH)
 887# -------------------
 888# Print the list of SVN properties for PATH.
 889sub cmd_proplist {
 890        my $path = shift;
 891        $path = '.' if not defined $path;
 892        my $props = get_svnprops($path);
 893        print "Properties on '$path':\n";
 894        foreach (sort keys %{$props}) {
 895                print "  $_\n";
 896        }
 897}
 898
 899sub cmd_multi_init {
 900        my $url = shift;
 901        unless (defined $_trunk || @_branches || @_tags) {
 902                usage(1);
 903        }
 904
 905        $_prefix = '' unless defined $_prefix;
 906        if (defined $url) {
 907                $url = canonicalize_url($url);
 908                init_subdir(@_);
 909        }
 910        do_git_init_db();
 911        if (defined $_trunk) {
 912                my $trunk_ref = $_prefix . 'trunk';
 913                # try both old-style and new-style lookups:
 914                my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
 915                unless ($gs_trunk) {
 916                        my ($trunk_url, $trunk_path) =
 917                                              complete_svn_url($url, $_trunk);
 918                        $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
 919                                                   undef, $trunk_ref);
 920                }
 921        }
 922        return unless @_branches || @_tags;
 923        my $ra = $url ? Git::SVN::Ra->new($url) : undef;
 924        foreach my $path (@_branches) {
 925                complete_url_ls_init($ra, $path, '--branches/-b', $_prefix);
 926        }
 927        foreach my $path (@_tags) {
 928                complete_url_ls_init($ra, $path, '--tags/-t', $_prefix.'tags/');
 929        }
 930}
 931
 932sub cmd_multi_fetch {
 933        my $remotes = Git::SVN::read_all_remotes();
 934        foreach my $repo_id (sort keys %$remotes) {
 935                if ($remotes->{$repo_id}->{url}) {
 936                        Git::SVN::fetch_all($repo_id, $remotes);
 937                }
 938        }
 939}
 940
 941# this command is special because it requires no metadata
 942sub cmd_commit_diff {
 943        my ($ta, $tb, $url) = @_;
 944        my $usage = "Usage: $0 commit-diff -r<revision> ".
 945                    "<tree-ish> <tree-ish> [<URL>]";
 946        fatal($usage) if (!defined $ta || !defined $tb);
 947        my $svn_path = '';
 948        if (!defined $url) {
 949                my $gs = eval { Git::SVN->new };
 950                if (!$gs) {
 951                        fatal("Needed URL or usable git-svn --id in ",
 952                              "the command-line\n", $usage);
 953                }
 954                $url = $gs->{url};
 955                $svn_path = $gs->{path};
 956        }
 957        unless (defined $_revision) {
 958                fatal("-r|--revision is a required argument\n", $usage);
 959        }
 960        if (defined $_message && defined $_file) {
 961                fatal("Both --message/-m and --file/-F specified ",
 962                      "for the commit message.\n",
 963                      "I have no idea what you mean");
 964        }
 965        if (defined $_file) {
 966                $_message = file_to_s($_file);
 967        } else {
 968                $_message ||= get_commit_entry($tb)->{log};
 969        }
 970        my $ra ||= Git::SVN::Ra->new($url);
 971        my $r = $_revision;
 972        if ($r eq 'HEAD') {
 973                $r = $ra->get_latest_revnum;
 974        } elsif ($r !~ /^\d+$/) {
 975                die "revision argument: $r not understood by git-svn\n";
 976        }
 977        my %ed_opts = ( r => $r,
 978                        log => $_message,
 979                        ra => $ra,
 980                        tree_a => $ta,
 981                        tree_b => $tb,
 982                        editor_cb => sub { print "Committed r$_[0]\n" },
 983                        svn_path => $svn_path );
 984        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
 985                print "No changes\n$ta == $tb\n";
 986        }
 987}
 988
 989sub escape_uri_only {
 990        my ($uri) = @_;
 991        my @tmp;
 992        foreach (split m{/}, $uri) {
 993                s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
 994                push @tmp, $_;
 995        }
 996        join('/', @tmp);
 997}
 998
 999sub escape_url {
1000        my ($url) = @_;
1001        if ($url =~ m#^([^:]+)://([^/]*)(.*)$#) {
1002                my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
1003                $url = "$scheme://$domain$uri";
1004        }
1005        $url;
1006}
1007
1008sub cmd_info {
1009        my $path = canonicalize_path(defined($_[0]) ? $_[0] : ".");
1010        my $fullpath = canonicalize_path($cmd_dir_prefix . $path);
1011        if (exists $_[1]) {
1012                die "Too many arguments specified\n";
1013        }
1014
1015        my ($file_type, $diff_status) = find_file_type_and_diff_status($path);
1016
1017        if (!$file_type && !$diff_status) {
1018                print STDERR "svn: '$path' is not under version control\n";
1019                exit 1;
1020        }
1021
1022        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1023        unless ($gs) {
1024                die "Unable to determine upstream SVN information from ",
1025                    "working tree history\n";
1026        }
1027
1028        # canonicalize_path() will return "" to make libsvn 1.5.x happy,
1029        $path = "." if $path eq "";
1030
1031        my $full_url = $url . ($fullpath eq "" ? "" : "/$fullpath");
1032
1033        if ($_url) {
1034                print escape_url($full_url), "\n";
1035                return;
1036        }
1037
1038        my $result = "Path: $path\n";
1039        $result .= "Name: " . basename($path) . "\n" if $file_type ne "dir";
1040        $result .= "URL: " . escape_url($full_url) . "\n";
1041
1042        eval {
1043                my $repos_root = $gs->repos_root;
1044                Git::SVN::remove_username($repos_root);
1045                $result .= "Repository Root: " . escape_url($repos_root) . "\n";
1046        };
1047        if ($@) {
1048                $result .= "Repository Root: (offline)\n";
1049        }
1050        $result .= "Repository UUID: $uuid\n" unless $diff_status eq "A" &&
1051                ($SVN::Core::VERSION le '1.5.4' || $file_type ne "dir");
1052        $result .= "Revision: " . ($diff_status eq "A" ? 0 : $rev) . "\n";
1053
1054        $result .= "Node Kind: " .
1055                   ($file_type eq "dir" ? "directory" : "file") . "\n";
1056
1057        my $schedule = $diff_status eq "A"
1058                       ? "add"
1059                       : ($diff_status eq "D" ? "delete" : "normal");
1060        $result .= "Schedule: $schedule\n";
1061
1062        if ($diff_status eq "A") {
1063                print $result, "\n";
1064                return;
1065        }
1066
1067        my ($lc_author, $lc_rev, $lc_date_utc);
1068        my @args = Git::SVN::Log::git_svn_log_cmd($rev, $rev, "--", $fullpath);
1069        my $log = command_output_pipe(@args);
1070        my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
1071        while (<$log>) {
1072                if (/^${esc_color}author (.+) <[^>]+> (\d+) ([\-\+]?\d+)$/o) {
1073                        $lc_author = $1;
1074                        $lc_date_utc = Git::SVN::Log::parse_git_date($2, $3);
1075                } elsif (/^${esc_color}    (git-svn-id:.+)$/o) {
1076                        (undef, $lc_rev, undef) = ::extract_metadata($1);
1077                }
1078        }
1079        close $log;
1080
1081        Git::SVN::Log::set_local_timezone();
1082
1083        $result .= "Last Changed Author: $lc_author\n";
1084        $result .= "Last Changed Rev: $lc_rev\n";
1085        $result .= "Last Changed Date: " .
1086                   Git::SVN::Log::format_svn_date($lc_date_utc) . "\n";
1087
1088        if ($file_type ne "dir") {
1089                my $text_last_updated_date =
1090                    ($diff_status eq "D" ? $lc_date_utc : (stat $path)[9]);
1091                $result .=
1092                    "Text Last Updated: " .
1093                    Git::SVN::Log::format_svn_date($text_last_updated_date) .
1094                    "\n";
1095                my $checksum;
1096                if ($diff_status eq "D") {
1097                        my ($fh, $ctx) =
1098                            command_output_pipe(qw(cat-file blob), "HEAD:$path");
1099                        if ($file_type eq "link") {
1100                                my $file_name = <$fh>;
1101                                $checksum = md5sum("link $file_name");
1102                        } else {
1103                                $checksum = md5sum($fh);
1104                        }
1105                        command_close_pipe($fh, $ctx);
1106                } elsif ($file_type eq "link") {
1107                        my $file_name =
1108                            command(qw(cat-file blob), "HEAD:$path");
1109                        $checksum =
1110                            md5sum("link " . $file_name);
1111                } else {
1112                        open FILE, "<", $path or die $!;
1113                        $checksum = md5sum(\*FILE);
1114                        close FILE or die $!;
1115                }
1116                $result .= "Checksum: " . $checksum . "\n";
1117        }
1118
1119        print $result, "\n";
1120}
1121
1122sub cmd_reset {
1123        my $target = shift || $_revision or die "SVN revision required\n";
1124        $target = $1 if $target =~ /^r(\d+)$/;
1125        $target =~ /^\d+$/ or die "Numeric SVN revision expected\n";
1126        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1127        unless ($gs) {
1128                die "Unable to determine upstream SVN information from ".
1129                    "history\n";
1130        }
1131        my ($r, $c) = $gs->find_rev_before($target, not $_fetch_parent);
1132        $gs->rev_map_set($r, $c, 'reset', $uuid);
1133        print "r$r = $c ($gs->{ref_id})\n";
1134}
1135
1136sub cmd_gc {
1137        if (!$can_compress) {
1138                warn "Compress::Zlib could not be found; unhandled.log " .
1139                     "files will not be compressed.\n";
1140        }
1141        find({ wanted => \&gc_directory, no_chdir => 1}, "$ENV{GIT_DIR}/svn");
1142}
1143
1144########################### utility functions #########################
1145
1146sub rebase_cmd {
1147        my @cmd = qw/rebase/;
1148        push @cmd, '-v' if $_verbose;
1149        push @cmd, qw/--merge/ if $_merge;
1150        push @cmd, "--strategy=$_strategy" if $_strategy;
1151        @cmd;
1152}
1153
1154sub post_fetch_checkout {
1155        return if $_no_checkout;
1156        my $gs = $Git::SVN::_head or return;
1157        return if verify_ref('refs/heads/master^0');
1158
1159        # look for "trunk" ref if it exists
1160        my $remote = Git::SVN::read_all_remotes()->{$gs->{repo_id}};
1161        my $fetch = $remote->{fetch};
1162        if ($fetch) {
1163                foreach my $p (keys %$fetch) {
1164                        basename($fetch->{$p}) eq 'trunk' or next;
1165                        $gs = Git::SVN->new($fetch->{$p}, $gs->{repo_id}, $p);
1166                        last;
1167                }
1168        }
1169
1170        my $valid_head = verify_ref('HEAD^0');
1171        command_noisy(qw(update-ref refs/heads/master), $gs->refname);
1172        return if ($valid_head || !verify_ref('HEAD^0'));
1173
1174        return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
1175        my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
1176        return if -f $index;
1177
1178        return if command_oneline(qw/rev-parse --is-inside-work-tree/) eq 'false';
1179        return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
1180        command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
1181        print STDERR "Checked out HEAD:\n  ",
1182                     $gs->full_url, " r", $gs->last_rev, "\n";
1183}
1184
1185sub complete_svn_url {
1186        my ($url, $path) = @_;
1187        $path =~ s#/+$##;
1188        if ($path !~ m#^[a-z\+]+://#) {
1189                if (!defined $url || $url !~ m#^[a-z\+]+://#) {
1190                        fatal("E: '$path' is not a complete URL ",
1191                              "and a separate URL is not specified");
1192                }
1193                return ($url, $path);
1194        }
1195        return ($path, '');
1196}
1197
1198sub complete_url_ls_init {
1199        my ($ra, $repo_path, $switch, $pfx) = @_;
1200        unless ($repo_path) {
1201                print STDERR "W: $switch not specified\n";
1202                return;
1203        }
1204        $repo_path =~ s#/+$##;
1205        if ($repo_path =~ m#^[a-z\+]+://#) {
1206                $ra = Git::SVN::Ra->new($repo_path);
1207                $repo_path = '';
1208        } else {
1209                $repo_path =~ s#^/+##;
1210                unless ($ra) {
1211                        fatal("E: '$repo_path' is not a complete URL ",
1212                              "and a separate URL is not specified");
1213                }
1214        }
1215        my $url = $ra->{url};
1216        my $gs = Git::SVN->init($url, undef, undef, undef, 1);
1217        my $k = "svn-remote.$gs->{repo_id}.url";
1218        my $orig_url = eval { command_oneline(qw/config --get/, $k) };
1219        if ($orig_url && ($orig_url ne $gs->{url})) {
1220                die "$k already set: $orig_url\n",
1221                    "wanted to set to: $gs->{url}\n";
1222        }
1223        command_oneline('config', $k, $gs->{url}) unless $orig_url;
1224        my $remote_path = "$gs->{path}/$repo_path";
1225        $remote_path =~ s#/+#/#g;
1226        $remote_path =~ s#^/##g;
1227        $remote_path .= "/*" if $remote_path !~ /\*/;
1228        my ($n) = ($switch =~ /^--(\w+)/);
1229        if (length $pfx && $pfx !~ m#/$#) {
1230                die "--prefix='$pfx' must have a trailing slash '/'\n";
1231        }
1232        command_noisy('config',
1233                      '--add',
1234                      "svn-remote.$gs->{repo_id}.$n",
1235                      "$remote_path:refs/remotes/$pfx*" .
1236                        ('/*' x (($remote_path =~ tr/*/*/) - 1)) );
1237}
1238
1239sub verify_ref {
1240        my ($ref) = @_;
1241        eval { command_oneline([ 'rev-parse', '--verify', $ref ],
1242                               { STDERR => 0 }); };
1243}
1244
1245sub get_tree_from_treeish {
1246        my ($treeish) = @_;
1247        # $treeish can be a symbolic ref, too:
1248        my $type = command_oneline(qw/cat-file -t/, $treeish);
1249        my $expected;
1250        while ($type eq 'tag') {
1251                ($treeish, $type) = command(qw/cat-file tag/, $treeish);
1252        }
1253        if ($type eq 'commit') {
1254                $expected = (grep /^tree /, command(qw/cat-file commit/,
1255                                                    $treeish))[0];
1256                ($expected) = ($expected =~ /^tree ($sha1)$/o);
1257                die "Unable to get tree from $treeish\n" unless $expected;
1258        } elsif ($type eq 'tree') {
1259                $expected = $treeish;
1260        } else {
1261                die "$treeish is a $type, expected tree, tag or commit\n";
1262        }
1263        return $expected;
1264}
1265
1266sub get_commit_entry {
1267        my ($treeish) = shift;
1268        my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
1269        my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
1270        my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
1271        open my $log_fh, '>', $commit_editmsg or croak $!;
1272
1273        my $type = command_oneline(qw/cat-file -t/, $treeish);
1274        if ($type eq 'commit' || $type eq 'tag') {
1275                my ($msg_fh, $ctx) = command_output_pipe('cat-file',
1276                                                         $type, $treeish);
1277                my $in_msg = 0;
1278                my $author;
1279                my $saw_from = 0;
1280                my $msgbuf = "";
1281                while (<$msg_fh>) {
1282                        if (!$in_msg) {
1283                                $in_msg = 1 if (/^\s*$/);
1284                                $author = $1 if (/^author (.*>)/);
1285                        } elsif (/^git-svn-id: /) {
1286                                # skip this for now, we regenerate the
1287                                # correct one on re-fetch anyways
1288                                # TODO: set *:merge properties or like...
1289                        } else {
1290                                if (/^From:/ || /^Signed-off-by:/) {
1291                                        $saw_from = 1;
1292                                }
1293                                $msgbuf .= $_;
1294                        }
1295                }
1296                $msgbuf =~ s/\s+$//s;
1297                if ($Git::SVN::_add_author_from && defined($author)
1298                    && !$saw_from) {
1299                        $msgbuf .= "\n\nFrom: $author";
1300                }
1301                print $log_fh $msgbuf or croak $!;
1302                command_close_pipe($msg_fh, $ctx);
1303        }
1304        close $log_fh or croak $!;
1305
1306        if ($_edit || ($type eq 'tree')) {
1307                my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
1308                # TODO: strip out spaces, comments, like git-commit.sh
1309                system($editor, $commit_editmsg);
1310        }
1311        rename $commit_editmsg, $commit_msg or croak $!;
1312        {
1313                require Encode;
1314                # SVN requires messages to be UTF-8 when entering the repo
1315                local $/;
1316                open $log_fh, '<', $commit_msg or croak $!;
1317                binmode $log_fh;
1318                chomp($log_entry{log} = <$log_fh>);
1319
1320                my $enc = Git::config('i18n.commitencoding') || 'UTF-8';
1321                my $msg = $log_entry{log};
1322
1323                eval { $msg = Encode::decode($enc, $msg, 1) };
1324                if ($@) {
1325                        die "Could not decode as $enc:\n", $msg,
1326                            "\nPerhaps you need to set i18n.commitencoding\n";
1327                }
1328
1329                eval { $msg = Encode::encode('UTF-8', $msg, 1) };
1330                die "Could not encode as UTF-8:\n$msg\n" if $@;
1331
1332                $log_entry{log} = $msg;
1333
1334                close $log_fh or croak $!;
1335        }
1336        unlink $commit_msg;
1337        \%log_entry;
1338}
1339
1340sub s_to_file {
1341        my ($str, $file, $mode) = @_;
1342        open my $fd,'>',$file or croak $!;
1343        print $fd $str,"\n" or croak $!;
1344        close $fd or croak $!;
1345        chmod ($mode &~ umask, $file) if (defined $mode);
1346}
1347
1348sub file_to_s {
1349        my $file = shift;
1350        open my $fd,'<',$file or croak "$!: file: $file\n";
1351        local $/;
1352        my $ret = <$fd>;
1353        close $fd or croak $!;
1354        $ret =~ s/\s*$//s;
1355        return $ret;
1356}
1357
1358# '<svn username> = real-name <email address>' mapping based on git-svnimport:
1359sub load_authors {
1360        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1361        my $log = $cmd eq 'log';
1362        while (<$authors>) {
1363                chomp;
1364                next unless /^(.+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
1365                my ($user, $name, $email) = ($1, $2, $3);
1366                if ($log) {
1367                        $Git::SVN::Log::rusers{"$name <$email>"} = $user;
1368                } else {
1369                        $users{$user} = [$name, $email];
1370                }
1371        }
1372        close $authors or croak $!;
1373}
1374
1375# convert GetOpt::Long specs for use by git-config
1376sub read_repo_config {
1377        return unless -d $ENV{GIT_DIR};
1378        my $opts = shift;
1379        my @config_only;
1380        foreach my $o (keys %$opts) {
1381                # if we have mixedCase and a long option-only, then
1382                # it's a config-only variable that we don't need for
1383                # the command-line.
1384                push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
1385                my $v = $opts->{$o};
1386                my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
1387                $key =~ s/-//g;
1388                my $arg = 'git config';
1389                $arg .= ' --int' if ($o =~ /[:=]i$/);
1390                $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
1391                if (ref $v eq 'ARRAY') {
1392                        chomp(my @tmp = `$arg --get-all svn.$key`);
1393                        @$v = @tmp if @tmp;
1394                } else {
1395                        chomp(my $tmp = `$arg --get svn.$key`);
1396                        if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
1397                                $$v = $tmp;
1398                        }
1399                }
1400        }
1401        delete @$opts{@config_only} if @config_only;
1402}
1403
1404sub extract_metadata {
1405        my $id = shift or return (undef, undef, undef);
1406        my ($url, $rev, $uuid) = ($id =~ /^\s*git-svn-id:\s+(.*)\@(\d+)
1407                                                        \s([a-f\d\-]+)$/ix);
1408        if (!defined $rev || !$uuid || !$url) {
1409                # some of the original repositories I made had
1410                # identifiers like this:
1411                ($rev, $uuid) = ($id =~/^\s*git-svn-id:\s(\d+)\@([a-f\d\-]+)/i);
1412        }
1413        return ($url, $rev, $uuid);
1414}
1415
1416sub cmt_metadata {
1417        return extract_metadata((grep(/^git-svn-id: /,
1418                command(qw/cat-file commit/, shift)))[-1]);
1419}
1420
1421sub cmt_sha2rev_batch {
1422        my %s2r;
1423        my ($pid, $in, $out, $ctx) = command_bidi_pipe(qw/cat-file --batch/);
1424        my $list = shift;
1425
1426        foreach my $sha (@{$list}) {
1427                my $first = 1;
1428                my $size = 0;
1429                print $out $sha, "\n";
1430
1431                while (my $line = <$in>) {
1432                        if ($first && $line =~ /^[[:xdigit:]]{40}\smissing$/) {
1433                                last;
1434                        } elsif ($first &&
1435                               $line =~ /^[[:xdigit:]]{40}\scommit\s(\d+)$/) {
1436                                $first = 0;
1437                                $size = $1;
1438                                next;
1439                        } elsif ($line =~ /^(git-svn-id: )/) {
1440                                my (undef, $rev, undef) =
1441                                                      extract_metadata($line);
1442                                $s2r{$sha} = $rev;
1443                        }
1444
1445                        $size -= length($line);
1446                        last if ($size == 0);
1447                }
1448        }
1449
1450        command_close_bidi_pipe($pid, $in, $out, $ctx);
1451
1452        return \%s2r;
1453}
1454
1455sub working_head_info {
1456        my ($head, $refs) = @_;
1457        my @args = ('log', '--no-color', '--first-parent', '--pretty=medium');
1458        my ($fh, $ctx) = command_output_pipe(@args, $head);
1459        my $hash;
1460        my %max;
1461        while (<$fh>) {
1462                if ( m{^commit ($::sha1)$} ) {
1463                        unshift @$refs, $hash if $hash and $refs;
1464                        $hash = $1;
1465                        next;
1466                }
1467                next unless s{^\s*(git-svn-id:)}{$1};
1468                my ($url, $rev, $uuid) = extract_metadata($_);
1469                if (defined $url && defined $rev) {
1470                        next if $max{$url} and $max{$url} < $rev;
1471                        if (my $gs = Git::SVN->find_by_url($url)) {
1472                                my $c = $gs->rev_map_get($rev, $uuid);
1473                                if ($c && $c eq $hash) {
1474                                        close $fh; # break the pipe
1475                                        return ($url, $rev, $uuid, $gs);
1476                                } else {
1477                                        $max{$url} ||= $gs->rev_map_max;
1478                                }
1479                        }
1480                }
1481        }
1482        command_close_pipe($fh, $ctx);
1483        (undef, undef, undef, undef);
1484}
1485
1486sub read_commit_parents {
1487        my ($parents, $c) = @_;
1488        chomp(my $p = command_oneline(qw/rev-list --parents -1/, $c));
1489        $p =~ s/^($c)\s*// or die "rev-list --parents -1 $c failed!\n";
1490        @{$parents->{$c}} = split(/ /, $p);
1491}
1492
1493sub linearize_history {
1494        my ($gs, $refs) = @_;
1495        my %parents;
1496        foreach my $c (@$refs) {
1497                read_commit_parents(\%parents, $c);
1498        }
1499
1500        my @linear_refs;
1501        my %skip = ();
1502        my $last_svn_commit = $gs->last_commit;
1503        foreach my $c (reverse @$refs) {
1504                next if $c eq $last_svn_commit;
1505                last if $skip{$c};
1506
1507                unshift @linear_refs, $c;
1508                $skip{$c} = 1;
1509
1510                # we only want the first parent to diff against for linear
1511                # history, we save the rest to inject when we finalize the
1512                # svn commit
1513                my $fp_a = verify_ref("$c~1");
1514                my $fp_b = shift @{$parents{$c}} if $parents{$c};
1515                if (!$fp_a || !$fp_b) {
1516                        die "Commit $c\n",
1517                            "has no parent commit, and therefore ",
1518                            "nothing to diff against.\n",
1519                            "You should be working from a repository ",
1520                            "originally created by git-svn\n";
1521                }
1522                if ($fp_a ne $fp_b) {
1523                        die "$c~1 = $fp_a, however parsing commit $c ",
1524                            "revealed that:\n$c~1 = $fp_b\nBUG!\n";
1525                }
1526
1527                foreach my $p (@{$parents{$c}}) {
1528                        $skip{$p} = 1;
1529                }
1530        }
1531        (\@linear_refs, \%parents);
1532}
1533
1534sub find_file_type_and_diff_status {
1535        my ($path) = @_;
1536        return ('dir', '') if $path eq '';
1537
1538        my $diff_output =
1539            command_oneline(qw(diff --cached --name-status --), $path) || "";
1540        my $diff_status = (split(' ', $diff_output))[0] || "";
1541
1542        my $ls_tree = command_oneline(qw(ls-tree HEAD), $path) || "";
1543
1544        return (undef, undef) if !$diff_status && !$ls_tree;
1545
1546        if ($diff_status eq "A") {
1547                return ("link", $diff_status) if -l $path;
1548                return ("dir", $diff_status) if -d $path;
1549                return ("file", $diff_status);
1550        }
1551
1552        my $mode = (split(' ', $ls_tree))[0] || "";
1553
1554        return ("link", $diff_status) if $mode eq "120000";
1555        return ("dir", $diff_status) if $mode eq "040000";
1556        return ("file", $diff_status);
1557}
1558
1559sub md5sum {
1560        my $arg = shift;
1561        my $ref = ref $arg;
1562        my $md5 = Digest::MD5->new();
1563        if ($ref eq 'GLOB' || $ref eq 'IO::File' || $ref eq 'File::Temp') {
1564                $md5->addfile($arg) or croak $!;
1565        } elsif ($ref eq 'SCALAR') {
1566                $md5->add($$arg) or croak $!;
1567        } elsif (!$ref) {
1568                $md5->add($arg) or croak $!;
1569        } else {
1570                ::fatal "Can't provide MD5 hash for unknown ref type: '", $ref, "'";
1571        }
1572        return $md5->hexdigest();
1573}
1574
1575sub gc_directory {
1576        if ($can_compress && -f $_ && basename($_) eq "unhandled.log") {
1577                my $out_filename = $_ . ".gz";
1578                open my $in_fh, "<", $_ or die "Unable to open $_: $!\n";
1579                binmode $in_fh;
1580                my $gz = Compress::Zlib::gzopen($out_filename, "ab") or
1581                                die "Unable to open $out_filename: $!\n";
1582
1583                my $res;
1584                while ($res = sysread($in_fh, my $str, 1024)) {
1585                        $gz->gzwrite($str) or
1586                                die "Unable to write: ".$gz->gzerror()."!\n";
1587                }
1588                unlink $_ or die "unlink $File::Find::name: $!\n";
1589        } elsif (-f $_ && basename($_) eq "index") {
1590                unlink $_ or die "unlink $_: $!\n";
1591        }
1592}
1593
1594package Git::SVN;
1595use strict;
1596use warnings;
1597use Fcntl qw/:DEFAULT :seek/;
1598use constant rev_map_fmt => 'NH40';
1599use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
1600            $_repack $_repack_flags $_use_svm_props $_head
1601            $_use_svnsync_props $no_reuse_existing $_minimize_url
1602            $_use_log_author $_add_author_from $_localtime/;
1603use Carp qw/croak/;
1604use File::Path qw/mkpath/;
1605use File::Copy qw/copy/;
1606use IPC::Open3;
1607
1608my ($_gc_nr, $_gc_period);
1609
1610# properties that we do not log:
1611my %SKIP_PROP;
1612BEGIN {
1613        %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
1614                                        svn:special svn:executable
1615                                        svn:entry:committed-rev
1616                                        svn:entry:last-author
1617                                        svn:entry:uuid
1618                                        svn:entry:committed-date/;
1619
1620        # some options are read globally, but can be overridden locally
1621        # per [svn-remote "..."] section.  Command-line options will *NOT*
1622        # override options set in an [svn-remote "..."] section
1623        no strict 'refs';
1624        for my $option (qw/follow_parent no_metadata use_svm_props
1625                           use_svnsync_props/) {
1626                my $key = $option;
1627                $key =~ tr/_//d;
1628                my $prop = "-$option";
1629                *$option = sub {
1630                        my ($self) = @_;
1631                        return $self->{$prop} if exists $self->{$prop};
1632                        my $k = "svn-remote.$self->{repo_id}.$key";
1633                        eval { command_oneline(qw/config --get/, $k) };
1634                        if ($@) {
1635                                $self->{$prop} = ${"Git::SVN::_$option"};
1636                        } else {
1637                                my $v = command_oneline(qw/config --bool/,$k);
1638                                $self->{$prop} = $v eq 'false' ? 0 : 1;
1639                        }
1640                        return $self->{$prop};
1641                }
1642        }
1643}
1644
1645
1646my (%LOCKFILES, %INDEX_FILES);
1647END {
1648        unlink keys %LOCKFILES if %LOCKFILES;
1649        unlink keys %INDEX_FILES if %INDEX_FILES;
1650}
1651
1652sub resolve_local_globs {
1653        my ($url, $fetch, $glob_spec) = @_;
1654        return unless defined $glob_spec;
1655        my $ref = $glob_spec->{ref};
1656        my $path = $glob_spec->{path};
1657        foreach (command(qw#for-each-ref --format=%(refname) refs/remotes#)) {
1658                next unless m#^refs/remotes/$ref->{regex}$#;
1659                my $p = $1;
1660                my $pathname = desanitize_refname($path->full_path($p));
1661                my $refname = desanitize_refname($ref->full_path($p));
1662                if (my $existing = $fetch->{$pathname}) {
1663                        if ($existing ne $refname) {
1664                                die "Refspec conflict:\n",
1665                                    "existing: refs/remotes/$existing\n",
1666                                    " globbed: refs/remotes/$refname\n";
1667                        }
1668                        my $u = (::cmt_metadata("refs/remotes/$refname"))[0];
1669                        $u =~ s!^\Q$url\E(/|$)!! or die
1670                          "refs/remotes/$refname: '$url' not found in '$u'\n";
1671                        if ($pathname ne $u) {
1672                                warn "W: Refspec glob conflict ",
1673                                     "(ref: refs/remotes/$refname):\n",
1674                                     "expected path: $pathname\n",
1675                                     "    real path: $u\n",
1676                                     "Continuing ahead with $u\n";
1677                                next;
1678                        }
1679                } else {
1680                        $fetch->{$pathname} = $refname;
1681                }
1682        }
1683}
1684
1685sub parse_revision_argument {
1686        my ($base, $head) = @_;
1687        if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
1688                return ($base, $head);
1689        }
1690        return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
1691        return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
1692        return ($head, $head) if ($::_revision eq 'HEAD');
1693        return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
1694        return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
1695        die "revision argument: $::_revision not understood by git-svn\n";
1696}
1697
1698sub fetch_all {
1699        my ($repo_id, $remotes) = @_;
1700        if (ref $repo_id) {
1701                my $gs = $repo_id;
1702                $repo_id = undef;
1703                $repo_id = $gs->{repo_id};
1704        }
1705        $remotes ||= read_all_remotes();
1706        my $remote = $remotes->{$repo_id} or
1707                     die "[svn-remote \"$repo_id\"] unknown\n";
1708        my $fetch = $remote->{fetch};
1709        my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
1710        my (@gs, @globs);
1711        my $ra = Git::SVN::Ra->new($url);
1712        my $uuid = $ra->get_uuid;
1713        my $head = $ra->get_latest_revnum;
1714        $ra->get_log("", $head, 0, 1, 0, 1, sub { $head = $_[1] });
1715        my $base = defined $fetch ? $head : 0;
1716
1717        # read the max revs for wildcard expansion (branches/*, tags/*)
1718        foreach my $t (qw/branches tags/) {
1719                defined $remote->{$t} or next;
1720                push @globs, @{$remote->{$t}};
1721
1722                my $max_rev = eval { tmp_config(qw/--int --get/,
1723                                         "svn-remote.$repo_id.${t}-maxRev") };
1724                if (defined $max_rev && ($max_rev < $base)) {
1725                        $base = $max_rev;
1726                } elsif (!defined $max_rev) {
1727                        $base = 0;
1728                }
1729        }
1730
1731        if ($fetch) {
1732                foreach my $p (sort keys %$fetch) {
1733                        my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
1734                        my $lr = $gs->rev_map_max;
1735                        if (defined $lr) {
1736                                $base = $lr if ($lr < $base);
1737                        }
1738                        push @gs, $gs;
1739                }
1740        }
1741
1742        ($base, $head) = parse_revision_argument($base, $head);
1743        $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
1744}
1745
1746sub read_all_remotes {
1747        my $r = {};
1748        my $use_svm_props = eval { command_oneline(qw/config --bool
1749            svn.useSvmProps/) };
1750        $use_svm_props = $use_svm_props eq 'true' if $use_svm_props;
1751        foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
1752                if (m!^(.+)\.fetch=\s*(.*)\s*:\s*(.+)\s*$!) {
1753                        my ($remote, $local_ref, $_remote_ref) = ($1, $2, $3);
1754                        die("svn-remote.$remote: remote ref '$_remote_ref' "
1755                            . "must start with 'refs/remotes/'\n")
1756                                unless $_remote_ref =~ m{^refs/remotes/(.+)};
1757                        my $remote_ref = $1;
1758                        $local_ref =~ s{^/}{};
1759                        $r->{$remote}->{fetch}->{$local_ref} = $remote_ref;
1760                        $r->{$remote}->{svm} = {} if $use_svm_props;
1761                } elsif (m!^(.+)\.usesvmprops=\s*(.*)\s*$!) {
1762                        $r->{$1}->{svm} = {};
1763                } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
1764                        $r->{$1}->{url} = $2;
1765                } elsif (m!^(.+)\.(branches|tags)=
1766                           (.*):refs/remotes/(.+)\s*$/!x) {
1767                        my ($p, $g) = ($3, $4);
1768                        my $rs = {
1769                            t => $2,
1770                            remote => $1,
1771                            path => Git::SVN::GlobSpec->new($p),
1772                            ref => Git::SVN::GlobSpec->new($g) };
1773                        if (length($rs->{ref}->{right}) != 0) {
1774                                die "The '*' glob character must be the last ",
1775                                    "character of '$g'\n";
1776                        }
1777                        push @{ $r->{$1}->{$2} }, $rs;
1778                }
1779        }
1780
1781        map {
1782                if (defined $r->{$_}->{svm}) {
1783                        my $svm;
1784                        eval {
1785                                my $section = "svn-remote.$_";
1786                                $svm = {
1787                                        source => tmp_config('--get',
1788                                            "$section.svm-source"),
1789                                        replace => tmp_config('--get',
1790                                            "$section.svm-replace"),
1791                                }
1792                        };
1793                        $r->{$_}->{svm} = $svm;
1794                }
1795        } keys %$r;
1796
1797        $r;
1798}
1799
1800sub init_vars {
1801        $_gc_nr = $_gc_period = 1000;
1802        if (defined $_repack || defined $_repack_flags) {
1803               warn "Repack options are obsolete; they have no effect.\n";
1804        }
1805}
1806
1807sub verify_remotes_sanity {
1808        return unless -d $ENV{GIT_DIR};
1809        my %seen;
1810        foreach (command(qw/config -l/)) {
1811                if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
1812                        if ($seen{$1}) {
1813                                die "Remote ref refs/remote/$1 is tracked by",
1814                                    "\n  \"$_\"\nand\n  \"$seen{$1}\"\n",
1815                                    "Please resolve this ambiguity in ",
1816                                    "your git configuration file before ",
1817                                    "continuing\n";
1818                        }
1819                        $seen{$1} = $_;
1820                }
1821        }
1822}
1823
1824sub find_existing_remote {
1825        my ($url, $remotes) = @_;
1826        return undef if $no_reuse_existing;
1827        my $existing;
1828        foreach my $repo_id (keys %$remotes) {
1829                my $u = $remotes->{$repo_id}->{url} or next;
1830                next if $u ne $url;
1831                $existing = $repo_id;
1832                last;
1833        }
1834        $existing;
1835}
1836
1837sub init_remote_config {
1838        my ($self, $url, $no_write) = @_;
1839        $url =~ s!/+$!!; # strip trailing slash
1840        my $r = read_all_remotes();
1841        my $existing = find_existing_remote($url, $r);
1842        if ($existing) {
1843                unless ($no_write) {
1844                        print STDERR "Using existing ",
1845                                     "[svn-remote \"$existing\"]\n";
1846                }
1847                $self->{repo_id} = $existing;
1848        } elsif ($_minimize_url) {
1849                my $min_url = Git::SVN::Ra->new($url)->minimize_url;
1850                $existing = find_existing_remote($min_url, $r);
1851                if ($existing) {
1852                        unless ($no_write) {
1853                                print STDERR "Using existing ",
1854                                             "[svn-remote \"$existing\"]\n";
1855                        }
1856                        $self->{repo_id} = $existing;
1857                }
1858                if ($min_url ne $url) {
1859                        unless ($no_write) {
1860                                print STDERR "Using higher level of URL: ",
1861                                             "$url => $min_url\n";
1862                        }
1863                        my $old_path = $self->{path};
1864                        $self->{path} = $url;
1865                        $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
1866                        if (length $old_path) {
1867                                $self->{path} .= "/$old_path";
1868                        }
1869                        $url = $min_url;
1870                }
1871        }
1872        my $orig_url;
1873        if (!$existing) {
1874                # verify that we aren't overwriting anything:
1875                $orig_url = eval {
1876                        command_oneline('config', '--get',
1877                                        "svn-remote.$self->{repo_id}.url")
1878                };
1879                if ($orig_url && ($orig_url ne $url)) {
1880                        die "svn-remote.$self->{repo_id}.url already set: ",
1881                            "$orig_url\nwanted to set to: $url\n";
1882                }
1883        }
1884        my ($xrepo_id, $xpath) = find_ref($self->refname);
1885        if (defined $xpath) {
1886                die "svn-remote.$xrepo_id.fetch already set to track ",
1887                    "$xpath:refs/remotes/", $self->refname, "\n";
1888        }
1889        unless ($no_write) {
1890                command_noisy('config',
1891                              "svn-remote.$self->{repo_id}.url", $url);
1892                $self->{path} =~ s{^/}{};
1893                command_noisy('config', '--add',
1894                              "svn-remote.$self->{repo_id}.fetch",
1895                              "$self->{path}:".$self->refname);
1896        }
1897        $self->{url} = $url;
1898}
1899
1900sub find_by_url { # repos_root and, path are optional
1901        my ($class, $full_url, $repos_root, $path) = @_;
1902
1903        return undef unless defined $full_url;
1904        remove_username($full_url);
1905        remove_username($repos_root) if defined $repos_root;
1906        my $remotes = read_all_remotes();
1907        if (defined $full_url && defined $repos_root && !defined $path) {
1908                $path = $full_url;
1909                $path =~ s#^\Q$repos_root\E(?:/|$)##;
1910        }
1911        foreach my $repo_id (keys %$remotes) {
1912                my $u = $remotes->{$repo_id}->{url} or next;
1913                remove_username($u);
1914                next if defined $repos_root && $repos_root ne $u;
1915
1916                my $fetch = $remotes->{$repo_id}->{fetch} || {};
1917                foreach my $t (qw/branches tags/) {
1918                        foreach my $globspec (@{$remotes->{$repo_id}->{$t}}) {
1919                                resolve_local_globs($u, $fetch, $globspec);
1920                        }
1921                }
1922                my $p = $path;
1923                my $rwr = rewrite_root({repo_id => $repo_id});
1924                my $svm = $remotes->{$repo_id}->{svm}
1925                        if defined $remotes->{$repo_id}->{svm};
1926                unless (defined $p) {
1927                        $p = $full_url;
1928                        my $z = $u;
1929                        my $prefix = '';
1930                        if ($rwr) {
1931                                $z = $rwr;
1932                                remove_username($z);
1933                        } elsif (defined $svm) {
1934                                $z = $svm->{source};
1935                                $prefix = $svm->{replace};
1936                                $prefix =~ s#^\Q$u\E(?:/|$)##;
1937                                $prefix =~ s#/$##;
1938                        }
1939                        $p =~ s#^\Q$z\E(?:/|$)#$prefix# or next;
1940                }
1941                foreach my $f (keys %$fetch) {
1942                        next if $f ne $p;
1943                        return Git::SVN->new($fetch->{$f}, $repo_id, $f);
1944                }
1945        }
1946        undef;
1947}
1948
1949sub init {
1950        my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
1951        my $self = _new($class, $repo_id, $ref_id, $path);
1952        if (defined $url) {
1953                $self->init_remote_config($url, $no_write);
1954        }
1955        $self;
1956}
1957
1958sub find_ref {
1959        my ($ref_id) = @_;
1960        foreach (command(qw/config -l/)) {
1961                next unless m!^svn-remote\.(.+)\.fetch=
1962                              \s*(.*)\s*:\s*refs/remotes/(.+)\s*$!x;
1963                my ($repo_id, $path, $ref) = ($1, $2, $3);
1964                if ($ref eq $ref_id) {
1965                        $path = '' if ($path =~ m#^\./?#);
1966                        return ($repo_id, $path);
1967                }
1968        }
1969        (undef, undef, undef);
1970}
1971
1972sub new {
1973        my ($class, $ref_id, $repo_id, $path) = @_;
1974        if (defined $ref_id && !defined $repo_id && !defined $path) {
1975                ($repo_id, $path) = find_ref($ref_id);
1976                if (!defined $repo_id) {
1977                        die "Could not find a \"svn-remote.*.fetch\" key ",
1978                            "in the repository configuration matching: ",
1979                            "refs/remotes/$ref_id\n";
1980                }
1981        }
1982        my $self = _new($class, $repo_id, $ref_id, $path);
1983        if (!defined $self->{path} || !length $self->{path}) {
1984                my $fetch = command_oneline('config', '--get',
1985                                            "svn-remote.$repo_id.fetch",
1986                                            ":refs/remotes/$ref_id\$") or
1987                     die "Failed to read \"svn-remote.$repo_id.fetch\" ",
1988                         "\":refs/remotes/$ref_id\$\" in config\n";
1989                ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
1990        }
1991        $self->{url} = command_oneline('config', '--get',
1992                                       "svn-remote.$repo_id.url") or
1993                  die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
1994        $self->rebuild;
1995        $self;
1996}
1997
1998sub refname {
1999        my ($refname) = "refs/remotes/$_[0]->{ref_id}" ;
2000
2001        # It cannot end with a slash /, we'll throw up on this because
2002        # SVN can't have directories with a slash in their name, either:
2003        if ($refname =~ m{/$}) {
2004                die "ref: '$refname' ends with a trailing slash, this is ",
2005                    "not permitted by git nor Subversion\n";
2006        }
2007
2008        # It cannot have ASCII control character space, tilde ~, caret ^,
2009        # colon :, question-mark ?, asterisk *, space, or open bracket [
2010        # anywhere.
2011        #
2012        # Additionally, % must be escaped because it is used for escaping
2013        # and we want our escaped refname to be reversible
2014        $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
2015
2016        # no slash-separated component can begin with a dot .
2017        # /.* becomes /%2E*
2018        $refname =~ s{/\.}{/%2E}g;
2019
2020        # It cannot have two consecutive dots .. anywhere
2021        # .. becomes %2E%2E
2022        $refname =~ s{\.\.}{%2E%2E}g;
2023
2024        return $refname;
2025}
2026
2027sub desanitize_refname {
2028        my ($refname) = @_;
2029        $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
2030        return $refname;
2031}
2032
2033sub svm_uuid {
2034        my ($self) = @_;
2035        return $self->{svm}->{uuid} if $self->svm;
2036        $self->ra;
2037        unless ($self->{svm}) {
2038                die "SVM UUID not cached, and reading remotely failed\n";
2039        }
2040        $self->{svm}->{uuid};
2041}
2042
2043sub svm {
2044        my ($self) = @_;
2045        return $self->{svm} if $self->{svm};
2046        my $svm;
2047        # see if we have it in our config, first:
2048        eval {
2049                my $section = "svn-remote.$self->{repo_id}";
2050                $svm = {
2051                  source => tmp_config('--get', "$section.svm-source"),
2052                  uuid => tmp_config('--get', "$section.svm-uuid"),
2053                  replace => tmp_config('--get', "$section.svm-replace"),
2054                }
2055        };
2056        if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
2057                $self->{svm} = $svm;
2058        }
2059        $self->{svm};
2060}
2061
2062sub _set_svm_vars {
2063        my ($self, $ra) = @_;
2064        return $ra if $self->svm;
2065
2066        my @err = ( "useSvmProps set, but failed to read SVM properties\n",
2067                    "(svm:source, svm:uuid) ",
2068                    "from the following URLs:\n" );
2069        sub read_svm_props {
2070                my ($self, $ra, $path, $r) = @_;
2071                my $props = ($ra->get_dir($path, $r))[2];
2072                my $src = $props->{'svm:source'};
2073                my $uuid = $props->{'svm:uuid'};
2074                return undef if (!$src || !$uuid);
2075
2076                chomp($src, $uuid);
2077
2078                $uuid =~ m{^[0-9a-f\-]{30,}$}i
2079                    or die "doesn't look right - svm:uuid is '$uuid'\n";
2080
2081                # the '!' is used to mark the repos_root!/relative/path
2082                $src =~ s{/?!/?}{/};
2083                $src =~ s{/+$}{}; # no trailing slashes please
2084                # username is of no interest
2085                $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
2086
2087                my $replace = $ra->{url};
2088                $replace .= "/$path" if length $path;
2089
2090                my $section = "svn-remote.$self->{repo_id}";
2091                tmp_config("$section.svm-source", $src);
2092                tmp_config("$section.svm-replace", $replace);
2093                tmp_config("$section.svm-uuid", $uuid);
2094                $self->{svm} = {
2095                        source => $src,
2096                        uuid => $uuid,
2097                        replace => $replace
2098                };
2099        }
2100
2101        my $r = $ra->get_latest_revnum;
2102        my $path = $self->{path};
2103        my %tried;
2104        while (length $path) {
2105                unless ($tried{"$self->{url}/$path"}) {
2106                        return $ra if $self->read_svm_props($ra, $path, $r);
2107                        $tried{"$self->{url}/$path"} = 1;
2108                }
2109                $path =~ s#/?[^/]+$##;
2110        }
2111        die "Path: '$path' should be ''\n" if $path ne '';
2112        return $ra if $self->read_svm_props($ra, $path, $r);
2113        $tried{"$self->{url}/$path"} = 1;
2114
2115        if ($ra->{repos_root} eq $self->{url}) {
2116                die @err, (map { "  $_\n" } keys %tried), "\n";
2117        }
2118
2119        # nope, make sure we're connected to the repository root:
2120        my $ok;
2121        my @tried_b;
2122        $path = $ra->{svn_path};
2123        $ra = Git::SVN::Ra->new($ra->{repos_root});
2124        while (length $path) {
2125                unless ($tried{"$ra->{url}/$path"}) {
2126                        $ok = $self->read_svm_props($ra, $path, $r);
2127                        last if $ok;
2128                        $tried{"$ra->{url}/$path"} = 1;
2129                }
2130                $path =~ s#/?[^/]+$##;
2131        }
2132        die "Path: '$path' should be ''\n" if $path ne '';
2133        $ok ||= $self->read_svm_props($ra, $path, $r);
2134        $tried{"$ra->{url}/$path"} = 1;
2135        if (!$ok) {
2136                die @err, (map { "  $_\n" } keys %tried), "\n";
2137        }
2138        Git::SVN::Ra->new($self->{url});
2139}
2140
2141sub svnsync {
2142        my ($self) = @_;
2143        return $self->{svnsync} if $self->{svnsync};
2144
2145        if ($self->no_metadata) {
2146                die "Can't have both 'noMetadata' and ",
2147                    "'useSvnsyncProps' options set!\n";
2148        }
2149        if ($self->rewrite_root) {
2150                die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
2151                    "options set!\n";
2152        }
2153
2154        my $svnsync;
2155        # see if we have it in our config, first:
2156        eval {
2157                my $section = "svn-remote.$self->{repo_id}";
2158
2159                my $url = tmp_config('--get', "$section.svnsync-url");
2160                ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2161                   die "doesn't look right - svn:sync-from-url is '$url'\n";
2162
2163                my $uuid = tmp_config('--get', "$section.svnsync-uuid");
2164                ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2165                   die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2166
2167                $svnsync = { url => $url, uuid => $uuid }
2168        };
2169        if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
2170                return $self->{svnsync} = $svnsync;
2171        }
2172
2173        my $err = "useSvnsyncProps set, but failed to read " .
2174                  "svnsync property: svn:sync-from-";
2175        my $rp = $self->ra->rev_proplist(0);
2176
2177        my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
2178        ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2179                   die "doesn't look right - svn:sync-from-url is '$url'\n";
2180
2181        my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
2182        ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2183                   die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2184
2185        my $section = "svn-remote.$self->{repo_id}";
2186        tmp_config('--add', "$section.svnsync-uuid", $uuid);
2187        tmp_config('--add', "$section.svnsync-url", $url);
2188        return $self->{svnsync} = { url => $url, uuid => $uuid };
2189}
2190
2191# this allows us to memoize our SVN::Ra UUID locally and avoid a
2192# remote lookup (useful for 'git svn log').
2193sub ra_uuid {
2194        my ($self) = @_;
2195        unless ($self->{ra_uuid}) {
2196                my $key = "svn-remote.$self->{repo_id}.uuid";
2197                my $uuid = eval { tmp_config('--get', $key) };
2198                if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/i) {
2199                        $self->{ra_uuid} = $uuid;
2200                } else {
2201                        die "ra_uuid called without URL\n" unless $self->{url};
2202                        $self->{ra_uuid} = $self->ra->get_uuid;
2203                        tmp_config('--add', $key, $self->{ra_uuid});
2204                }
2205        }
2206        $self->{ra_uuid};
2207}
2208
2209sub _set_repos_root {
2210        my ($self, $repos_root) = @_;
2211        my $k = "svn-remote.$self->{repo_id}.reposRoot";
2212        $repos_root ||= $self->ra->{repos_root};
2213        tmp_config($k, $repos_root);
2214        $repos_root;
2215}
2216
2217sub repos_root {
2218        my ($self) = @_;
2219        my $k = "svn-remote.$self->{repo_id}.reposRoot";
2220        eval { tmp_config('--get', $k) } || $self->_set_repos_root;
2221}
2222
2223sub ra {
2224        my ($self) = shift;
2225        my $ra = Git::SVN::Ra->new($self->{url});
2226        $self->_set_repos_root($ra->{repos_root});
2227        if ($self->use_svm_props && !$self->{svm}) {
2228                if ($self->no_metadata) {
2229                        die "Can't have both 'noMetadata' and ",
2230                            "'useSvmProps' options set!\n";
2231                } elsif ($self->use_svnsync_props) {
2232                        die "Can't have both 'useSvnsyncProps' and ",
2233                            "'useSvmProps' options set!\n";
2234                }
2235                $ra = $self->_set_svm_vars($ra);
2236                $self->{-want_revprops} = 1;
2237        }
2238        $ra;
2239}
2240
2241# prop_walk(PATH, REV, SUB)
2242# -------------------------
2243# Recursively traverse PATH at revision REV and invoke SUB for each
2244# directory that contains a SVN property.  SUB will be invoked as
2245# follows:  &SUB(gs, path, props);  where `gs' is this instance of
2246# Git::SVN, `path' the path to the directory where the properties
2247# `props' were found.  The `path' will be relative to point of checkout,
2248# that is, if url://repo/trunk is the current Git branch, and that
2249# directory contains a sub-directory `d', SUB will be invoked with `/d/'
2250# as `path' (note the trailing `/').
2251sub prop_walk {
2252        my ($self, $path, $rev, $sub) = @_;
2253
2254        $path =~ s#^/##;
2255        my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev);
2256        $path =~ s#^/*#/#g;
2257        my $p = $path;
2258        # Strip the irrelevant part of the path.
2259        $p =~ s#^/+\Q$self->{path}\E(/|$)#/#;
2260        # Ensure the path is terminated by a `/'.
2261        $p =~ s#/*$#/#;
2262
2263        # The properties contain all the internal SVN stuff nobody
2264        # (usually) cares about.
2265        my $interesting_props = 0;
2266        foreach (keys %{$props}) {
2267                # If it doesn't start with `svn:', it must be a
2268                # user-defined property.
2269                ++$interesting_props and next if $_ !~ /^svn:/;
2270                # FIXME: Fragile, if SVN adds new public properties,
2271                # this needs to be updated.
2272                ++$interesting_props if /^svn:(?:ignore|keywords|executable
2273                                                 |eol-style|mime-type
2274                                                 |externals|needs-lock)$/x;
2275        }
2276        &$sub($self, $p, $props) if $interesting_props;
2277
2278        foreach (sort keys %$dirent) {
2279                next if $dirent->{$_}->{kind} != $SVN::Node::dir;
2280                $self->prop_walk($self->{path} . $p . $_, $rev, $sub);
2281        }
2282}
2283
2284sub last_rev { ($_[0]->last_rev_commit)[0] }
2285sub last_commit { ($_[0]->last_rev_commit)[1] }
2286
2287# returns the newest SVN revision number and newest commit SHA1
2288sub last_rev_commit {
2289        my ($self) = @_;
2290        if (defined $self->{last_rev} && defined $self->{last_commit}) {
2291                return ($self->{last_rev}, $self->{last_commit});
2292        }
2293        my $c = ::verify_ref($self->refname.'^0');
2294        if ($c && !$self->use_svm_props && !$self->no_metadata) {
2295                my $rev = (::cmt_metadata($c))[1];
2296                if (defined $rev) {
2297                        ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
2298                        return ($rev, $c);
2299                }
2300        }
2301        my $map_path = $self->map_path;
2302        unless (-e $map_path) {
2303                ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
2304                return (undef, undef);
2305        }
2306        my ($rev, $commit) = $self->rev_map_max(1);
2307        ($self->{last_rev}, $self->{last_commit}) = ($rev, $commit);
2308        return ($rev, $commit);
2309}
2310
2311sub get_fetch_range {
2312        my ($self, $min, $max) = @_;
2313        $max ||= $self->ra->get_latest_revnum;
2314        $min ||= $self->rev_map_max;
2315        (++$min, $max);
2316}
2317
2318sub tmp_config {
2319        my (@args) = @_;
2320        my $old_def_config = "$ENV{GIT_DIR}/svn/config";
2321        my $config = "$ENV{GIT_DIR}/svn/.metadata";
2322        if (! -f $config && -f $old_def_config) {
2323                rename $old_def_config, $config or
2324                       die "Failed rename $old_def_config => $config: $!\n";
2325        }
2326        my $old_config = $ENV{GIT_CONFIG};
2327        $ENV{GIT_CONFIG} = $config;
2328        $@ = undef;
2329        my @ret = eval {
2330                unless (-f $config) {
2331                        mkfile($config);
2332                        open my $fh, '>', $config or
2333                            die "Can't open $config: $!\n";
2334                        print $fh "; This file is used internally by ",
2335                                  "git-svn\n" or die
2336                                  "Couldn't write to $config: $!\n";
2337                        print $fh "; You should not have to edit it\n" or
2338                              die "Couldn't write to $config: $!\n";
2339                        close $fh or die "Couldn't close $config: $!\n";
2340                }
2341                command('config', @args);
2342        };
2343        my $err = $@;
2344        if (defined $old_config) {
2345                $ENV{GIT_CONFIG} = $old_config;
2346        } else {
2347                delete $ENV{GIT_CONFIG};
2348        }
2349        die $err if $err;
2350        wantarray ? @ret : $ret[0];
2351}
2352
2353sub tmp_index_do {
2354        my ($self, $sub) = @_;
2355        my $old_index = $ENV{GIT_INDEX_FILE};
2356        $ENV{GIT_INDEX_FILE} = $self->{index};
2357        $@ = undef;
2358        my @ret = eval {
2359                my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
2360                mkpath([$dir]) unless -d $dir;
2361                &$sub;
2362        };
2363        my $err = $@;
2364        if (defined $old_index) {
2365                $ENV{GIT_INDEX_FILE} = $old_index;
2366        } else {
2367                delete $ENV{GIT_INDEX_FILE};
2368        }
2369        die $err if $err;
2370        wantarray ? @ret : $ret[0];
2371}
2372
2373sub assert_index_clean {
2374        my ($self, $treeish) = @_;
2375
2376        $self->tmp_index_do(sub {
2377                command_noisy('read-tree', $treeish) unless -e $self->{index};
2378                my $x = command_oneline('write-tree');
2379                my ($y) = (command(qw/cat-file commit/, $treeish) =~
2380                           /^tree ($::sha1)/mo);
2381                return if $y eq $x;
2382
2383                warn "Index mismatch: $y != $x\nrereading $treeish\n";
2384                unlink $self->{index} or die "unlink $self->{index}: $!\n";
2385                command_noisy('read-tree', $treeish);
2386                $x = command_oneline('write-tree');
2387                if ($y ne $x) {
2388                        ::fatal "trees ($treeish) $y != $x\n",
2389                                "Something is seriously wrong...";
2390                }
2391        });
2392}
2393
2394sub get_commit_parents {
2395        my ($self, $log_entry) = @_;
2396        my (%seen, @ret, @tmp);
2397        # legacy support for 'set-tree'; this is only used by set_tree_cb:
2398        if (my $ip = $self->{inject_parents}) {
2399                if (my $commit = delete $ip->{$log_entry->{revision}}) {
2400                        push @tmp, $commit;
2401                }
2402        }
2403        if (my $cur = ::verify_ref($self->refname.'^0')) {
2404                push @tmp, $cur;
2405        }
2406        if (my $ipd = $self->{inject_parents_dcommit}) {
2407                if (my $commit = delete $ipd->{$log_entry->{revision}}) {
2408                        push @tmp, @$commit;
2409                }
2410        }
2411        push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
2412        while (my $p = shift @tmp) {
2413                next if $seen{$p};
2414                $seen{$p} = 1;
2415                push @ret, $p;
2416                # MAXPARENT is defined to 16 in commit-tree.c:
2417                last if @ret >= 16;
2418        }
2419        if (@tmp) {
2420                die "r$log_entry->{revision}: No room for parents:\n\t",
2421                    join("\n\t", @tmp), "\n";
2422        }
2423        @ret;
2424}
2425
2426sub rewrite_root {
2427        my ($self) = @_;
2428        return $self->{-rewrite_root} if exists $self->{-rewrite_root};
2429        my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
2430        my $rwr = eval { command_oneline(qw/config --get/, $k) };
2431        if ($rwr) {
2432                $rwr =~ s#/+$##;
2433                if ($rwr !~ m#^[a-z\+]+://#) {
2434                        die "$rwr is not a valid URL (key: $k)\n";
2435                }
2436        }
2437        $self->{-rewrite_root} = $rwr;
2438}
2439
2440sub metadata_url {
2441        my ($self) = @_;
2442        ($self->rewrite_root || $self->{url}) .
2443           (length $self->{path} ? '/' . $self->{path} : '');
2444}
2445
2446sub full_url {
2447        my ($self) = @_;
2448        $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
2449}
2450
2451
2452sub set_commit_header_env {
2453        my ($log_entry) = @_;
2454        my %env;
2455        foreach my $ned (qw/NAME EMAIL DATE/) {
2456                foreach my $ac (qw/AUTHOR COMMITTER/) {
2457                        $env{"GIT_${ac}_${ned}"} = $ENV{"GIT_${ac}_${ned}"};
2458                }
2459        }
2460
2461        $ENV{GIT_AUTHOR_NAME} = $log_entry->{name};
2462        $ENV{GIT_AUTHOR_EMAIL} = $log_entry->{email};
2463        $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
2464
2465        $ENV{GIT_COMMITTER_NAME} = (defined $log_entry->{commit_name})
2466                                                ? $log_entry->{commit_name}
2467                                                : $log_entry->{name};
2468        $ENV{GIT_COMMITTER_EMAIL} = (defined $log_entry->{commit_email})
2469                                                ? $log_entry->{commit_email}
2470                                                : $log_entry->{email};
2471        \%env;
2472}
2473
2474sub restore_commit_header_env {
2475        my ($env) = @_;
2476        foreach my $ned (qw/NAME EMAIL DATE/) {
2477                foreach my $ac (qw/AUTHOR COMMITTER/) {
2478                        my $k = "GIT_${ac}_${ned}";
2479                        if (defined $env->{$k}) {
2480                                $ENV{$k} = $env->{$k};
2481                        } else {
2482                                delete $ENV{$k};
2483                        }
2484                }
2485        }
2486}
2487
2488sub gc {
2489        command_noisy('gc', '--auto');
2490};
2491
2492sub do_git_commit {
2493        my ($self, $log_entry) = @_;
2494        my $lr = $self->last_rev;
2495        if (defined $lr && $lr >= $log_entry->{revision}) {
2496                die "Last fetched revision of ", $self->refname,
2497                    " was r$lr, but we are about to fetch: ",
2498                    "r$log_entry->{revision}!\n";
2499        }
2500        if (my $c = $self->rev_map_get($log_entry->{revision})) {
2501                croak "$log_entry->{revision} = $c already exists! ",
2502                      "Why are we refetching it?\n";
2503        }
2504        my $old_env = set_commit_header_env($log_entry);
2505        my $tree = $log_entry->{tree};
2506        if (!defined $tree) {
2507                $tree = $self->tmp_index_do(sub {
2508                                            command_oneline('write-tree') });
2509        }
2510        die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
2511
2512        my @exec = ('git', 'commit-tree', $tree);
2513        foreach ($self->get_commit_parents($log_entry)) {
2514                push @exec, '-p', $_;
2515        }
2516        defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
2517                                                                   or croak $!;
2518        binmode $msg_fh;
2519
2520        # we always get UTF-8 from SVN, but we may want our commits in
2521        # a different encoding.
2522        if (my $enc = Git::config('i18n.commitencoding')) {
2523                require Encode;
2524                Encode::from_to($log_entry->{log}, 'UTF-8', $enc);
2525        }
2526        print $msg_fh $log_entry->{log} or croak $!;
2527        restore_commit_header_env($old_env);
2528        unless ($self->no_metadata) {
2529                print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
2530                              or croak $!;
2531        }
2532        $msg_fh->flush == 0 or croak $!;
2533        close $msg_fh or croak $!;
2534        chomp(my $commit = do { local $/; <$out_fh> });
2535        close $out_fh or croak $!;
2536        waitpid $pid, 0;
2537        croak $? if $?;
2538        if ($commit !~ /^$::sha1$/o) {
2539                die "Failed to commit, invalid sha1: $commit\n";
2540        }
2541
2542        $self->rev_map_set($log_entry->{revision}, $commit, 1);
2543
2544        $self->{last_rev} = $log_entry->{revision};
2545        $self->{last_commit} = $commit;
2546        print "r$log_entry->{revision}" unless $::_q > 1;
2547        if (defined $log_entry->{svm_revision}) {
2548                 print " (\@$log_entry->{svm_revision})" unless $::_q > 1;
2549                 $self->rev_map_set($log_entry->{svm_revision}, $commit,
2550                                   0, $self->svm_uuid);
2551        }
2552        print " = $commit ($self->{ref_id})\n" unless $::_q > 1;
2553        if (--$_gc_nr == 0) {
2554                $_gc_nr = $_gc_period;
2555                gc();
2556        }
2557        return $commit;
2558}
2559
2560sub match_paths {
2561        my ($self, $paths, $r) = @_;
2562        return 1 if $self->{path} eq '';
2563        if (my $path = $paths->{"/$self->{path}"}) {
2564                return ($path->{action} eq 'D') ? 0 : 1;
2565        }
2566        $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
2567        if (grep /$self->{path_regex}/, keys %$paths) {
2568                return 1;
2569        }
2570        my $c = '';
2571        foreach (split m#/#, $self->{path}) {
2572                $c .= "/$_";
2573                next unless ($paths->{$c} &&
2574                             ($paths->{$c}->{action} =~ /^[AR]$/));
2575                if ($self->ra->check_path($self->{path}, $r) ==
2576                    $SVN::Node::dir) {
2577                        return 1;
2578                }
2579        }
2580        return 0;
2581}
2582
2583sub find_parent_branch {
2584        my ($self, $paths, $rev) = @_;
2585        return undef unless $self->follow_parent;
2586        unless (defined $paths) {
2587                my $err_handler = $SVN::Error::handler;
2588                $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
2589                $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1,
2590                                   sub { $paths = $_[0] });
2591                $SVN::Error::handler = $err_handler;
2592        }
2593        return undef unless defined $paths;
2594
2595        # look for a parent from another branch:
2596        my @b_path_components = split m#/#, $self->{path};
2597        my @a_path_components;
2598        my $i;
2599        while (@b_path_components) {
2600                $i = $paths->{'/'.join('/', @b_path_components)};
2601                last if $i && defined $i->{copyfrom_path};
2602                unshift(@a_path_components, pop(@b_path_components));
2603        }
2604        return undef unless defined $i && defined $i->{copyfrom_path};
2605        my $branch_from = $i->{copyfrom_path};
2606        if (@a_path_components) {
2607                print STDERR "branch_from: $branch_from => ";
2608                $branch_from .= '/'.join('/', @a_path_components);
2609                print STDERR $branch_from, "\n";
2610        }
2611        my $r = $i->{copyfrom_rev};
2612        my $repos_root = $self->ra->{repos_root};
2613        my $url = $self->ra->{url};
2614        my $new_url = $url . $branch_from;
2615        print STDERR  "Found possible branch point: ",
2616                      "$new_url => ", $self->full_url, ", $r\n";
2617        $branch_from =~ s#^/##;
2618        my $gs = $self->other_gs($new_url, $url,
2619                                 $branch_from, $r, $self->{ref_id});
2620        my ($r0, $parent) = $gs->find_rev_before($r, 1);
2621        {
2622                my ($base, $head);
2623                if (!defined $r0 || !defined $parent) {
2624                        ($base, $head) = parse_revision_argument(0, $r);
2625                } else {
2626                        if ($r0 < $r) {
2627                                $gs->ra->get_log([$gs->{path}], $r0 + 1, $r, 1,
2628                                        0, 1, sub { $base = $_[1] - 1 });
2629                        }
2630                }
2631                if (defined $base && $base <= $r) {
2632                        $gs->fetch($base, $r);
2633                }
2634                ($r0, $parent) = $gs->find_rev_before($r, 1);
2635        }
2636        if (defined $r0 && defined $parent) {
2637                print STDERR "Found branch parent: ($self->{ref_id}) $parent\n";
2638                my $ed;
2639                if ($self->ra->can_do_switch) {
2640                        $self->assert_index_clean($parent);
2641                        print STDERR "Following parent with do_switch\n";
2642                        # do_switch works with svn/trunk >= r22312, but that
2643                        # is not included with SVN 1.4.3 (the latest version
2644                        # at the moment), so we can't rely on it
2645                        $self->{last_rev} = $r0;
2646                        $self->{last_commit} = $parent;
2647                        $ed = SVN::Git::Fetcher->new($self, $gs->{path});
2648                        $gs->ra->gs_do_switch($r0, $rev, $gs,
2649                                              $self->full_url, $ed)
2650                          or die "SVN connection failed somewhere...\n";
2651                } elsif ($self->ra->trees_match($new_url, $r0,
2652                                                $self->full_url, $rev)) {
2653                        print STDERR "Trees match:\n",
2654                                     "  $new_url\@$r0\n",
2655                                     "  ${\$self->full_url}\@$rev\n",
2656                                     "Following parent with no changes\n";
2657                        $self->tmp_index_do(sub {
2658                            command_noisy('read-tree', $parent);
2659                        });
2660                        $self->{last_commit} = $parent;
2661                } else {
2662                        print STDERR "Following parent with do_update\n";
2663                        $ed = SVN::Git::Fetcher->new($self);
2664                        $self->ra->gs_do_update($rev, $rev, $self, $ed)
2665                          or die "SVN connection failed somewhere...\n";
2666                }
2667                print STDERR "Successfully followed parent\n";
2668                return $self->make_log_entry($rev, [$parent], $ed);
2669        }
2670        return undef;
2671}
2672
2673sub do_fetch {
2674        my ($self, $paths, $rev) = @_;
2675        my $ed;
2676        my ($last_rev, @parents);
2677        if (my $lc = $self->last_commit) {
2678                # we can have a branch that was deleted, then re-added
2679                # under the same name but copied from another path, in
2680                # which case we'll have multiple parents (we don't
2681                # want to break the original ref, nor lose copypath info):
2682                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2683                        push @{$log_entry->{parents}}, $lc;
2684                        return $log_entry;
2685                }
2686                $ed = SVN::Git::Fetcher->new($self);
2687                $last_rev = $self->{last_rev};
2688                $ed->{c} = $lc;
2689                @parents = ($lc);
2690        } else {
2691                $last_rev = $rev;
2692                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2693                        return $log_entry;
2694                }
2695                $ed = SVN::Git::Fetcher->new($self);
2696        }
2697        unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
2698                die "SVN connection failed somewhere...\n";
2699        }
2700        $self->make_log_entry($rev, \@parents, $ed);
2701}
2702
2703sub get_untracked {
2704        my ($self, $ed) = @_;
2705        my @out;
2706        my $h = $ed->{empty};
2707        foreach (sort keys %$h) {
2708                my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
2709                push @out, "  $act: " . uri_encode($_);
2710                warn "W: $act: $_\n";
2711        }
2712        foreach my $t (qw/dir_prop file_prop/) {
2713                $h = $ed->{$t} or next;
2714                foreach my $path (sort keys %$h) {
2715                        my $ppath = $path eq '' ? '.' : $path;
2716                        foreach my $prop (sort keys %{$h->{$path}}) {
2717                                next if $SKIP_PROP{$prop};
2718                                my $v = $h->{$path}->{$prop};
2719                                my $t_ppath_prop = "$t: " .
2720                                                    uri_encode($ppath) . ' ' .
2721                                                    uri_encode($prop);
2722                                if (defined $v) {
2723                                        push @out, "  +$t_ppath_prop " .
2724                                                   uri_encode($v);
2725                                } else {
2726                                        push @out, "  -$t_ppath_prop";
2727                                }
2728                        }
2729                }
2730        }
2731        foreach my $t (qw/absent_file absent_directory/) {
2732                $h = $ed->{$t} or next;
2733                foreach my $parent (sort keys %$h) {
2734                        foreach my $path (sort @{$h->{$parent}}) {
2735                                push @out, "  $t: " .
2736                                           uri_encode("$parent/$path");
2737                                warn "W: $t: $parent/$path ",
2738                                     "Insufficient permissions?\n";
2739                        }
2740                }
2741        }
2742        \@out;
2743}
2744
2745# parse_svn_date(DATE)
2746# --------------------
2747# Given a date (in UTC) from Subversion, return a string in the format
2748# "<TZ Offset> <local date/time>" that Git will use.
2749#
2750# By default the parsed date will be in UTC; if $Git::SVN::_localtime
2751# is true we'll convert it to the local timezone instead.
2752sub parse_svn_date {
2753        my $date = shift || return '+0000 1970-01-01 00:00:00';
2754        my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2755                                            (\d\d)\:(\d\d)\:(\d\d)\.\d*Z$/x) or
2756                                         croak "Unable to parse date: $date\n";
2757        my $parsed_date;    # Set next.
2758
2759        if ($Git::SVN::_localtime) {
2760                # Translate the Subversion datetime to an epoch time.
2761                # Begin by switching ourselves to $date's timezone, UTC.
2762                my $old_env_TZ = $ENV{TZ};
2763                $ENV{TZ} = 'UTC';
2764
2765                my $epoch_in_UTC =
2766                    POSIX::strftime('%s', $S, $M, $H, $d, $m - 1, $Y - 1900);
2767
2768                # Determine our local timezone (including DST) at the
2769                # time of $epoch_in_UTC.  $Git::SVN::Log::TZ stored the
2770                # value of TZ, if any, at the time we were run.
2771                if (defined $Git::SVN::Log::TZ) {
2772                        $ENV{TZ} = $Git::SVN::Log::TZ;
2773                } else {
2774                        delete $ENV{TZ};
2775                }
2776
2777                my $our_TZ =
2778                    POSIX::strftime('%Z', $S, $M, $H, $d, $m - 1, $Y - 1900);
2779
2780                # This converts $epoch_in_UTC into our local timezone.
2781                my ($sec, $min, $hour, $mday, $mon, $year,
2782                    $wday, $yday, $isdst) = localtime($epoch_in_UTC);
2783
2784                $parsed_date = sprintf('%s %04d-%02d-%02d %02d:%02d:%02d',
2785                                       $our_TZ, $year + 1900, $mon + 1,
2786                                       $mday, $hour, $min, $sec);
2787
2788                # Reset us to the timezone in effect when we entered
2789                # this routine.
2790                if (defined $old_env_TZ) {
2791                        $ENV{TZ} = $old_env_TZ;
2792                } else {
2793                        delete $ENV{TZ};
2794                }
2795        } else {
2796                $parsed_date = "+0000 $Y-$m-$d $H:$M:$S";
2797        }
2798
2799        return $parsed_date;
2800}
2801
2802sub other_gs {
2803        my ($self, $new_url, $url,
2804            $branch_from, $r, $old_ref_id) = @_;
2805        my $gs = Git::SVN->find_by_url($new_url, $url, $branch_from);
2806        unless ($gs) {
2807                my $ref_id = $old_ref_id;
2808                $ref_id =~ s/\@\d+$//;
2809                $ref_id .= "\@$r";
2810                # just grow a tail if we're not unique enough :x
2811                $ref_id .= '-' while find_ref($ref_id);
2812                print STDERR "Initializing parent: $ref_id\n";
2813                my ($u, $p, $repo_id) = ($new_url, '', $ref_id);
2814                if ($u =~ s#^\Q$url\E(/|$)##) {
2815                        $p = $u;
2816                        $u = $url;
2817                        $repo_id = $self->{repo_id};
2818                }
2819                $gs = Git::SVN->init($u, $p, $repo_id, $ref_id, 1);
2820        }
2821        $gs
2822}
2823
2824sub call_authors_prog {
2825        my ($orig_author) = @_;
2826        my $author = `$::_authors_prog $orig_author`;
2827        if ($? != 0) {
2828                die "$::_authors_prog failed with exit code $?\n"
2829        }
2830        if ($author =~ /^\s*(.+?)\s*<(.*)>\s*$/) {
2831                my ($name, $email) = ($1, $2);
2832                $email = undef if length $2 == 0;
2833                return [$name, $email];
2834        } else {
2835                die "Author: $orig_author: $::_authors_prog returned "
2836                        . "invalid author format: $author\n";
2837        }
2838}
2839
2840sub check_author {
2841        my ($author) = @_;
2842        if (!defined $author || length $author == 0) {
2843                $author = '(no author)';
2844        }
2845        if (!defined $::users{$author}) {
2846                if (defined $::_authors_prog) {
2847                        $::users{$author} = call_authors_prog($author);
2848                } elsif (defined $::_authors) {
2849                        die "Author: $author not defined in $::_authors file\n";
2850                }
2851        }
2852        $author;
2853}
2854
2855sub make_log_entry {
2856        my ($self, $rev, $parents, $ed) = @_;
2857        my $untracked = $self->get_untracked($ed);
2858
2859        open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
2860        print $un "r$rev\n" or croak $!;
2861        print $un $_, "\n" foreach @$untracked;
2862        my %log_entry = ( parents => $parents || [], revision => $rev,
2863                          log => '');
2864
2865        my $headrev;
2866        my $logged = delete $self->{logged_rev_props};
2867        if (!$logged || $self->{-want_revprops}) {
2868                my $rp = $self->ra->rev_proplist($rev);
2869                foreach (sort keys %$rp) {
2870                        my $v = $rp->{$_};
2871                        if (/^svn:(author|date|log)$/) {
2872                                $log_entry{$1} = $v;
2873                        } elsif ($_ eq 'svm:headrev') {
2874                                $headrev = $v;
2875                        } else {
2876                                print $un "  rev_prop: ", uri_encode($_), ' ',
2877                                          uri_encode($v), "\n";
2878                        }
2879                }
2880        } else {
2881                map { $log_entry{$_} = $logged->{$_} } keys %$logged;
2882        }
2883        close $un or croak $!;
2884
2885        $log_entry{date} = parse_svn_date($log_entry{date});
2886        $log_entry{log} .= "\n";
2887        my $author = $log_entry{author} = check_author($log_entry{author});
2888        my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
2889                                                       : ($author, undef);
2890
2891        my ($commit_name, $commit_email) = ($name, $email);
2892        if ($_use_log_author) {
2893                my $name_field;
2894                if ($log_entry{log} =~ /From:\s+(.*\S)\s*\n/i) {
2895                        $name_field = $1;
2896                } elsif ($log_entry{log} =~ /Signed-off-by:\s+(.*\S)\s*\n/i) {
2897                        $name_field = $1;
2898                }
2899                if (!defined $name_field) {
2900                        if (!defined $email) {
2901                                $email = $name;
2902                        }
2903                } elsif ($name_field =~ /(.*?)\s+<(.*)>/) {
2904                        ($name, $email) = ($1, $2);
2905                } elsif ($name_field =~ /(.*)@/) {
2906                        ($name, $email) = ($1, $name_field);
2907                } else {
2908                        ($name, $email) = ($name_field, $name_field);
2909                }
2910        }
2911        if (defined $headrev && $self->use_svm_props) {
2912                if ($self->rewrite_root) {
2913                        die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
2914                            "options set!\n";
2915                }
2916                my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$}i;
2917                # we don't want "SVM: initializing mirror for junk" ...
2918                return undef if $r == 0;
2919                my $svm = $self->svm;
2920                if ($uuid ne $svm->{uuid}) {
2921                        die "UUID mismatch on SVM path:\n",
2922                            "expected: $svm->{uuid}\n",
2923                            "     got: $uuid\n";
2924                }
2925                my $full_url = $self->full_url;
2926                $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
2927                             die "Failed to replace '$svm->{replace}' with ",
2928                                 "'$svm->{source}' in $full_url\n";
2929                # throw away username for storing in records
2930                remove_username($full_url);
2931                $log_entry{metadata} = "$full_url\@$r $uuid";
2932                $log_entry{svm_revision} = $r;
2933                $email ||= "$author\@$uuid";
2934                $commit_email ||= "$author\@$uuid";
2935        } elsif ($self->use_svnsync_props) {
2936                my $full_url = $self->svnsync->{url};
2937                $full_url .= "/$self->{path}" if length $self->{path};
2938                remove_username($full_url);
2939                my $uuid = $self->svnsync->{uuid};
2940                $log_entry{metadata} = "$full_url\@$rev $uuid";
2941                $email ||= "$author\@$uuid";
2942                $commit_email ||= "$author\@$uuid";
2943        } else {
2944                my $url = $self->metadata_url;
2945                remove_username($url);
2946                $log_entry{metadata} = "$url\@$rev " .
2947                                       $self->ra->get_uuid;
2948                $email ||= "$author\@" . $self->ra->get_uuid;
2949                $commit_email ||= "$author\@" . $self->ra->get_uuid;
2950        }
2951        $log_entry{name} = $name;
2952        $log_entry{email} = $email;
2953        $log_entry{commit_name} = $commit_name;
2954        $log_entry{commit_email} = $commit_email;
2955        \%log_entry;
2956}
2957
2958sub fetch {
2959        my ($self, $min_rev, $max_rev, @parents) = @_;
2960        my ($last_rev, $last_commit) = $self->last_rev_commit;
2961        my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
2962        $self->ra->gs_fetch_loop_common($base, $head, [$self]);
2963}
2964
2965sub set_tree_cb {
2966        my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
2967        $self->{inject_parents} = { $rev => $tree };
2968        $self->fetch(undef, undef);
2969}
2970
2971sub set_tree {
2972        my ($self, $tree) = (shift, shift);
2973        my $log_entry = ::get_commit_entry($tree);
2974        unless ($self->{last_rev}) {
2975                ::fatal("Must have an existing revision to commit");
2976        }
2977        my %ed_opts = ( r => $self->{last_rev},
2978                        log => $log_entry->{log},
2979                        ra => $self->ra,
2980                        tree_a => $self->{last_commit},
2981                        tree_b => $tree,
2982                        editor_cb => sub {
2983                               $self->set_tree_cb($log_entry, $tree, @_) },
2984                        svn_path => $self->{path} );
2985        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
2986                print "No changes\nr$self->{last_rev} = $tree\n";
2987        }
2988}
2989
2990sub rebuild_from_rev_db {
2991        my ($self, $path) = @_;
2992        my $r = -1;
2993        open my $fh, '<', $path or croak "open: $!";
2994        binmode $fh or croak "binmode: $!";
2995        while (<$fh>) {
2996                length($_) == 41 or croak "inconsistent size in ($_) != 41";
2997                chomp($_);
2998                ++$r;
2999                next if $_ eq ('0' x 40);
3000                $self->rev_map_set($r, $_);
3001                print "r$r = $_\n";
3002        }
3003        close $fh or croak "close: $!";
3004        unlink $path or croak "unlink: $!";
3005}
3006
3007sub rebuild {
3008        my ($self) = @_;
3009        my $map_path = $self->map_path;
3010        my $partial = (-e $map_path && ! -z $map_path);
3011        return unless ::verify_ref($self->refname.'^0');
3012        if (!$partial && ($self->use_svm_props || $self->no_metadata)) {
3013                my $rev_db = $self->rev_db_path;
3014                $self->rebuild_from_rev_db($rev_db);
3015                if ($self->use_svm_props) {
3016                        my $svm_rev_db = $self->rev_db_path($self->svm_uuid);
3017                        $self->rebuild_from_rev_db($svm_rev_db);
3018                }
3019                $self->unlink_rev_db_symlink;
3020                return;
3021        }
3022        print "Rebuilding $map_path ...\n" if (!$partial);
3023        my ($base_rev, $head) = ($partial ? $self->rev_map_max_norebuild(1) :
3024                (undef, undef));
3025        my ($log, $ctx) =
3026            command_output_pipe(qw/rev-list --pretty=raw --no-color --reverse/,
3027                                ($head ? "$head.." : "") . $self->refname,
3028                                '--');
3029        my $metadata_url = $self->metadata_url;
3030        remove_username($metadata_url);
3031        my $svn_uuid = $self->ra_uuid;
3032        my $c;
3033        while (<$log>) {
3034                if ( m{^commit ($::sha1)$} ) {
3035                        $c = $1;
3036                        next;
3037                }
3038                next unless s{^\s*(git-svn-id:)}{$1};
3039                my ($url, $rev, $uuid) = ::extract_metadata($_);
3040                remove_username($url);
3041
3042                # ignore merges (from set-tree)
3043                next if (!defined $rev || !$uuid);
3044
3045                # if we merged or otherwise started elsewhere, this is
3046                # how we break out of it
3047                if (($uuid ne $svn_uuid) ||
3048                    ($metadata_url && $url && ($url ne $metadata_url))) {
3049                        next;
3050                }
3051                if ($partial && $head) {
3052                        print "Partial-rebuilding $map_path ...\n";
3053                        print "Currently at $base_rev = $head\n";
3054                        $head = undef;
3055                }
3056
3057                $self->rev_map_set($rev, $c);
3058                print "r$rev = $c\n";
3059        }
3060        command_close_pipe($log, $ctx);
3061        print "Done rebuilding $map_path\n" if (!$partial || !$head);
3062        my $rev_db_path = $self->rev_db_path;
3063        if (-f $self->rev_db_path) {
3064                unlink $self->rev_db_path or croak "unlink: $!";
3065        }
3066        $self->unlink_rev_db_symlink;
3067}
3068
3069# rev_map:
3070# Tie::File seems to be prone to offset errors if revisions get sparse,
3071# it's not that fast, either.  Tie::File is also not in Perl 5.6.  So
3072# one of my favorite modules is out :<  Next up would be one of the DBM
3073# modules, but I'm not sure which is most portable...
3074#
3075# This is the replacement for the rev_db format, which was too big
3076# and inefficient for large repositories with a lot of sparse history
3077# (mainly tags)
3078#
3079# The format is this:
3080#   - 24 bytes for every record,
3081#     * 4 bytes for the integer representing an SVN revision number
3082#     * 20 bytes representing the sha1 of a git commit
3083#   - No empty padding records like the old format
3084#     (except the last record, which can be overwritten)
3085#   - new records are written append-only since SVN revision numbers
3086#     increase monotonically
3087#   - lookups on SVN revision number are done via a binary search
3088#   - Piping the file to xxd -c24 is a good way of dumping it for
3089#     viewing or editing (piped back through xxd -r), should the need
3090#     ever arise.
3091#   - The last record can be padding revision with an all-zero sha1
3092#     This is used to optimize fetch performance when using multiple
3093#     "fetch" directives in .git/config
3094#
3095# These files are disposable unless noMetadata or useSvmProps is set
3096
3097sub _rev_map_set {
3098        my ($fh, $rev, $commit) = @_;
3099
3100        binmode $fh or croak "binmode: $!";
3101        my $size = (stat($fh))[7];
3102        ($size % 24) == 0 or croak "inconsistent size: $size";
3103
3104        my $wr_offset = 0;
3105        if ($size > 0) {
3106                sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3107                my $read = sysread($fh, my $buf, 24) or croak "read: $!";
3108                $read == 24 or croak "read only $read bytes (!= 24)";
3109                my ($last_rev, $last_commit) = unpack(rev_map_fmt, $buf);
3110                if ($last_commit eq ('0' x40)) {
3111                        if ($size >= 48) {
3112                                sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3113                                $read = sysread($fh, $buf, 24) or
3114                                    croak "read: $!";
3115                                $read == 24 or
3116                                    croak "read only $read bytes (!= 24)";
3117                                ($last_rev, $last_commit) =
3118                                    unpack(rev_map_fmt, $buf);
3119                                if ($last_commit eq ('0' x40)) {
3120                                        croak "inconsistent .rev_map\n";
3121                                }
3122                        }
3123                        if ($last_rev >= $rev) {
3124                                croak "last_rev is higher!: $last_rev >= $rev";
3125                        }
3126                        $wr_offset = -24;
3127                }
3128        }
3129        sysseek($fh, $wr_offset, SEEK_END) or croak "seek: $!";
3130        syswrite($fh, pack(rev_map_fmt, $rev, $commit), 24) == 24 or
3131          croak "write: $!";
3132}
3133
3134sub _rev_map_reset {
3135        my ($fh, $rev, $commit) = @_;
3136        my $c = _rev_map_get($fh, $rev);
3137        $c eq $commit or die "_rev_map_reset(@_) commit $c does not match!\n";
3138        my $offset = sysseek($fh, 0, SEEK_CUR) or croak "seek: $!";
3139        truncate $fh, $offset or croak "truncate: $!";
3140}
3141
3142sub mkfile {
3143        my ($path) = @_;
3144        unless (-e $path) {
3145                my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
3146                mkpath([$dir]) unless -d $dir;
3147                open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
3148                close $fh or die "Couldn't close (create) $path: $!\n";
3149        }
3150}
3151
3152sub rev_map_set {
3153        my ($self, $rev, $commit, $update_ref, $uuid) = @_;
3154        length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
3155        my $db = $self->map_path($uuid);
3156        my $db_lock = "$db.lock";
3157        my $sig;
3158        $update_ref ||= 0;
3159        if ($update_ref) {
3160                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3161                            $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
3162        }
3163        mkfile($db);
3164
3165        $LOCKFILES{$db_lock} = 1;
3166        my $sync;
3167        # both of these options make our .rev_db file very, very important
3168        # and we can't afford to lose it because rebuild() won't work
3169        if ($self->use_svm_props || $self->no_metadata) {
3170                $sync = 1;
3171                copy($db, $db_lock) or die "rev_map_set(@_): ",
3172                                           "Failed to copy: ",
3173                                           "$db => $db_lock ($!)\n";
3174        } else {
3175                rename $db, $db_lock or die "rev_map_set(@_): ",
3176                                            "Failed to rename: ",
3177                                            "$db => $db_lock ($!)\n";
3178        }
3179
3180        sysopen(my $fh, $db_lock, O_RDWR | O_CREAT)
3181             or croak "Couldn't open $db_lock: $!\n";
3182        $update_ref eq 'reset' ? _rev_map_reset($fh, $rev, $commit) :
3183                                 _rev_map_set($fh, $rev, $commit);
3184        if ($sync) {
3185                $fh->flush or die "Couldn't flush $db_lock: $!\n";
3186                $fh->sync or die "Couldn't sync $db_lock: $!\n";
3187        }
3188        close $fh or croak $!;
3189        if ($update_ref) {
3190                $_head = $self;
3191                my $note = "";
3192                $note = " ($update_ref)" if ($update_ref !~ /^\d*$/);
3193                command_noisy('update-ref', '-m', "r$rev$note",
3194                              $self->refname, $commit);
3195        }
3196        rename $db_lock, $db or die "rev_map_set(@_): ", "Failed to rename: ",
3197                                    "$db_lock => $db ($!)\n";
3198        delete $LOCKFILES{$db_lock};
3199        if ($update_ref) {
3200                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3201                            $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
3202                kill $sig, $$ if defined $sig;
3203        }
3204}
3205
3206# If want_commit, this will return an array of (rev, commit) where
3207# commit _must_ be a valid commit in the archive.
3208# Otherwise, it'll return the max revision (whether or not the
3209# commit is valid or just a 0x40 placeholder).
3210sub rev_map_max {
3211        my ($self, $want_commit) = @_;
3212        $self->rebuild;
3213        my ($r, $c) = $self->rev_map_max_norebuild($want_commit);
3214        $want_commit ? ($r, $c) : $r;
3215}
3216
3217sub rev_map_max_norebuild {
3218        my ($self, $want_commit) = @_;
3219        my $map_path = $self->map_path;
3220        stat $map_path or return $want_commit ? (0, undef) : 0;
3221        sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3222        binmode $fh or croak "binmode: $!";
3223        my $size = (stat($fh))[7];
3224        ($size % 24) == 0 or croak "inconsistent size: $size";
3225
3226        if ($size == 0) {
3227                close $fh or croak "close: $!";
3228                return $want_commit ? (0, undef) : 0;
3229        }
3230
3231        sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3232        sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3233        my ($r, $c) = unpack(rev_map_fmt, $buf);
3234        if ($want_commit && $c eq ('0' x40)) {
3235                if ($size < 48) {
3236                        return $want_commit ? (0, undef) : 0;
3237                }
3238                sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3239                sysread($fh, $buf, 24) == 24 or croak "read: $!";
3240                ($r, $c) = unpack(rev_map_fmt, $buf);
3241                if ($c eq ('0'x40)) {
3242                        croak "Penultimate record is all-zeroes in $map_path";
3243                }
3244        }
3245        close $fh or croak "close: $!";
3246        $want_commit ? ($r, $c) : $r;
3247}
3248
3249sub rev_map_get {
3250        my ($self, $rev, $uuid) = @_;
3251        my $map_path = $self->map_path($uuid);
3252        return undef unless -e $map_path;
3253
3254        sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3255        my $c = _rev_map_get($fh, $rev);
3256        close($fh) or croak "close: $!";
3257        $c
3258}
3259
3260sub _rev_map_get {
3261        my ($fh, $rev) = @_;
3262
3263        binmode $fh or croak "binmode: $!";
3264        my $size = (stat($fh))[7];
3265        ($size % 24) == 0 or croak "inconsistent size: $size";
3266
3267        if ($size == 0) {
3268                return undef;
3269        }
3270
3271        my ($l, $u) = (0, $size - 24);
3272        my ($r, $c, $buf);
3273
3274        while ($l <= $u) {
3275                my $i = int(($l/24 + $u/24) / 2) * 24;
3276                sysseek($fh, $i, SEEK_SET) or croak "seek: $!";
3277                sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3278                my ($r, $c) = unpack('NH40', $buf);
3279
3280                if ($r < $rev) {
3281                        $l = $i + 24;
3282                } elsif ($r > $rev) {
3283                        $u = $i - 24;
3284                } else { # $r == $rev
3285                        return $c eq ('0' x 40) ? undef : $c;
3286                }
3287        }
3288        undef;
3289}
3290
3291# Finds the first svn revision that exists on (if $eq_ok is true) or
3292# before $rev for the current branch.  It will not search any lower
3293# than $min_rev.  Returns the git commit hash and svn revision number
3294# if found, else (undef, undef).
3295sub find_rev_before {
3296        my ($self, $rev, $eq_ok, $min_rev) = @_;
3297        --$rev unless $eq_ok;
3298        $min_rev ||= 1;
3299        my $max_rev = $self->rev_map_max;
3300        $rev = $max_rev if ($rev > $max_rev);
3301        while ($rev >= $min_rev) {
3302                if (my $c = $self->rev_map_get($rev)) {
3303                        return ($rev, $c);
3304                }
3305                --$rev;
3306        }
3307        return (undef, undef);
3308}
3309
3310# Finds the first svn revision that exists on (if $eq_ok is true) or
3311# after $rev for the current branch.  It will not search any higher
3312# than $max_rev.  Returns the git commit hash and svn revision number
3313# if found, else (undef, undef).
3314sub find_rev_after {
3315        my ($self, $rev, $eq_ok, $max_rev) = @_;
3316        ++$rev unless $eq_ok;
3317        $max_rev ||= $self->rev_map_max;
3318        while ($rev <= $max_rev) {
3319                if (my $c = $self->rev_map_get($rev)) {
3320                        return ($rev, $c);
3321                }
3322                ++$rev;
3323        }
3324        return (undef, undef);
3325}
3326
3327sub _new {
3328        my ($class, $repo_id, $ref_id, $path) = @_;
3329        unless (defined $repo_id && length $repo_id) {
3330                $repo_id = $Git::SVN::default_repo_id;
3331        }
3332        unless (defined $ref_id && length $ref_id) {
3333                $_prefix = '' unless defined($_prefix);
3334                $_[2] = $ref_id = $_prefix . $Git::SVN::default_ref_id;
3335        }
3336        $_[1] = $repo_id;
3337        my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
3338        $_[3] = $path = '' unless (defined $path);
3339        mkpath(["$ENV{GIT_DIR}/svn"]);
3340        bless {
3341                ref_id => $ref_id, dir => $dir, index => "$dir/index",
3342                path => $path, config => "$ENV{GIT_DIR}/svn/config",
3343                map_root => "$dir/.rev_map", repo_id => $repo_id }, $class;
3344}
3345
3346# for read-only access of old .rev_db formats
3347sub unlink_rev_db_symlink {
3348        my ($self) = @_;
3349        my $link = $self->rev_db_path;
3350        $link =~ s/\.[\w-]+$// or croak "missing UUID at the end of $link";
3351        if (-l $link) {
3352                unlink $link or croak "unlink: $link failed!";
3353        }
3354}
3355
3356sub rev_db_path {
3357        my ($self, $uuid) = @_;
3358        my $db_path = $self->map_path($uuid);
3359        $db_path =~ s{/\.rev_map\.}{/\.rev_db\.}
3360            or croak "map_path: $db_path does not contain '/.rev_map.' !";
3361        $db_path;
3362}
3363
3364# the new replacement for .rev_db
3365sub map_path {
3366        my ($self, $uuid) = @_;
3367        $uuid ||= $self->ra_uuid;
3368        "$self->{map_root}.$uuid";
3369}
3370
3371sub uri_encode {
3372        my ($f) = @_;
3373        $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
3374        $f
3375}
3376
3377sub remove_username {
3378        $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
3379}
3380
3381package Git::SVN::Prompt;
3382use strict;
3383use warnings;
3384require SVN::Core;
3385use vars qw/$_no_auth_cache $_username/;
3386
3387sub simple {
3388        my ($cred, $realm, $default_username, $may_save, $pool) = @_;
3389        $may_save = undef if $_no_auth_cache;
3390        $default_username = $_username if defined $_username;
3391        if (defined $default_username && length $default_username) {
3392                if (defined $realm && length $realm) {
3393                        print STDERR "Authentication realm: $realm\n";
3394                        STDERR->flush;
3395                }
3396                $cred->username($default_username);
3397        } else {
3398                username($cred, $realm, $may_save, $pool);
3399        }
3400        $cred->password(_read_password("Password for '" .
3401                                       $cred->username . "': ", $realm));
3402        $cred->may_save($may_save);
3403        $SVN::_Core::SVN_NO_ERROR;
3404}
3405
3406sub ssl_server_trust {
3407        my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
3408        $may_save = undef if $_no_auth_cache;
3409        print STDERR "Error validating server certificate for '$realm':\n";
3410        {
3411                no warnings 'once';
3412                # All variables SVN::Auth::SSL::* are used only once,
3413                # so we're shutting up Perl warnings about this.
3414                if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
3415                        print STDERR " - The certificate is not issued ",
3416                            "by a trusted authority. Use the\n",
3417                            "   fingerprint to validate ",
3418                            "the certificate manually!\n";
3419                }
3420                if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
3421                        print STDERR " - The certificate hostname ",
3422                            "does not match.\n";
3423                }
3424                if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
3425                        print STDERR " - The certificate is not yet valid.\n";
3426                }
3427                if ($failures & $SVN::Auth::SSL::EXPIRED) {
3428                        print STDERR " - The certificate has expired.\n";
3429                }
3430                if ($failures & $SVN::Auth::SSL::OTHER) {
3431                        print STDERR " - The certificate has ",
3432                            "an unknown error.\n";
3433                }
3434        } # no warnings 'once'
3435        printf STDERR
3436                "Certificate information:\n".
3437                " - Hostname: %s\n".
3438                " - Valid: from %s until %s\n".
3439                " - Issuer: %s\n".
3440                " - Fingerprint: %s\n",
3441                map $cert_info->$_, qw(hostname valid_from valid_until
3442                                       issuer_dname fingerprint);
3443        my $choice;
3444prompt:
3445        print STDERR $may_save ?
3446              "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
3447              "(R)eject or accept (t)emporarily? ";
3448        STDERR->flush;
3449        $choice = lc(substr(<STDIN> || 'R', 0, 1));
3450        if ($choice =~ /^t$/i) {
3451                $cred->may_save(undef);
3452        } elsif ($choice =~ /^r$/i) {
3453                return -1;
3454        } elsif ($may_save && $choice =~ /^p$/i) {
3455                $cred->may_save($may_save);
3456        } else {
3457                goto prompt;
3458        }
3459        $cred->accepted_failures($failures);
3460        $SVN::_Core::SVN_NO_ERROR;
3461}
3462
3463sub ssl_client_cert {
3464        my ($cred, $realm, $may_save, $pool) = @_;
3465        $may_save = undef if $_no_auth_cache;
3466        print STDERR "Client certificate filename: ";
3467        STDERR->flush;
3468        chomp(my $filename = <STDIN>);
3469        $cred->cert_file($filename);
3470        $cred->may_save($may_save);
3471        $SVN::_Core::SVN_NO_ERROR;
3472}
3473
3474sub ssl_client_cert_pw {
3475        my ($cred, $realm, $may_save, $pool) = @_;
3476        $may_save = undef if $_no_auth_cache;
3477        $cred->password(_read_password("Password: ", $realm));
3478        $cred->may_save($may_save);
3479        $SVN::_Core::SVN_NO_ERROR;
3480}
3481
3482sub username {
3483        my ($cred, $realm, $may_save, $pool) = @_;
3484        $may_save = undef if $_no_auth_cache;
3485        if (defined $realm && length $realm) {
3486                print STDERR "Authentication realm: $realm\n";
3487        }
3488        my $username;
3489        if (defined $_username) {
3490                $username = $_username;
3491        } else {
3492                print STDERR "Username: ";
3493                STDERR->flush;
3494                chomp($username = <STDIN>);
3495        }
3496        $cred->username($username);
3497        $cred->may_save($may_save);
3498        $SVN::_Core::SVN_NO_ERROR;
3499}
3500
3501sub _read_password {
3502        my ($prompt, $realm) = @_;
3503        print STDERR $prompt;
3504        STDERR->flush;
3505        require Term::ReadKey;
3506        Term::ReadKey::ReadMode('noecho');
3507        my $password = '';
3508        while (defined(my $key = Term::ReadKey::ReadKey(0))) {
3509                last if $key =~ /[\012\015]/; # \n\r
3510                $password .= $key;
3511        }
3512        Term::ReadKey::ReadMode('restore');
3513        print STDERR "\n";
3514        STDERR->flush;
3515        $password;
3516}
3517
3518package SVN::Git::Fetcher;
3519use vars qw/@ISA/;
3520use strict;
3521use warnings;
3522use Carp qw/croak/;
3523use File::Temp qw/tempfile/;
3524use IO::File qw//;
3525use vars qw/$_ignore_regex/;
3526
3527# file baton members: path, mode_a, mode_b, pool, fh, blob, base
3528sub new {
3529        my ($class, $git_svn, $switch_path) = @_;
3530        my $self = SVN::Delta::Editor->new;
3531        bless $self, $class;
3532        if (exists $git_svn->{last_commit}) {
3533                $self->{c} = $git_svn->{last_commit};
3534                $self->{empty_symlinks} =
3535                                  _mark_empty_symlinks($git_svn, $switch_path);
3536        }
3537        $self->{ignore_regex} = eval { command_oneline('config', '--get',
3538                             "svn-remote.$git_svn->{repo_id}.ignore-paths") };
3539        $self->{empty} = {};
3540        $self->{dir_prop} = {};
3541        $self->{file_prop} = {};
3542        $self->{absent_dir} = {};
3543        $self->{absent_file} = {};
3544        $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
3545        $self;
3546}
3547
3548# this uses the Ra object, so it must be called before do_{switch,update},
3549# not inside them (when the Git::SVN::Fetcher object is passed) to
3550# do_{switch,update}
3551sub _mark_empty_symlinks {
3552        my ($git_svn, $switch_path) = @_;
3553        my $bool = Git::config_bool('svn.brokenSymlinkWorkaround');
3554        return {} if (!defined($bool)) || (defined($bool) && ! $bool);
3555
3556        my %ret;
3557        my ($rev, $cmt) = $git_svn->last_rev_commit;
3558        return {} unless ($rev && $cmt);
3559
3560        # allow the warning to be printed for each revision we fetch to
3561        # ensure the user sees it.  The user can also disable the workaround
3562        # on the repository even while git svn is running and the next
3563        # revision fetched will skip this expensive function.
3564        my $printed_warning;
3565        chomp(my $empty_blob = `git hash-object -t blob --stdin < /dev/null`);
3566        my ($ls, $ctx) = command_output_pipe(qw/ls-tree -r -z/, $cmt);
3567        local $/ = "\0";
3568        my $pfx = defined($switch_path) ? $switch_path : $git_svn->{path};
3569        $pfx .= '/' if length($pfx);
3570        while (<$ls>) {
3571                chomp;
3572                s/\A100644 blob $empty_blob\t//o or next;
3573                unless ($printed_warning) {
3574                        print STDERR "Scanning for empty symlinks, ",
3575                                     "this may take a while if you have ",
3576                                     "many empty files\n",
3577                                     "You may disable this with `",
3578                                     "git config svn.brokenSymlinkWorkaround ",
3579                                     "false'.\n",
3580                                     "This may be done in a different ",
3581                                     "terminal without restarting ",
3582                                     "git svn\n";
3583                        $printed_warning = 1;
3584                }
3585                my $path = $_;
3586                my (undef, $props) =
3587                               $git_svn->ra->get_file($pfx.$path, $rev, undef);
3588                if ($props->{'svn:special'}) {
3589                        $ret{$path} = 1;
3590                }
3591        }
3592        command_close_pipe($ls, $ctx);
3593        \%ret;
3594}
3595
3596# returns true if a given path is inside a ".git" directory
3597sub in_dot_git {
3598        $_[0] =~ m{(?:^|/)\.git(?:/|$)};
3599}
3600
3601# return value: 0 -- don't ignore, 1 -- ignore
3602sub is_path_ignored {
3603        my ($self, $path) = @_;
3604        return 1 if in_dot_git($path);
3605        return 1 if defined($self->{ignore_regex}) &&
3606                    $path =~ m!$self->{ignore_regex}!;
3607        return 0 unless defined($_ignore_regex);
3608        return 1 if $path =~ m!$_ignore_regex!o;
3609        return 0;
3610}
3611
3612sub set_path_strip {
3613        my ($self, $path) = @_;
3614        $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
3615}
3616
3617sub open_root {
3618        { path => '' };
3619}
3620
3621sub open_directory {
3622        my ($self, $path, $pb, $rev) = @_;
3623        { path => $path };
3624}
3625
3626sub git_path {
3627        my ($self, $path) = @_;
3628        if ($self->{path_strip}) {
3629                $path =~ s!$self->{path_strip}!! or
3630                  die "Failed to strip path '$path' ($self->{path_strip})\n";
3631        }
3632        $path;
3633}
3634
3635sub delete_entry {
3636        my ($self, $path, $rev, $pb) = @_;
3637        return undef if $self->is_path_ignored($path);
3638
3639        my $gpath = $self->git_path($path);
3640        return undef if ($gpath eq '');
3641
3642        # remove entire directories.
3643        my ($tree) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
3644                         =~ /\A040000 tree ([a-f\d]{40})\t\Q$gpath\E\0/);
3645        if ($tree) {
3646                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3647                                                     -r --name-only -z/,
3648                                                     $tree);
3649                local $/ = "\0";
3650                while (<$ls>) {
3651                        chomp;
3652                        my $rmpath = "$gpath/$_";
3653                        $self->{gii}->remove($rmpath);
3654                        print "\tD\t$rmpath\n" unless $::_q;
3655                }
3656                print "\tD\t$gpath/\n" unless $::_q;
3657                command_close_pipe($ls, $ctx);
3658                $self->{empty}->{$path} = 0
3659        } else {
3660                $self->{gii}->remove($gpath);
3661                print "\tD\t$gpath\n" unless $::_q;
3662        }
3663        undef;
3664}
3665
3666sub open_file {
3667        my ($self, $path, $pb, $rev) = @_;
3668        my ($mode, $blob);
3669
3670        goto out if $self->is_path_ignored($path);
3671
3672        my $gpath = $self->git_path($path);
3673        ($mode, $blob) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
3674                             =~ /\A(\d{6}) blob ([a-f\d]{40})\t\Q$gpath\E\0/);
3675        unless (defined $mode && defined $blob) {
3676                die "$path was not found in commit $self->{c} (r$rev)\n";
3677        }
3678        if ($mode eq '100644' && $self->{empty_symlinks}->{$path}) {
3679                $mode = '120000';
3680        }
3681out:
3682        { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
3683          pool => SVN::Pool->new, action => 'M' };
3684}
3685
3686sub add_file {
3687        my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
3688        my $mode;
3689
3690        if (!$self->is_path_ignored($path)) {
3691                my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3692                delete $self->{empty}->{$dir};
3693                $mode = '100644';
3694        }
3695        { path => $path, mode_a => $mode, mode_b => $mode,
3696          pool => SVN::Pool->new, action => 'A' };
3697}
3698
3699sub add_directory {
3700        my ($self, $path, $cp_path, $cp_rev) = @_;
3701        goto out if $self->is_path_ignored($path);
3702        my $gpath = $self->git_path($path);
3703        if ($gpath eq '') {
3704                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3705                                                     -r --name-only -z/,
3706                                                     $self->{c});
3707                local $/ = "\0";
3708                while (<$ls>) {
3709                        chomp;
3710                        $self->{gii}->remove($_);
3711                        print "\tD\t$_\n" unless $::_q;
3712                }
3713                command_close_pipe($ls, $ctx);
3714                $self->{empty}->{$path} = 0;
3715        }
3716        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3717        delete $self->{empty}->{$dir};
3718        $self->{empty}->{$path} = 1;
3719out:
3720        { path => $path };
3721}
3722
3723sub change_dir_prop {
3724        my ($self, $db, $prop, $value) = @_;
3725        return undef if $self->is_path_ignored($db->{path});
3726        $self->{dir_prop}->{$db->{path}} ||= {};
3727        $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
3728        undef;
3729}
3730
3731sub absent_directory {
3732        my ($self, $path, $pb) = @_;
3733        return undef if $self->is_path_ignored($path);
3734        $self->{absent_dir}->{$pb->{path}} ||= [];
3735        push @{$self->{absent_dir}->{$pb->{path}}}, $path;
3736        undef;
3737}
3738
3739sub absent_file {
3740        my ($self, $path, $pb) = @_;
3741        return undef if $self->is_path_ignored($path);
3742        $self->{absent_file}->{$pb->{path}} ||= [];
3743        push @{$self->{absent_file}->{$pb->{path}}}, $path;
3744        undef;
3745}
3746
3747sub change_file_prop {
3748        my ($self, $fb, $prop, $value) = @_;
3749        return undef if $self->is_path_ignored($fb->{path});
3750        if ($prop eq 'svn:executable') {
3751                if ($fb->{mode_b} != 120000) {
3752                        $fb->{mode_b} = defined $value ? 100755 : 100644;
3753                }
3754        } elsif ($prop eq 'svn:special') {
3755                $fb->{mode_b} = defined $value ? 120000 : 100644;
3756        } else {
3757                $self->{file_prop}->{$fb->{path}} ||= {};
3758                $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
3759        }
3760        undef;
3761}
3762
3763sub apply_textdelta {
3764        my ($self, $fb, $exp) = @_;
3765        return undef if $self->is_path_ignored($fb->{path});
3766        my $fh = $::_repository->temp_acquire('svn_delta');
3767        # $fh gets auto-closed() by SVN::TxDelta::apply(),
3768        # (but $base does not,) so dup() it for reading in close_file
3769        open my $dup, '<&', $fh or croak $!;
3770        my $base = $::_repository->temp_acquire('git_blob');
3771
3772        if ($fb->{blob}) {
3773                my ($base_is_link, $size);
3774
3775                if ($fb->{mode_a} eq '120000' &&
3776                    ! $self->{empty_symlinks}->{$fb->{path}}) {
3777                        print $base 'link ' or die "print $!\n";
3778                        $base_is_link = 1;
3779                }
3780        retry:
3781                $size = $::_repository->cat_blob($fb->{blob}, $base);
3782                die "Failed to read object $fb->{blob}" if ($size < 0);
3783
3784                if (defined $exp) {
3785                        seek $base, 0, 0 or croak $!;
3786                        my $got = ::md5sum($base);
3787                        if ($got ne $exp) {
3788                                my $err = "Checksum mismatch: ".
3789                                       "$fb->{path} $fb->{blob}\n" .
3790                                       "expected: $exp\n" .
3791                                       "     got: $got\n";
3792                                if ($base_is_link) {
3793                                        warn $err,
3794                                             "Retrying... (possibly ",
3795                                             "a bad symlink from SVN)\n";
3796                                        $::_repository->temp_reset($base);
3797                                        $base_is_link = 0;
3798                                        goto retry;
3799                                }
3800                                die $err;
3801                        }
3802                }
3803        }
3804        seek $base, 0, 0 or croak $!;
3805        $fb->{fh} = $fh;
3806        $fb->{base} = $base;
3807        [ SVN::TxDelta::apply($base, $dup, undef, $fb->{path}, $fb->{pool}) ];
3808}
3809
3810sub close_file {
3811        my ($self, $fb, $exp) = @_;
3812        return undef if $self->is_path_ignored($fb->{path});
3813
3814        my $hash;
3815        my $path = $self->git_path($fb->{path});
3816        if (my $fh = $fb->{fh}) {
3817                if (defined $exp) {
3818                        seek($fh, 0, 0) or croak $!;
3819                        my $got = ::md5sum($fh);
3820                        if ($got ne $exp) {
3821                                die "Checksum mismatch: $path\n",
3822                                    "expected: $exp\n    got: $got\n";
3823                        }
3824                }
3825                if ($fb->{mode_b} == 120000) {
3826                        sysseek($fh, 0, 0) or croak $!;
3827                        my $rd = sysread($fh, my $buf, 5);
3828
3829                        if (!defined $rd) {
3830                                croak "sysread: $!\n";
3831                        } elsif ($rd == 0) {
3832                                warn "$path has mode 120000",
3833                                     " but it points to nothing\n",
3834                                     "converting to an empty file with mode",
3835                                     " 100644\n";
3836                                $fb->{mode_b} = '100644';
3837                        } elsif ($buf ne 'link ') {
3838                                warn "$path has mode 120000",
3839                                     " but is not a link\n";
3840                        } else {
3841                                my $tmp_fh = $::_repository->temp_acquire(
3842                                        'svn_hash');
3843                                my $res;
3844                                while ($res = sysread($fh, my $str, 1024)) {
3845                                        my $out = syswrite($tmp_fh, $str, $res);
3846                                        defined($out) && $out == $res
3847                                                or croak("write ",
3848                                                        Git::temp_path($tmp_fh),
3849                                                        ": $!\n");
3850                                }
3851                                defined $res or croak $!;
3852
3853                                ($fh, $tmp_fh) = ($tmp_fh, $fh);
3854                                Git::temp_release($tmp_fh, 1);
3855                        }
3856                }
3857
3858                $hash = $::_repository->hash_and_insert_object(
3859                                Git::temp_path($fh));
3860                $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
3861
3862                Git::temp_release($fb->{base}, 1);
3863                Git::temp_release($fh, 1);
3864        } else {
3865                $hash = $fb->{blob} or die "no blob information\n";
3866        }
3867        $fb->{pool}->clear;
3868        $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
3869        print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
3870        undef;
3871}
3872
3873sub abort_edit {
3874        my $self = shift;
3875        $self->{nr} = $self->{gii}->{nr};
3876        delete $self->{gii};
3877        $self->SUPER::abort_edit(@_);
3878}
3879
3880sub close_edit {
3881        my $self = shift;
3882        $self->{git_commit_ok} = 1;
3883        $self->{nr} = $self->{gii}->{nr};
3884        delete $self->{gii};
3885        $self->SUPER::close_edit(@_);
3886}
3887
3888package SVN::Git::Editor;
3889use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
3890use strict;
3891use warnings;
3892use Carp qw/croak/;
3893use IO::File;
3894
3895sub new {
3896        my ($class, $opts) = @_;
3897        foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
3898                die "$_ required!\n" unless (defined $opts->{$_});
3899        }
3900
3901        my $pool = SVN::Pool->new;
3902        my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
3903        my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
3904                                     $opts->{r}, $mods);
3905
3906        # $opts->{ra} functions should not be used after this:
3907        my @ce  = $opts->{ra}->get_commit_editor($opts->{log},
3908                                                $opts->{editor_cb}, $pool);
3909        my $self = SVN::Delta::Editor->new(@ce, $pool);
3910        bless $self, $class;
3911        foreach (qw/svn_path r tree_a tree_b/) {
3912                $self->{$_} = $opts->{$_};
3913        }
3914        $self->{url} = $opts->{ra}->{url};
3915        $self->{mods} = $mods;
3916        $self->{types} = $types;
3917        $self->{pool} = $pool;
3918        $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
3919        $self->{rm} = { };
3920        $self->{path_prefix} = length $self->{svn_path} ?
3921                               "$self->{svn_path}/" : '';
3922        $self->{config} = $opts->{config};
3923        return $self;
3924}
3925
3926sub generate_diff {
3927        my ($tree_a, $tree_b) = @_;
3928        my @diff_tree = qw(diff-tree -z -r);
3929        if ($_cp_similarity) {
3930                push @diff_tree, "-C$_cp_similarity";
3931        } else {
3932                push @diff_tree, '-C';
3933        }
3934        push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
3935        push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
3936        push @diff_tree, $tree_a, $tree_b;
3937        my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
3938        local $/ = "\0";
3939        my $state = 'meta';
3940        my @mods;
3941        while (<$diff_fh>) {
3942                chomp $_; # this gets rid of the trailing "\0"
3943                if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
3944                                        ($::sha1)\s($::sha1)\s
3945                                        ([MTCRAD])\d*$/xo) {
3946                        push @mods, {   mode_a => $1, mode_b => $2,
3947                                        sha1_a => $3, sha1_b => $4,
3948                                        chg => $5 };
3949                        if ($5 =~ /^(?:C|R)$/) {
3950                                $state = 'file_a';
3951                        } else {
3952                                $state = 'file_b';
3953                        }
3954                } elsif ($state eq 'file_a') {
3955                        my $x = $mods[$#mods] or croak "Empty array\n";
3956                        if ($x->{chg} !~ /^(?:C|R)$/) {
3957                                croak "Error parsing $_, $x->{chg}\n";
3958                        }
3959                        $x->{file_a} = $_;
3960                        $state = 'file_b';
3961                } elsif ($state eq 'file_b') {
3962                        my $x = $mods[$#mods] or croak "Empty array\n";
3963                        if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
3964                                croak "Error parsing $_, $x->{chg}\n";
3965                        }
3966                        if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
3967                                croak "Error parsing $_, $x->{chg}\n";
3968                        }
3969                        $x->{file_b} = $_;
3970                        $state = 'meta';
3971                } else {
3972                        croak "Error parsing $_\n";
3973                }
3974        }
3975        command_close_pipe($diff_fh, $ctx);
3976        \@mods;
3977}
3978
3979sub check_diff_paths {
3980        my ($ra, $pfx, $rev, $mods) = @_;
3981        my %types;
3982        $pfx .= '/' if length $pfx;
3983
3984        sub type_diff_paths {
3985                my ($ra, $types, $path, $rev) = @_;
3986                my @p = split m#/+#, $path;
3987                my $c = shift @p;
3988                unless (defined $types->{$c}) {
3989                        $types->{$c} = $ra->check_path($c, $rev);
3990                }
3991                while (@p) {
3992                        $c .= '/' . shift @p;
3993                        next if defined $types->{$c};
3994                        $types->{$c} = $ra->check_path($c, $rev);
3995                }
3996        }
3997
3998        foreach my $m (@$mods) {
3999                foreach my $f (qw/file_a file_b/) {
4000                        next unless defined $m->{$f};
4001                        my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
4002                        if (length $pfx.$dir && ! defined $types{$dir}) {
4003                                type_diff_paths($ra, \%types, $pfx.$dir, $rev);
4004                        }
4005                }
4006        }
4007        \%types;
4008}
4009
4010sub split_path {
4011        return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
4012}
4013
4014sub repo_path {
4015        my ($self, $path) = @_;
4016        $self->{path_prefix}.(defined $path ? $path : '');
4017}
4018
4019sub url_path {
4020        my ($self, $path) = @_;
4021        if ($self->{url} =~ m#^https?://#) {
4022                $path =~ s!([^~a-zA-Z0-9_./-])!uc sprintf("%%%02x",ord($1))!eg;
4023        }
4024        $self->{url} . '/' . $self->repo_path($path);
4025}
4026
4027sub rmdirs {
4028        my ($self) = @_;
4029        my $rm = $self->{rm};
4030        delete $rm->{''}; # we never delete the url we're tracking
4031        return unless %$rm;
4032
4033        foreach (keys %$rm) {
4034                my @d = split m#/#, $_;
4035                my $c = shift @d;
4036                $rm->{$c} = 1;
4037                while (@d) {
4038                        $c .= '/' . shift @d;
4039                        $rm->{$c} = 1;
4040                }
4041        }
4042        delete $rm->{$self->{svn_path}};
4043        delete $rm->{''}; # we never delete the url we're tracking
4044        return unless %$rm;
4045
4046        my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
4047                                             $self->{tree_b});
4048        local $/ = "\0";
4049        while (<$fh>) {
4050                chomp;
4051                my @dn = split m#/#, $_;
4052                while (pop @dn) {
4053                        delete $rm->{join '/', @dn};
4054                }
4055                unless (%$rm) {
4056                        close $fh;
4057                        return;
4058                }
4059        }
4060        command_close_pipe($fh, $ctx);
4061
4062        my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
4063        foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
4064                $self->close_directory($bat->{$d}, $p);
4065                my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
4066                print "\tD+\t$d/\n" unless $::_q;
4067                $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
4068                delete $bat->{$d};
4069        }
4070}
4071
4072sub open_or_add_dir {
4073        my ($self, $full_path, $baton) = @_;
4074        my $t = $self->{types}->{$full_path};
4075        if (!defined $t) {
4076                die "$full_path not known in r$self->{r} or we have a bug!\n";
4077        }
4078        {
4079                no warnings 'once';
4080                # SVN::Node::none and SVN::Node::file are used only once,
4081                # so we're shutting up Perl's warnings about them.
4082                if ($t == $SVN::Node::none) {
4083                        return $self->add_directory($full_path, $baton,
4084                            undef, -1, $self->{pool});
4085                } elsif ($t == $SVN::Node::dir) {
4086                        return $self->open_directory($full_path, $baton,
4087                            $self->{r}, $self->{pool});
4088                } # no warnings 'once'
4089                print STDERR "$full_path already exists in repository at ",
4090                    "r$self->{r} and it is not a directory (",
4091                    ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
4092        } # no warnings 'once'
4093        exit 1;
4094}
4095
4096sub ensure_path {
4097        my ($self, $path) = @_;
4098        my $bat = $self->{bat};
4099        my $repo_path = $self->repo_path($path);
4100        return $bat->{''} unless (length $repo_path);
4101        my @p = split m#/+#, $repo_path;
4102        my $c = shift @p;
4103        $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
4104        while (@p) {
4105                my $c0 = $c;
4106                $c .= '/' . shift @p;
4107                $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
4108        }
4109        return $bat->{$c};
4110}
4111
4112# Subroutine to convert a globbing pattern to a regular expression.
4113# From perl cookbook.
4114sub glob2pat {
4115        my $globstr = shift;
4116        my %patmap = ('*' => '.*', '?' => '.', '[' => '[', ']' => ']');
4117        $globstr =~ s{(.)} { $patmap{$1} || "\Q$1" }ge;
4118        return '^' . $globstr . '$';
4119}
4120
4121sub check_autoprop {
4122        my ($self, $pattern, $properties, $file, $fbat) = @_;
4123        # Convert the globbing pattern to a regular expression.
4124        my $regex = glob2pat($pattern);
4125        # Check if the pattern matches the file name.
4126        if($file =~ m/($regex)/) {
4127                # Parse the list of properties to set.
4128                my @props = split(/;/, $properties);
4129                foreach my $prop (@props) {
4130                        # Parse 'name=value' syntax and set the property.
4131                        if ($prop =~ /([^=]+)=(.*)/) {
4132                                my ($n,$v) = ($1,$2);
4133                                for ($n, $v) {
4134                                        s/^\s+//; s/\s+$//;
4135                                }
4136                                $self->change_file_prop($fbat, $n, $v);
4137                        }
4138                }
4139        }
4140}
4141
4142sub apply_autoprops {
4143        my ($self, $file, $fbat) = @_;
4144        my $conf_t = ${$self->{config}}{'config'};
4145        no warnings 'once';
4146        # Check [miscellany]/enable-auto-props in svn configuration.
4147        if (SVN::_Core::svn_config_get_bool(
4148                $conf_t,
4149                $SVN::_Core::SVN_CONFIG_SECTION_MISCELLANY,
4150                $SVN::_Core::SVN_CONFIG_OPTION_ENABLE_AUTO_PROPS,
4151                0)) {
4152                # Auto-props are enabled.  Enumerate them to look for matches.
4153                my $callback = sub {
4154                        $self->check_autoprop($_[0], $_[1], $file, $fbat);
4155                };
4156                SVN::_Core::svn_config_enumerate(
4157                        $conf_t,
4158                        $SVN::_Core::SVN_CONFIG_SECTION_AUTO_PROPS,
4159                        $callback);
4160        }
4161}
4162
4163sub A {
4164        my ($self, $m) = @_;
4165        my ($dir, $file) = split_path($m->{file_b});
4166        my $pbat = $self->ensure_path($dir);
4167        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4168                                        undef, -1);
4169        print "\tA\t$m->{file_b}\n" unless $::_q;
4170        $self->apply_autoprops($file, $fbat);
4171        $self->chg_file($fbat, $m);
4172        $self->close_file($fbat,undef,$self->{pool});
4173}
4174
4175sub C {
4176        my ($self, $m) = @_;
4177        my ($dir, $file) = split_path($m->{file_b});
4178        my $pbat = $self->ensure_path($dir);
4179        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4180                                $self->url_path($m->{file_a}), $self->{r});
4181        print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
4182        $self->chg_file($fbat, $m);
4183        $self->close_file($fbat,undef,$self->{pool});
4184}
4185
4186sub delete_entry {
4187        my ($self, $path, $pbat) = @_;
4188        my $rpath = $self->repo_path($path);
4189        my ($dir, $file) = split_path($rpath);
4190        $self->{rm}->{$dir} = 1;
4191        $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
4192}
4193
4194sub R {
4195        my ($self, $m) = @_;
4196        my ($dir, $file) = split_path($m->{file_b});
4197        my $pbat = $self->ensure_path($dir);
4198        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4199                                $self->url_path($m->{file_a}), $self->{r});
4200        print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
4201        $self->apply_autoprops($file, $fbat);
4202        $self->chg_file($fbat, $m);
4203        $self->close_file($fbat,undef,$self->{pool});
4204
4205        ($dir, $file) = split_path($m->{file_a});
4206        $pbat = $self->ensure_path($dir);
4207        $self->delete_entry($m->{file_a}, $pbat);
4208}
4209
4210sub M {
4211        my ($self, $m) = @_;
4212        my ($dir, $file) = split_path($m->{file_b});
4213        my $pbat = $self->ensure_path($dir);
4214        my $fbat = $self->open_file($self->repo_path($m->{file_b}),
4215                                $pbat,$self->{r},$self->{pool});
4216        print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
4217        $self->chg_file($fbat, $m);
4218        $self->close_file($fbat,undef,$self->{pool});
4219}
4220
4221sub T { shift->M(@_) }
4222
4223sub change_file_prop {
4224        my ($self, $fbat, $pname, $pval) = @_;
4225        $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
4226}
4227
4228sub _chg_file_get_blob ($$$$) {
4229        my ($self, $fbat, $m, $which) = @_;
4230        my $fh = $::_repository->temp_acquire("git_blob_$which");
4231        if ($m->{"mode_$which"} =~ /^120/) {
4232                print $fh 'link ' or croak $!;
4233                $self->change_file_prop($fbat,'svn:special','*');
4234        } elsif ($m->{mode_a} =~ /^120/ && $m->{"mode_$which"} !~ /^120/) {
4235                $self->change_file_prop($fbat,'svn:special',undef);
4236        }
4237        my $blob = $m->{"sha1_$which"};
4238        return ($fh,) if ($blob =~ /^0{40}$/);
4239        my $size = $::_repository->cat_blob($blob, $fh);
4240        croak "Failed to read object $blob" if ($size < 0);
4241        $fh->flush == 0 or croak $!;
4242        seek $fh, 0, 0 or croak $!;
4243
4244        my $exp = ::md5sum($fh);
4245        seek $fh, 0, 0 or croak $!;
4246        return ($fh, $exp);
4247}
4248
4249sub chg_file {
4250        my ($self, $fbat, $m) = @_;
4251        if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
4252                $self->change_file_prop($fbat,'svn:executable','*');
4253        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
4254                $self->change_file_prop($fbat,'svn:executable',undef);
4255        }
4256        my ($fh_a, $exp_a) = _chg_file_get_blob $self, $fbat, $m, 'a';
4257        my ($fh_b, $exp_b) = _chg_file_get_blob $self, $fbat, $m, 'b';
4258        my $pool = SVN::Pool->new;
4259        my $atd = $self->apply_textdelta($fbat, $exp_a, $pool);
4260        if (-s $fh_a) {
4261                my $txstream = SVN::TxDelta::new ($fh_a, $fh_b, $pool);
4262                my $res = SVN::TxDelta::send_txstream($txstream, @$atd, $pool);
4263                if (defined $res) {
4264                        die "Unexpected result from send_txstream: $res\n",
4265                            "(SVN::Core::VERSION: $SVN::Core::VERSION)\n";
4266                }
4267        } else {
4268                my $got = SVN::TxDelta::send_stream($fh_b, @$atd, $pool);
4269                die "Checksum mismatch\nexpected: $exp_b\ngot: $got\n"
4270                    if ($got ne $exp_b);
4271        }
4272        Git::temp_release($fh_b, 1);
4273        Git::temp_release($fh_a, 1);
4274        $pool->clear;
4275}
4276
4277sub D {
4278        my ($self, $m) = @_;
4279        my ($dir, $file) = split_path($m->{file_b});
4280        my $pbat = $self->ensure_path($dir);
4281        print "\tD\t$m->{file_b}\n" unless $::_q;
4282        $self->delete_entry($m->{file_b}, $pbat);
4283}
4284
4285sub close_edit {
4286        my ($self) = @_;
4287        my ($p,$bat) = ($self->{pool}, $self->{bat});
4288        foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
4289                next if $_ eq '';
4290                $self->close_directory($bat->{$_}, $p);
4291        }
4292        $self->close_directory($bat->{''}, $p);
4293        $self->SUPER::close_edit($p);
4294        $p->clear;
4295}
4296
4297sub abort_edit {
4298        my ($self) = @_;
4299        $self->SUPER::abort_edit($self->{pool});
4300}
4301
4302sub DESTROY {
4303        my $self = shift;
4304        $self->SUPER::DESTROY(@_);
4305        $self->{pool}->clear;
4306}
4307
4308# this drives the editor
4309sub apply_diff {
4310        my ($self) = @_;
4311        my $mods = $self->{mods};
4312        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
4313        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
4314                my $f = $m->{chg};
4315                if (defined $o{$f}) {
4316                        $self->$f($m);
4317                } else {
4318                        fatal("Invalid change type: $f");
4319                }
4320        }
4321        $self->rmdirs if $_rmdir;
4322        if (@$mods == 0) {
4323                $self->abort_edit;
4324        } else {
4325                $self->close_edit;
4326        }
4327        return scalar @$mods;
4328}
4329
4330package Git::SVN::Ra;
4331use vars qw/@ISA $config_dir $_log_window_size/;
4332use strict;
4333use warnings;
4334my ($ra_invalid, $can_do_switch, %ignored_err, $RA);
4335
4336BEGIN {
4337        # enforce temporary pool usage for some simple functions
4338        no strict 'refs';
4339        for my $f (qw/rev_proplist get_latest_revnum get_uuid get_repos_root
4340                      get_file/) {
4341                my $SUPER = "SUPER::$f";
4342                *$f = sub {
4343                        my $self = shift;
4344                        my $pool = SVN::Pool->new;
4345                        my @ret = $self->$SUPER(@_,$pool);
4346                        $pool->clear;
4347                        wantarray ? @ret : $ret[0];
4348                };
4349        }
4350}
4351
4352sub _auth_providers () {
4353        [
4354          SVN::Client::get_simple_provider(),
4355          SVN::Client::get_ssl_server_trust_file_provider(),
4356          SVN::Client::get_simple_prompt_provider(
4357            \&Git::SVN::Prompt::simple, 2),
4358          SVN::Client::get_ssl_client_cert_file_provider(),
4359          SVN::Client::get_ssl_client_cert_prompt_provider(
4360            \&Git::SVN::Prompt::ssl_client_cert, 2),
4361          SVN::Client::get_ssl_client_cert_pw_file_provider(),
4362          SVN::Client::get_ssl_client_cert_pw_prompt_provider(
4363            \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
4364          SVN::Client::get_username_provider(),
4365          SVN::Client::get_ssl_server_trust_prompt_provider(
4366            \&Git::SVN::Prompt::ssl_server_trust),
4367          SVN::Client::get_username_prompt_provider(
4368            \&Git::SVN::Prompt::username, 2)
4369        ]
4370}
4371
4372sub escape_uri_only {
4373        my ($uri) = @_;
4374        my @tmp;
4375        foreach (split m{/}, $uri) {
4376                s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
4377                push @tmp, $_;
4378        }
4379        join('/', @tmp);
4380}
4381
4382sub escape_url {
4383        my ($url) = @_;
4384        if ($url =~ m#^(https?)://([^/]+)(.*)$#) {
4385                my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
4386                $url = "$scheme://$domain$uri";
4387        }
4388        $url;
4389}
4390
4391sub new {
4392        my ($class, $url) = @_;
4393        $url =~ s!/+$!!;
4394        return $RA if ($RA && $RA->{url} eq $url);
4395
4396        SVN::_Core::svn_config_ensure($config_dir, undef);
4397        my ($baton, $callbacks) = SVN::Core::auth_open_helper(_auth_providers);
4398        my $config = SVN::Core::config_get_config($config_dir);
4399        $RA = undef;
4400        my $dont_store_passwords = 1;
4401        my $conf_t = ${$config}{'config'};
4402        {
4403                no warnings 'once';
4404                # The usage of $SVN::_Core::SVN_CONFIG_* variables
4405                # produces warnings that variables are used only once.
4406                # I had not found the better way to shut them up, so
4407                # the warnings of type 'once' are disabled in this block.
4408                if (SVN::_Core::svn_config_get_bool($conf_t,
4409                    $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4410                    $SVN::_Core::SVN_CONFIG_OPTION_STORE_PASSWORDS,
4411                    1) == 0) {
4412                        SVN::_Core::svn_auth_set_parameter($baton,
4413                            $SVN::_Core::SVN_AUTH_PARAM_DONT_STORE_PASSWORDS,
4414                            bless (\$dont_store_passwords, "_p_void"));
4415                }
4416                if (SVN::_Core::svn_config_get_bool($conf_t,
4417                    $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4418                    $SVN::_Core::SVN_CONFIG_OPTION_STORE_AUTH_CREDS,
4419                    1) == 0) {
4420                        $Git::SVN::Prompt::_no_auth_cache = 1;
4421                }
4422        } # no warnings 'once'
4423        my $self = SVN::Ra->new(url => escape_url($url), auth => $baton,
4424                              config => $config,
4425                              pool => SVN::Pool->new,
4426                              auth_provider_callbacks => $callbacks);
4427        $self->{url} = $url;
4428        $self->{svn_path} = $url;
4429        $self->{repos_root} = $self->get_repos_root;
4430        $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
4431        $self->{cache} = { check_path => { r => 0, data => {} },
4432                           get_dir => { r => 0, data => {} } };
4433        $RA = bless $self, $class;
4434}
4435
4436sub check_path {
4437        my ($self, $path, $r) = @_;
4438        my $cache = $self->{cache}->{check_path};
4439        if ($r == $cache->{r} && exists $cache->{data}->{$path}) {
4440                return $cache->{data}->{$path};
4441        }
4442        my $pool = SVN::Pool->new;
4443        my $t = $self->SUPER::check_path($path, $r, $pool);
4444        $pool->clear;
4445        if ($r != $cache->{r}) {
4446                %{$cache->{data}} = ();
4447                $cache->{r} = $r;
4448        }
4449        $cache->{data}->{$path} = $t;
4450}
4451
4452sub get_dir {
4453        my ($self, $dir, $r) = @_;
4454        my $cache = $self->{cache}->{get_dir};
4455        if ($r == $cache->{r}) {
4456                if (my $x = $cache->{data}->{$dir}) {
4457                        return wantarray ? @$x : $x->[0];
4458                }
4459        }
4460        my $pool = SVN::Pool->new;
4461        my ($d, undef, $props) = $self->SUPER::get_dir($dir, $r, $pool);
4462        my %dirents = map { $_ => { kind => $d->{$_}->kind } } keys %$d;
4463        $pool->clear;
4464        if ($r != $cache->{r}) {
4465                %{$cache->{data}} = ();
4466                $cache->{r} = $r;
4467        }
4468        $cache->{data}->{$dir} = [ \%dirents, $r, $props ];
4469        wantarray ? (\%dirents, $r, $props) : \%dirents;
4470}
4471
4472sub DESTROY {
4473        # do not call the real DESTROY since we store ourselves in $RA
4474}
4475
4476# get_log(paths, start, end, limit,
4477#         discover_changed_paths, strict_node_history, receiver)
4478sub get_log {
4479        my ($self, @args) = @_;
4480        my $pool = SVN::Pool->new;
4481
4482        # svn_log_changed_path_t objects passed to get_log are likely to be
4483        # overwritten even if only the refs are copied to an external variable,
4484        # so we should dup the structures in their entirety.  Using an
4485        # externally passed pool (instead of our temporary and quickly cleared
4486        # pool in Git::SVN::Ra) does not help matters at all...
4487        my $receiver = pop @args;
4488        my $prefix = "/".$self->{svn_path};
4489        $prefix =~ s#/+($)##;
4490        my $prefix_regex = qr#^\Q$prefix\E#;
4491        push(@args, sub {
4492                my ($paths) = $_[0];
4493                return &$receiver(@_) unless $paths;
4494                $_[0] = ();
4495                foreach my $p (keys %$paths) {
4496                        my $i = $paths->{$p};
4497                        # Make path relative to our url, not repos_root
4498                        $p =~ s/$prefix_regex//;
4499                        my %s = map { $_ => $i->$_; }
4500                                qw/copyfrom_path copyfrom_rev action/;
4501                        if ($s{'copyfrom_path'}) {
4502                                $s{'copyfrom_path'} =~ s/$prefix_regex//;
4503                        }
4504                        $_[0]{$p} = \%s;
4505                }
4506                &$receiver(@_);
4507        });
4508
4509
4510        # the limit parameter was not supported in SVN 1.1.x, so we
4511        # drop it.  Therefore, the receiver callback passed to it
4512        # is made aware of this limitation by being wrapped if
4513        # the limit passed to is being wrapped.
4514        if ($SVN::Core::VERSION le '1.2.0') {
4515                my $limit = splice(@args, 3, 1);
4516                if ($limit > 0) {
4517                        my $receiver = pop @args;
4518                        push(@args, sub { &$receiver(@_) if (--$limit >= 0) });
4519                }
4520        }
4521        my $ret = $self->SUPER::get_log(@args, $pool);
4522        $pool->clear;
4523        $ret;
4524}
4525
4526sub trees_match {
4527        my ($self, $url1, $rev1, $url2, $rev2) = @_;
4528        my $ctx = SVN::Client->new(auth => _auth_providers);
4529        my $out = IO::File->new_tmpfile;
4530
4531        # older SVN (1.1.x) doesn't take $pool as the last parameter for
4532        # $ctx->diff(), so we'll create a default one
4533        my $pool = SVN::Pool->new_default_sub;
4534
4535        $ra_invalid = 1; # this will open a new SVN::Ra connection to $url1
4536        $ctx->diff([], $url1, $rev1, $url2, $rev2, 1, 1, 0, $out, $out);
4537        $out->flush;
4538        my $ret = (($out->stat)[7] == 0);
4539        close $out or croak $!;
4540
4541        $ret;
4542}
4543
4544sub get_commit_editor {
4545        my ($self, $log, $cb, $pool) = @_;
4546        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
4547        $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
4548}
4549
4550sub gs_do_update {
4551        my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
4552        my $new = ($rev_a == $rev_b);
4553        my $path = $gs->{path};
4554
4555        if ($new && -e $gs->{index}) {
4556                unlink $gs->{index} or die
4557                  "Couldn't unlink index: $gs->{index}: $!\n";
4558        }
4559        my $pool = SVN::Pool->new;
4560        $editor->set_path_strip($path);
4561        my (@pc) = split m#/#, $path;
4562        my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
4563                                        1, $editor, $pool);
4564        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
4565
4566        # Since we can't rely on svn_ra_reparent being available, we'll
4567        # just have to do some magic with set_path to make it so
4568        # we only want a partial path.
4569        my $sp = '';
4570        my $final = join('/', @pc);
4571        while (@pc) {
4572                $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
4573                $sp .= '/' if length $sp;
4574                $sp .= shift @pc;
4575        }
4576        die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
4577
4578        $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
4579
4580        $reporter->finish_report($pool);
4581        $pool->clear;
4582        $editor->{git_commit_ok};
4583}
4584
4585# this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
4586# svn_ra_reparent didn't work before 1.4)
4587sub gs_do_switch {
4588        my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
4589        my $path = $gs->{path};
4590        my $pool = SVN::Pool->new;
4591
4592        my $full_url = $self->{url};
4593        my $old_url = $full_url;
4594        $full_url .= '/' . $path if length $path;
4595        my ($ra, $reparented);
4596
4597        if ($old_url =~ m#^svn(\+ssh)?://# ||
4598            ($full_url =~ m#^https?://# &&
4599             escape_url($full_url) ne $full_url)) {
4600                $_[0] = undef;
4601                $self = undef;
4602                $RA = undef;
4603                $ra = Git::SVN::Ra->new($full_url);
4604                $ra_invalid = 1;
4605        } elsif ($old_url ne $full_url) {
4606                SVN::_Ra::svn_ra_reparent($self->{session}, $full_url, $pool);
4607                $self->{url} = $full_url;
4608                $reparented = 1;
4609        }
4610
4611        $ra ||= $self;
4612        $url_b = escape_url($url_b);
4613        my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
4614        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
4615        $reporter->set_path('', $rev_a, 0, @lock, $pool);
4616        $reporter->finish_report($pool);
4617
4618        if ($reparented) {
4619                SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
4620                $self->{url} = $old_url;
4621        }
4622
4623        $pool->clear;
4624        $editor->{git_commit_ok};
4625}
4626
4627sub longest_common_path {
4628        my ($gsv, $globs) = @_;
4629        my %common;
4630        my $common_max = scalar @$gsv;
4631
4632        foreach my $gs (@$gsv) {
4633                my @tmp = split m#/#, $gs->{path};
4634                my $p = '';
4635                foreach (@tmp) {
4636                        $p .= length($p) ? "/$_" : $_;
4637                        $common{$p} ||= 0;
4638                        $common{$p}++;
4639                }
4640        }
4641        $globs ||= [];
4642        $common_max += scalar @$globs;
4643        foreach my $glob (@$globs) {
4644                my @tmp = split m#/#, $glob->{path}->{left};
4645                my $p = '';
4646                foreach (@tmp) {
4647                        $p .= length($p) ? "/$_" : $_;
4648                        $common{$p} ||= 0;
4649                        $common{$p}++;
4650                }
4651        }
4652
4653        my $longest_path = '';
4654        foreach (sort {length $b <=> length $a} keys %common) {
4655                if ($common{$_} == $common_max) {
4656                        $longest_path = $_;
4657                        last;
4658                }
4659        }
4660        $longest_path;
4661}
4662
4663sub gs_fetch_loop_common {
4664        my ($self, $base, $head, $gsv, $globs) = @_;
4665        return if ($base > $head);
4666        my $inc = $_log_window_size;
4667        my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
4668        my $longest_path = longest_common_path($gsv, $globs);
4669        my $ra_url = $self->{url};
4670        my $find_trailing_edge;
4671        while (1) {
4672                my %revs;
4673                my $err;
4674                my $err_handler = $SVN::Error::handler;
4675                $SVN::Error::handler = sub {
4676                        ($err) = @_;
4677                        skip_unknown_revs($err);
4678                };
4679                sub _cb {
4680                        my ($paths, $r, $author, $date, $log) = @_;
4681                        [ $paths,
4682                          { author => $author, date => $date, log => $log } ];
4683                }
4684                $self->get_log([$longest_path], $min, $max, 0, 1, 1,
4685                               sub { $revs{$_[1]} = _cb(@_) });
4686                if ($err) {
4687                        print "Checked through r$max\r";
4688                } else {
4689                        $find_trailing_edge = 1;
4690                }
4691                if ($err and $find_trailing_edge) {
4692                        print STDERR "Path '$longest_path' ",
4693                                     "was probably deleted:\n",
4694                                     $err->expanded_message,
4695                                     "\nWill attempt to follow ",
4696                                     "revisions r$min .. r$max ",
4697                                     "committed before the deletion\n";
4698                        my $hi = $max;
4699                        while (--$hi >= $min) {
4700                                my $ok;
4701                                $self->get_log([$longest_path], $min, $hi,
4702                                               0, 1, 1, sub {
4703                                               $ok = $_[1];
4704                                               $revs{$_[1]} = _cb(@_) });
4705                                if ($ok) {
4706                                        print STDERR "r$min .. r$ok OK\n";
4707                                        last;
4708                                }
4709                        }
4710                        $find_trailing_edge = 0;
4711                }
4712                $SVN::Error::handler = $err_handler;
4713
4714                my %exists = map { $_->{path} => $_ } @$gsv;
4715                foreach my $r (sort {$a <=> $b} keys %revs) {
4716                        my ($paths, $logged) = @{$revs{$r}};
4717
4718                        foreach my $gs ($self->match_globs(\%exists, $paths,
4719                                                           $globs, $r)) {
4720                                if ($gs->rev_map_max >= $r) {
4721                                        next;
4722                                }
4723                                next unless $gs->match_paths($paths, $r);
4724                                $gs->{logged_rev_props} = $logged;
4725                                if (my $last_commit = $gs->last_commit) {
4726                                        $gs->assert_index_clean($last_commit);
4727                                }
4728                                my $log_entry = $gs->do_fetch($paths, $r);
4729                                if ($log_entry) {
4730                                        $gs->do_git_commit($log_entry);
4731                                }
4732                                $INDEX_FILES{$gs->{index}} = 1;
4733                        }
4734                        foreach my $g (@$globs) {
4735                                my $k = "svn-remote.$g->{remote}." .
4736                                        "$g->{t}-maxRev";
4737                                Git::SVN::tmp_config($k, $r);
4738                        }
4739                        if ($ra_invalid) {
4740                                $_[0] = undef;
4741                                $self = undef;
4742                                $RA = undef;
4743                                $self = Git::SVN::Ra->new($ra_url);
4744                                $ra_invalid = undef;
4745                        }
4746                }
4747                # pre-fill the .rev_db since it'll eventually get filled in
4748                # with '0' x40 if something new gets committed
4749                foreach my $gs (@$gsv) {
4750                        next if $gs->rev_map_max >= $max;
4751                        next if defined $gs->rev_map_get($max);
4752                        $gs->rev_map_set($max, 0 x40);
4753                }
4754                foreach my $g (@$globs) {
4755                        my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
4756                        Git::SVN::tmp_config($k, $max);
4757                }
4758                last if $max >= $head;
4759                $min = $max + 1;
4760                $max += $inc;
4761                $max = $head if ($max > $head);
4762        }
4763        Git::SVN::gc();
4764}
4765
4766sub get_dir_globbed {
4767        my ($self, $left, $depth, $r) = @_;
4768
4769        my @x = eval { $self->get_dir($left, $r) };
4770        return unless scalar @x == 3;
4771        my $dirents = $x[0];
4772        my @finalents;
4773        foreach my $de (keys %$dirents) {
4774                next if $dirents->{$de}->{kind} != $SVN::Node::dir;
4775                if ($depth > 1) {
4776                        my @args = ("$left/$de", $depth - 1, $r);
4777                        foreach my $dir ($self->get_dir_globbed(@args)) {
4778                                push @finalents, "$de/$dir";
4779                        }
4780                } else {
4781                        push @finalents, $de;
4782                }
4783        }
4784        @finalents;
4785}
4786
4787sub match_globs {
4788        my ($self, $exists, $paths, $globs, $r) = @_;
4789
4790        sub get_dir_check {
4791                my ($self, $exists, $g, $r) = @_;
4792
4793                my @dirs = $self->get_dir_globbed($g->{path}->{left},
4794                                                  $g->{path}->{depth},
4795                                                  $r);
4796
4797                foreach my $de (@dirs) {
4798                        my $p = $g->{path}->full_path($de);
4799                        next if $exists->{$p};
4800                        next if (length $g->{path}->{right} &&
4801                                 ($self->check_path($p, $r) !=
4802                                  $SVN::Node::dir));
4803                        $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
4804                                         $g->{ref}->full_path($de), 1);
4805                }
4806        }
4807        foreach my $g (@$globs) {
4808                if (my $path = $paths->{"/$g->{path}->{left}"}) {
4809                        if ($path->{action} =~ /^[AR]$/) {
4810                                get_dir_check($self, $exists, $g, $r);
4811                        }
4812                }
4813                foreach (keys %$paths) {
4814                        if (/$g->{path}->{left_regex}/ &&
4815                            !/$g->{path}->{regex}/) {
4816                                next if $paths->{$_}->{action} !~ /^[AR]$/;
4817                                get_dir_check($self, $exists, $g, $r);
4818                        }
4819                        next unless /$g->{path}->{regex}/;
4820                        my $p = $1;
4821                        my $pathname = $g->{path}->full_path($p);
4822                        next if $exists->{$pathname};
4823                        next if ($self->check_path($pathname, $r) !=
4824                                 $SVN::Node::dir);
4825                        $exists->{$pathname} = Git::SVN->init(
4826                                              $self->{url}, $pathname, undef,
4827                                              $g->{ref}->full_path($p), 1);
4828                }
4829                my $c = '';
4830                foreach (split m#/#, $g->{path}->{left}) {
4831                        $c .= "/$_";
4832                        next unless ($paths->{$c} &&
4833                                     ($paths->{$c}->{action} =~ /^[AR]$/));
4834                        get_dir_check($self, $exists, $g, $r);
4835                }
4836        }
4837        values %$exists;
4838}
4839
4840sub minimize_url {
4841        my ($self) = @_;
4842        return $self->{url} if ($self->{url} eq $self->{repos_root});
4843        my $url = $self->{repos_root};
4844        my @components = split(m!/!, $self->{svn_path});
4845        my $c = '';
4846        do {
4847                $url .= "/$c" if length $c;
4848                eval {
4849                        my $ra = (ref $self)->new($url);
4850                        my $latest = $ra->get_latest_revnum;
4851                        $ra->get_log("", $latest, 0, 1, 0, 1, sub {});
4852                };
4853        } while ($@ && ($c = shift @components));
4854        $url;
4855}
4856
4857sub can_do_switch {
4858        my $self = shift;
4859        unless (defined $can_do_switch) {
4860                my $pool = SVN::Pool->new;
4861                my $rep = eval {
4862                        $self->do_switch(1, '', 0, $self->{url},
4863                                         SVN::Delta::Editor->new, $pool);
4864                };
4865                if ($@) {
4866                        $can_do_switch = 0;
4867                } else {
4868                        $rep->abort_report($pool);
4869                        $can_do_switch = 1;
4870                }
4871                $pool->clear;
4872        }
4873        $can_do_switch;
4874}
4875
4876sub skip_unknown_revs {
4877        my ($err) = @_;
4878        my $errno = $err->apr_err();
4879        # Maybe the branch we're tracking didn't
4880        # exist when the repo started, so it's
4881        # not an error if it doesn't, just continue
4882        #
4883        # Wonderfully consistent library, eh?
4884        # 160013 - svn:// and file://
4885        # 175002 - http(s)://
4886        # 175007 - http(s):// (this repo required authorization, too...)
4887        #   More codes may be discovered later...
4888        if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
4889                my $err_key = $err->expanded_message;
4890                # revision numbers change every time, filter them out
4891                $err_key =~ s/\d+/\0/g;
4892                $err_key = "$errno\0$err_key";
4893                unless ($ignored_err{$err_key}) {
4894                        warn "W: Ignoring error from SVN, path probably ",
4895                             "does not exist: ($errno): ",
4896                             $err->expanded_message,"\n";
4897                        warn "W: Do not be alarmed at the above message ",
4898                             "git-svn is just searching aggressively for ",
4899                             "old history.\n",
4900                             "This may take a while on large repositories\n";
4901                        $ignored_err{$err_key} = 1;
4902                }
4903                return;
4904        }
4905        die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
4906}
4907
4908package Git::SVN::Log;
4909use strict;
4910use warnings;
4911use POSIX qw/strftime/;
4912use Time::Local;
4913use constant commit_log_separator => ('-' x 72) . "\n";
4914use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
4915            %rusers $show_commit $incremental/;
4916my $l_fmt;
4917
4918sub cmt_showable {
4919        my ($c) = @_;
4920        return 1 if defined $c->{r};
4921
4922        # big commit message got truncated by the 16k pretty buffer in rev-list
4923        if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
4924                                $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
4925                @{$c->{l}} = ();
4926                my @log = command(qw/cat-file commit/, $c->{c});
4927
4928                # shift off the headers
4929                shift @log while ($log[0] ne '');
4930                shift @log;
4931
4932                # TODO: make $c->{l} not have a trailing newline in the future
4933                @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
4934
4935                (undef, $c->{r}, undef) = ::extract_metadata(
4936                                (grep(/^git-svn-id: /, @log))[-1]);
4937        }
4938        return defined $c->{r};
4939}
4940
4941sub log_use_color {
4942        return $color || Git->repository->get_colorbool('color.diff');
4943}
4944
4945sub git_svn_log_cmd {
4946        my ($r_min, $r_max, @args) = @_;
4947        my $head = 'HEAD';
4948        my (@files, @log_opts);
4949        foreach my $x (@args) {
4950                if ($x eq '--' || @files) {
4951                        push @files, $x;
4952                } else {
4953                        if (::verify_ref("$x^0")) {
4954                                $head = $x;
4955                        } else {
4956                                push @log_opts, $x;
4957                        }
4958                }
4959        }
4960
4961        my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
4962        $gs ||= Git::SVN->_new;
4963        my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
4964                   $gs->refname);
4965        push @cmd, '-r' unless $non_recursive;
4966        push @cmd, qw/--raw --name-status/ if $verbose;
4967        push @cmd, '--color' if log_use_color();
4968        push @cmd, @log_opts;
4969        if (defined $r_max && $r_max == $r_min) {
4970                push @cmd, '--max-count=1';
4971                if (my $c = $gs->rev_map_get($r_max)) {
4972                        push @cmd, $c;
4973                }
4974        } elsif (defined $r_max) {
4975                if ($r_max < $r_min) {
4976                        ($r_min, $r_max) = ($r_max, $r_min);
4977                }
4978                my (undef, $c_max) = $gs->find_rev_before($r_max, 1, $r_min);
4979                my (undef, $c_min) = $gs->find_rev_after($r_min, 1, $r_max);
4980                # If there are no commits in the range, both $c_max and $c_min
4981                # will be undefined.  If there is at least 1 commit in the
4982                # range, both will be defined.
4983                return () if !defined $c_min || !defined $c_max;
4984                if ($c_min eq $c_max) {
4985                        push @cmd, '--max-count=1', $c_min;
4986                } else {
4987                        push @cmd, '--boundary', "$c_min..$c_max";
4988                }
4989        }
4990        return (@cmd, @files);
4991}
4992
4993# adapted from pager.c
4994sub config_pager {
4995        $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
4996        if (!defined $pager) {
4997                $pager = 'less';
4998        } elsif (length $pager == 0 || $pager eq 'cat') {
4999                $pager = undef;
5000        }
5001        $ENV{GIT_PAGER_IN_USE} = defined($pager);
5002}
5003
5004sub run_pager {
5005        return unless -t *STDOUT && defined $pager;
5006        pipe my ($rfd, $wfd) or return;
5007        defined(my $pid = fork) or ::fatal "Can't fork: $!";
5008        if (!$pid) {
5009                open STDOUT, '>&', $wfd or
5010                                     ::fatal "Can't redirect to stdout: $!";
5011                return;
5012        }
5013        open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!";
5014        $ENV{LESS} ||= 'FRSX';
5015        exec $pager or ::fatal "Can't run pager: $! ($pager)";
5016}
5017
5018sub format_svn_date {
5019        # some systmes don't handle or mishandle %z, so be creative.
5020        my $t = shift || time;
5021        my $gm = timelocal(gmtime($t));
5022        my $sign = qw( + + - )[ $t <=> $gm ];
5023        my $gmoff = sprintf("%s%02d%02d", $sign, (gmtime(abs($t - $gm)))[2,1]);
5024        return strftime("%Y-%m-%d %H:%M:%S $gmoff (%a, %d %b %Y)", localtime($t));
5025}
5026
5027sub parse_git_date {
5028        my ($t, $tz) = @_;
5029        # Date::Parse isn't in the standard Perl distro :(
5030        if ($tz =~ s/^\+//) {
5031                $t += tz_to_s_offset($tz);
5032        } elsif ($tz =~ s/^\-//) {
5033                $t -= tz_to_s_offset($tz);
5034        }
5035        return $t;
5036}
5037
5038sub set_local_timezone {
5039        if (defined $TZ) {
5040                $ENV{TZ} = $TZ;
5041        } else {
5042                delete $ENV{TZ};
5043        }
5044}
5045
5046sub tz_to_s_offset {
5047        my ($tz) = @_;
5048        $tz =~ s/(\d\d)$//;
5049        return ($1 * 60) + ($tz * 3600);
5050}
5051
5052sub get_author_info {
5053        my ($dest, $author, $t, $tz) = @_;
5054        $author =~ s/(?:^\s*|\s*$)//g;
5055        $dest->{a_raw} = $author;
5056        my $au;
5057        if ($::_authors) {
5058                $au = $rusers{$author} || undef;
5059        }
5060        if (!$au) {
5061                ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
5062        }
5063        $dest->{t} = $t;
5064        $dest->{tz} = $tz;
5065        $dest->{a} = $au;
5066        $dest->{t_utc} = parse_git_date($t, $tz);
5067}
5068
5069sub process_commit {
5070        my ($c, $r_min, $r_max, $defer) = @_;
5071        if (defined $r_min && defined $r_max) {
5072                if ($r_min == $c->{r} && $r_min == $r_max) {
5073                        show_commit($c);
5074                        return 0;
5075                }
5076                return 1 if $r_min == $r_max;
5077                if ($r_min < $r_max) {
5078                        # we need to reverse the print order
5079                        return 0 if (defined $limit && --$limit < 0);
5080                        push @$defer, $c;
5081                        return 1;
5082                }
5083                if ($r_min != $r_max) {
5084                        return 1 if ($r_min < $c->{r});
5085                        return 1 if ($r_max > $c->{r});
5086                }
5087        }
5088        return 0 if (defined $limit && --$limit < 0);
5089        show_commit($c);
5090        return 1;
5091}
5092
5093sub show_commit {
5094        my $c = shift;
5095        if ($oneline) {
5096                my $x = "\n";
5097                if (my $l = $c->{l}) {
5098                        while ($l->[0] =~ /^\s*$/) { shift @$l }
5099                        $x = $l->[0];
5100                }
5101                $l_fmt ||= 'A' . length($c->{r});
5102                print 'r',pack($l_fmt, $c->{r}),' | ';
5103                print "$c->{c} | " if $show_commit;
5104                print $x;
5105        } else {
5106                show_commit_normal($c);
5107        }
5108}
5109
5110sub show_commit_changed_paths {
5111        my ($c) = @_;
5112        return unless $c->{changed};
5113        print "Changed paths:\n", @{$c->{changed}};
5114}
5115
5116sub show_commit_normal {
5117        my ($c) = @_;
5118        print commit_log_separator, "r$c->{r} | ";
5119        print "$c->{c} | " if $show_commit;
5120        print "$c->{a} | ", format_svn_date($c->{t_utc}), ' | ';
5121        my $nr_line = 0;
5122
5123        if (my $l = $c->{l}) {
5124                while ($l->[$#$l] eq "\n" && $#$l > 0
5125                                          && $l->[($#$l - 1)] eq "\n") {
5126                        pop @$l;
5127                }
5128                $nr_line = scalar @$l;
5129                if (!$nr_line) {
5130                        print "1 line\n\n\n";
5131                } else {
5132                        if ($nr_line == 1) {
5133                                $nr_line = '1 line';
5134                        } else {
5135                                $nr_line .= ' lines';
5136                        }
5137                        print $nr_line, "\n";
5138                        show_commit_changed_paths($c);
5139                        print "\n";
5140                        print $_ foreach @$l;
5141                }
5142        } else {
5143                print "1 line\n";
5144                show_commit_changed_paths($c);
5145                print "\n";
5146
5147        }
5148        foreach my $x (qw/raw stat diff/) {
5149                if ($c->{$x}) {
5150                        print "\n";
5151                        print $_ foreach @{$c->{$x}}
5152                }
5153        }
5154}
5155
5156sub cmd_show_log {
5157        my (@args) = @_;
5158        my ($r_min, $r_max);
5159        my $r_last = -1; # prevent dupes
5160        set_local_timezone();
5161        if (defined $::_revision) {
5162                if ($::_revision =~ /^(\d+):(\d+)$/) {
5163                        ($r_min, $r_max) = ($1, $2);
5164                } elsif ($::_revision =~ /^\d+$/) {
5165                        $r_min = $r_max = $::_revision;
5166                } else {
5167                        ::fatal "-r$::_revision is not supported, use ",
5168                                "standard 'git log' arguments instead";
5169                }
5170        }
5171
5172        config_pager();
5173        @args = git_svn_log_cmd($r_min, $r_max, @args);
5174        if (!@args) {
5175                print commit_log_separator unless $incremental || $oneline;
5176                return;
5177        }
5178        my $log = command_output_pipe(@args);
5179        run_pager();
5180        my (@k, $c, $d, $stat);
5181        my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
5182        while (<$log>) {
5183                if (/^${esc_color}commit -?($::sha1_short)/o) {
5184                        my $cmt = $1;
5185                        if ($c && cmt_showable($c) && $c->{r} != $r_last) {
5186                                $r_last = $c->{r};
5187                                process_commit($c, $r_min, $r_max, \@k) or
5188                                                                goto out;
5189                        }
5190                        $d = undef;
5191                        $c = { c => $cmt };
5192                } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
5193                        get_author_info($c, $1, $2, $3);
5194                } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
5195                        # ignore
5196                } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
5197                        push @{$c->{raw}}, $_;
5198                } elsif (/^${esc_color}[ACRMDT]\t/) {
5199                        # we could add $SVN->{svn_path} here, but that requires
5200                        # remote access at the moment (repo_path_split)...
5201                        s#^(${esc_color})([ACRMDT])\t#$1   $2 #o;
5202                        push @{$c->{changed}}, $_;
5203                } elsif (/^${esc_color}diff /o) {
5204                        $d = 1;
5205                        push @{$c->{diff}}, $_;
5206                } elsif ($d) {
5207                        push @{$c->{diff}}, $_;
5208                } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
5209                          $esc_color*[\+\-]*$esc_color$/x) {
5210                        $stat = 1;
5211                        push @{$c->{stat}}, $_;
5212                } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
5213                        push @{$c->{stat}}, $_;
5214                        $stat = undef;
5215                } elsif (/^${esc_color}    (git-svn-id:.+)$/o) {
5216                        ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
5217                } elsif (s/^${esc_color}    //o) {
5218                        push @{$c->{l}}, $_;
5219                }
5220        }
5221        if ($c && defined $c->{r} && $c->{r} != $r_last) {
5222                $r_last = $c->{r};
5223                process_commit($c, $r_min, $r_max, \@k);
5224        }
5225        if (@k) {
5226                ($r_min, $r_max) = ($r_max, $r_min);
5227                process_commit($_, $r_min, $r_max) foreach reverse @k;
5228        }
5229out:
5230        close $log;
5231        print commit_log_separator unless $incremental || $oneline;
5232}
5233
5234sub cmd_blame {
5235        my $path = pop;
5236
5237        config_pager();
5238        run_pager();
5239
5240        my ($fh, $ctx, $rev);
5241
5242        if ($_git_format) {
5243                ($fh, $ctx) = command_output_pipe('blame', @_, $path);
5244                while (my $line = <$fh>) {
5245                        if ($line =~ /^\^?([[:xdigit:]]+)\s/) {
5246                                # Uncommitted edits show up as a rev ID of
5247                                # all zeros, which we can't look up with
5248                                # cmt_metadata
5249                                if ($1 !~ /^0+$/) {
5250                                        (undef, $rev, undef) =
5251                                                ::cmt_metadata($1);
5252                                        $rev = '0' if (!$rev);
5253                                } else {
5254                                        $rev = '0';
5255                                }
5256                                $rev = sprintf('%-10s', $rev);
5257                                $line =~ s/^\^?[[:xdigit:]]+(\s)/$rev$1/;
5258                        }
5259                        print $line;
5260                }
5261        } else {
5262                ($fh, $ctx) = command_output_pipe('blame', '-p', @_, 'HEAD',
5263                                                  '--', $path);
5264                my ($sha1);
5265                my %authors;
5266                my @buffer;
5267                my %dsha; #distinct sha keys
5268
5269                while (my $line = <$fh>) {
5270                        push @buffer, $line;
5271                        if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5272                                $dsha{$1} = 1;
5273                        }
5274                }
5275
5276                my $s2r = ::cmt_sha2rev_batch([keys %dsha]);
5277
5278                foreach my $line (@buffer) {
5279                        if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5280                                $rev = $s2r->{$1};
5281                                $rev = '0' if (!$rev)
5282                        }
5283                        elsif ($line =~ /^author (.*)/) {
5284                                $authors{$rev} = $1;
5285                                $authors{$rev} =~ s/\s/_/g;
5286                        }
5287                        elsif ($line =~ /^\t(.*)$/) {
5288                                printf("%6s %10s %s\n", $rev, $authors{$rev}, $1);
5289                        }
5290                }
5291        }
5292        command_close_pipe($fh, $ctx);
5293}
5294
5295package Git::SVN::Migration;
5296# these version numbers do NOT correspond to actual version numbers
5297# of git nor git-svn.  They are just relative.
5298#
5299# v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
5300#
5301# v1 layout: .git/$id/info/url, refs/remotes/$id
5302#
5303# v2 layout: .git/svn/$id/info/url, refs/remotes/$id
5304#
5305# v3 layout: .git/svn/$id, refs/remotes/$id
5306#            - info/url may remain for backwards compatibility
5307#            - this is what we migrate up to this layout automatically,
5308#            - this will be used by git svn init on single branches
5309# v3.1 layout (auto migrated):
5310#            - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
5311#              for backwards compatibility
5312#
5313# v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
5314#            - this is only created for newly multi-init-ed
5315#              repositories.  Similar in spirit to the
5316#              --use-separate-remotes option in git-clone (now default)
5317#            - we do not automatically migrate to this (following
5318#              the example set by core git)
5319#
5320# v5 layout: .rev_db.$UUID => .rev_map.$UUID
5321#            - newer, more-efficient format that uses 24-bytes per record
5322#              with no filler space.
5323#            - use xxd -c24 < .rev_map.$UUID to view and debug
5324#            - This is a one-way migration, repositories updated to the
5325#              new format will not be able to use old git-svn without
5326#              rebuilding the .rev_db.  Rebuilding the rev_db is not
5327#              possible if noMetadata or useSvmProps are set; but should
5328#              be no problem for users that use the (sensible) defaults.
5329use strict;
5330use warnings;
5331use Carp qw/croak/;
5332use File::Path qw/mkpath/;
5333use File::Basename qw/dirname basename/;
5334use vars qw/$_minimize/;
5335
5336sub migrate_from_v0 {
5337        my $git_dir = $ENV{GIT_DIR};
5338        return undef unless -d $git_dir;
5339        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5340        my $migrated = 0;
5341        while (<$fh>) {
5342                chomp;
5343                my ($id, $orig_ref) = ($_, $_);
5344                next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
5345                next unless -f "$git_dir/$id/info/url";
5346                my $new_ref = "refs/remotes/$id";
5347                if (::verify_ref("$new_ref^0")) {
5348                        print STDERR "W: $orig_ref is probably an old ",
5349                                     "branch used by an ancient version of ",
5350                                     "git-svn.\n",
5351                                     "However, $new_ref also exists.\n",
5352                                     "We will not be able ",
5353                                     "to use this branch until this ",
5354                                     "ambiguity is resolved.\n";
5355                        next;
5356                }
5357                print STDERR "Migrating from v0 layout...\n" if !$migrated;
5358                print STDERR "Renaming ref: $orig_ref => $new_ref\n";
5359                command_noisy('update-ref', $new_ref, $orig_ref);
5360                command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
5361                $migrated++;
5362        }
5363        command_close_pipe($fh, $ctx);
5364        print STDERR "Done migrating from v0 layout...\n" if $migrated;
5365        $migrated;
5366}
5367
5368sub migrate_from_v1 {
5369        my $git_dir = $ENV{GIT_DIR};
5370        my $migrated = 0;
5371        return $migrated unless -d $git_dir;
5372        my $svn_dir = "$git_dir/svn";
5373
5374        # just in case somebody used 'svn' as their $id at some point...
5375        return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
5376
5377        print STDERR "Migrating from a git-svn v1 layout...\n";
5378        mkpath([$svn_dir]);
5379        print STDERR "Data from a previous version of git-svn exists, but\n\t",
5380                     "$svn_dir\n\t(required for this version ",
5381                     "($::VERSION) of git-svn) does not exist.\n";
5382        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5383        while (<$fh>) {
5384                my $x = $_;
5385                next unless $x =~ s#^refs/remotes/##;
5386                chomp $x;
5387                next unless -f "$git_dir/$x/info/url";
5388                my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
5389                next unless $u;
5390                my $dn = dirname("$git_dir/svn/$x");
5391                mkpath([$dn]) unless -d $dn;
5392                if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
5393                        mkpath(["$git_dir/svn/svn"]);
5394                        print STDERR " - $git_dir/$x/info => ",
5395                                        "$git_dir/svn/$x/info\n";
5396                        rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
5397                               croak "$!: $x";
5398                        # don't worry too much about these, they probably
5399                        # don't exist with repos this old (save for index,
5400                        # and we can easily regenerate that)
5401                        foreach my $f (qw/unhandled.log index .rev_db/) {
5402                                rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
5403                        }
5404                } else {
5405                        print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
5406                        rename "$git_dir/$x", "$git_dir/svn/$x" or
5407                               croak "$!: $x";
5408                }
5409                $migrated++;
5410        }
5411        command_close_pipe($fh, $ctx);
5412        print STDERR "Done migrating from a git-svn v1 layout\n";
5413        $migrated;
5414}
5415
5416sub read_old_urls {
5417        my ($l_map, $pfx, $path) = @_;
5418        my @dir;
5419        foreach (<$path/*>) {
5420                if (-r "$_/info/url") {
5421                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
5422                        my $ref_id = $pfx . basename $_;
5423                        my $url = ::file_to_s("$_/info/url");
5424                        $l_map->{$ref_id} = $url;
5425                } elsif (-d $_) {
5426                        push @dir, $_;
5427                }
5428        }
5429        foreach (@dir) {
5430                my $x = $_;
5431                $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
5432                read_old_urls($l_map, $x, $_);
5433        }
5434}
5435
5436sub migrate_from_v2 {
5437        my @cfg = command(qw/config -l/);
5438        return if grep /^svn-remote\..+\.url=/, @cfg;
5439        my %l_map;
5440        read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
5441        my $migrated = 0;
5442
5443        foreach my $ref_id (sort keys %l_map) {
5444                eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
5445                if ($@) {
5446                        Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
5447                }
5448                $migrated++;
5449        }
5450        $migrated;
5451}
5452
5453sub minimize_connections {
5454        my $r = Git::SVN::read_all_remotes();
5455        my $new_urls = {};
5456        my $root_repos = {};
5457        foreach my $repo_id (keys %$r) {
5458                my $url = $r->{$repo_id}->{url} or next;
5459                my $fetch = $r->{$repo_id}->{fetch} or next;
5460                my $ra = Git::SVN::Ra->new($url);
5461
5462                # skip existing cases where we already connect to the root
5463                if (($ra->{url} eq $ra->{repos_root}) ||
5464                    ($ra->{repos_root} eq $repo_id)) {
5465                        $root_repos->{$ra->{url}} = $repo_id;
5466                        next;
5467                }
5468
5469                my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
5470                my $root_path = $ra->{url};
5471                $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
5472                foreach my $path (keys %$fetch) {
5473                        my $ref_id = $fetch->{$path};
5474                        my $gs = Git::SVN->new($ref_id, $repo_id, $path);
5475
5476                        # make sure we can read when connecting to
5477                        # a higher level of a repository
5478                        my ($last_rev, undef) = $gs->last_rev_commit;
5479                        if (!defined $last_rev) {
5480                                $last_rev = eval {
5481                                        $root_ra->get_latest_revnum;
5482                                };
5483                                next if $@;
5484                        }
5485                        my $new = $root_path;
5486                        $new .= length $path ? "/$path" : '';
5487                        eval {
5488                                $root_ra->get_log([$new], $last_rev, $last_rev,
5489                                                  0, 0, 1, sub { });
5490                        };
5491                        next if $@;
5492                        $new_urls->{$ra->{repos_root}}->{$new} =
5493                                { ref_id => $ref_id,
5494                                  old_repo_id => $repo_id,
5495                                  old_path => $path };
5496                }
5497        }
5498
5499        my @emptied;
5500        foreach my $url (keys %$new_urls) {
5501                # see if we can re-use an existing [svn-remote "repo_id"]
5502                # instead of creating a(n ugly) new section:
5503                my $repo_id = $root_repos->{$url} || $url;
5504
5505                my $fetch = $new_urls->{$url};
5506                foreach my $path (keys %$fetch) {
5507                        my $x = $fetch->{$path};
5508                        Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
5509                        my $pfx = "svn-remote.$x->{old_repo_id}";
5510
5511                        my $old_fetch = quotemeta("$x->{old_path}:".
5512                                                  "refs/remotes/$x->{ref_id}");
5513                        command_noisy(qw/config --unset/,
5514                                      "$pfx.fetch", '^'. $old_fetch . '$');
5515                        delete $r->{$x->{old_repo_id}}->
5516                               {fetch}->{$x->{old_path}};
5517                        if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
5518                                command_noisy(qw/config --unset/,
5519                                              "$pfx.url");
5520                                push @emptied, $x->{old_repo_id}
5521                        }
5522                }
5523        }
5524        if (@emptied) {
5525                my $file = $ENV{GIT_CONFIG} || "$ENV{GIT_DIR}/config";
5526                print STDERR <<EOF;
5527The following [svn-remote] sections in your config file ($file) are empty
5528and can be safely removed:
5529EOF
5530                print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
5531        }
5532}
5533
5534sub migration_check {
5535        migrate_from_v0();
5536        migrate_from_v1();
5537        migrate_from_v2();
5538        minimize_connections() if $_minimize;
5539}
5540
5541package Git::IndexInfo;
5542use strict;
5543use warnings;
5544use Git qw/command_input_pipe command_close_pipe/;
5545
5546sub new {
5547        my ($class) = @_;
5548        my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
5549        bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
5550}
5551
5552sub remove {
5553        my ($self, $path) = @_;
5554        if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
5555                return ++$self->{nr};
5556        }
5557        undef;
5558}
5559
5560sub update {
5561        my ($self, $mode, $hash, $path) = @_;
5562        if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
5563                return ++$self->{nr};
5564        }
5565        undef;
5566}
5567
5568sub DESTROY {
5569        my ($self) = @_;
5570        command_close_pipe($self->{gui}, $self->{ctx});
5571}
5572
5573package Git::SVN::GlobSpec;
5574use strict;
5575use warnings;
5576
5577sub new {
5578        my ($class, $glob) = @_;
5579        my $re = $glob;
5580        $re =~ s!/+$!!g; # no need for trailing slashes
5581        $re =~ m!^([^*]*)(\*(?:/\*)*)([^*]*)$!;
5582        my $temp = $re;
5583        my ($left, $right) = ($1, $3);
5584        $re = $2;
5585        my $depth = $re =~ tr/*/*/;
5586        if ($depth != $temp =~ tr/*/*/) {
5587                die "Only one set of wildcard directories " .
5588                        "(e.g. '*' or '*/*/*') is supported: '$glob'\n";
5589        }
5590        if ($depth == 0) {
5591                die "One '*' is needed for glob: '$glob'\n";
5592        }
5593        $re =~ s!\*!\[^/\]*!g;
5594        $re = quotemeta($left) . "($re)" . quotemeta($right);
5595        if (length $left && !($left =~ s!/+$!!g)) {
5596                die "Missing trailing '/' on left side of: '$glob' ($left)\n";
5597        }
5598        if (length $right && !($right =~ s!^/+!!g)) {
5599                die "Missing leading '/' on right side of: '$glob' ($right)\n";
5600        }
5601        my $left_re = qr/^\/\Q$left\E(\/|$)/;
5602        bless { left => $left, right => $right, left_regex => $left_re,
5603                regex => qr/$re/, glob => $glob, depth => $depth }, $class;
5604}
5605
5606sub full_path {
5607        my ($self, $path) = @_;
5608        return (length $self->{left} ? "$self->{left}/" : '') .
5609               $path . (length $self->{right} ? "/$self->{right}" : '');
5610}
5611
5612__END__
5613
5614Data structures:
5615
5616
5617$remotes = { # returned by read_all_remotes()
5618        'svn' => {
5619                # svn-remote.svn.url=https://svn.musicpd.org
5620                url => 'https://svn.musicpd.org',
5621                # svn-remote.svn.fetch=mpd/trunk:trunk
5622                fetch => {
5623                        'mpd/trunk' => 'trunk',
5624                },
5625                # svn-remote.svn.tags=mpd/tags/*:tags/*
5626                tags => {
5627                        path => {
5628                                left => 'mpd/tags',
5629                                right => '',
5630                                regex => qr!mpd/tags/([^/]+)$!,
5631                                glob => 'tags/*',
5632                        },
5633                        ref => {
5634                                left => 'tags',
5635                                right => '',
5636                                regex => qr!tags/([^/]+)$!,
5637                                glob => 'tags/*',
5638                        },
5639                }
5640        }
5641};
5642
5643$log_entry hashref as returned by libsvn_log_entry()
5644{
5645        log => 'whitespace-formatted log entry
5646',                                              # trailing newline is preserved
5647        revision => '8',                        # integer
5648        date => '2004-02-24T17:01:44.108345Z',  # commit date
5649        author => 'committer name'
5650};
5651
5652
5653# this is generated by generate_diff();
5654@mods = array of diff-index line hashes, each element represents one line
5655        of diff-index output
5656
5657diff-index line ($m hash)
5658{
5659        mode_a => first column of diff-index output, no leading ':',
5660        mode_b => second column of diff-index output,
5661        sha1_b => sha1sum of the final blob,
5662        chg => change type [MCRADT],
5663        file_a => original file name of a file (iff chg is 'C' or 'R')
5664        file_b => new/current file name of a file (any chg)
5665}
5666;
5667
5668# retval of read_url_paths{,_all}();
5669$l_map = {
5670        # repository root url
5671        'https://svn.musicpd.org' => {
5672                # repository path               # GIT_SVN_ID
5673                'mpd/trunk'             =>      'trunk',
5674                'mpd/tags/0.11.5'       =>      'tags/0.11.5',
5675        },
5676}
5677
5678Notes:
5679        I don't trust the each() function on unless I created %hash myself
5680        because the internal iterator may not have started at base.