5515e3ea549fd16df5f851b8b3ffc1e7cc1e2c22
   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 = 'refs/remotes/' . $_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{%([0-9A-F]{2})}{chr hex($1)}ieg;
1226        $remote_path =~ s#/+#/#g;
1227        $remote_path =~ s#^/##g;
1228        $remote_path .= "/*" if $remote_path !~ /\*/;
1229        my ($n) = ($switch =~ /^--(\w+)/);
1230        if (length $pfx && $pfx !~ m#/$#) {
1231                die "--prefix='$pfx' must have a trailing slash '/'\n";
1232        }
1233        command_noisy('config',
1234                      '--add',
1235                      "svn-remote.$gs->{repo_id}.$n",
1236                      "$remote_path:refs/remotes/$pfx*" .
1237                        ('/*' x (($remote_path =~ tr/*/*/) - 1)) );
1238}
1239
1240sub verify_ref {
1241        my ($ref) = @_;
1242        eval { command_oneline([ 'rev-parse', '--verify', $ref ],
1243                               { STDERR => 0 }); };
1244}
1245
1246sub get_tree_from_treeish {
1247        my ($treeish) = @_;
1248        # $treeish can be a symbolic ref, too:
1249        my $type = command_oneline(qw/cat-file -t/, $treeish);
1250        my $expected;
1251        while ($type eq 'tag') {
1252                ($treeish, $type) = command(qw/cat-file tag/, $treeish);
1253        }
1254        if ($type eq 'commit') {
1255                $expected = (grep /^tree /, command(qw/cat-file commit/,
1256                                                    $treeish))[0];
1257                ($expected) = ($expected =~ /^tree ($sha1)$/o);
1258                die "Unable to get tree from $treeish\n" unless $expected;
1259        } elsif ($type eq 'tree') {
1260                $expected = $treeish;
1261        } else {
1262                die "$treeish is a $type, expected tree, tag or commit\n";
1263        }
1264        return $expected;
1265}
1266
1267sub get_commit_entry {
1268        my ($treeish) = shift;
1269        my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
1270        my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
1271        my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
1272        open my $log_fh, '>', $commit_editmsg or croak $!;
1273
1274        my $type = command_oneline(qw/cat-file -t/, $treeish);
1275        if ($type eq 'commit' || $type eq 'tag') {
1276                my ($msg_fh, $ctx) = command_output_pipe('cat-file',
1277                                                         $type, $treeish);
1278                my $in_msg = 0;
1279                my $author;
1280                my $saw_from = 0;
1281                my $msgbuf = "";
1282                while (<$msg_fh>) {
1283                        if (!$in_msg) {
1284                                $in_msg = 1 if (/^\s*$/);
1285                                $author = $1 if (/^author (.*>)/);
1286                        } elsif (/^git-svn-id: /) {
1287                                # skip this for now, we regenerate the
1288                                # correct one on re-fetch anyways
1289                                # TODO: set *:merge properties or like...
1290                        } else {
1291                                if (/^From:/ || /^Signed-off-by:/) {
1292                                        $saw_from = 1;
1293                                }
1294                                $msgbuf .= $_;
1295                        }
1296                }
1297                $msgbuf =~ s/\s+$//s;
1298                if ($Git::SVN::_add_author_from && defined($author)
1299                    && !$saw_from) {
1300                        $msgbuf .= "\n\nFrom: $author";
1301                }
1302                print $log_fh $msgbuf or croak $!;
1303                command_close_pipe($msg_fh, $ctx);
1304        }
1305        close $log_fh or croak $!;
1306
1307        if ($_edit || ($type eq 'tree')) {
1308                my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
1309                # TODO: strip out spaces, comments, like git-commit.sh
1310                system($editor, $commit_editmsg);
1311        }
1312        rename $commit_editmsg, $commit_msg or croak $!;
1313        {
1314                require Encode;
1315                # SVN requires messages to be UTF-8 when entering the repo
1316                local $/;
1317                open $log_fh, '<', $commit_msg or croak $!;
1318                binmode $log_fh;
1319                chomp($log_entry{log} = <$log_fh>);
1320
1321                my $enc = Git::config('i18n.commitencoding') || 'UTF-8';
1322                my $msg = $log_entry{log};
1323
1324                eval { $msg = Encode::decode($enc, $msg, 1) };
1325                if ($@) {
1326                        die "Could not decode as $enc:\n", $msg,
1327                            "\nPerhaps you need to set i18n.commitencoding\n";
1328                }
1329
1330                eval { $msg = Encode::encode('UTF-8', $msg, 1) };
1331                die "Could not encode as UTF-8:\n$msg\n" if $@;
1332
1333                $log_entry{log} = $msg;
1334
1335                close $log_fh or croak $!;
1336        }
1337        unlink $commit_msg;
1338        \%log_entry;
1339}
1340
1341sub s_to_file {
1342        my ($str, $file, $mode) = @_;
1343        open my $fd,'>',$file or croak $!;
1344        print $fd $str,"\n" or croak $!;
1345        close $fd or croak $!;
1346        chmod ($mode &~ umask, $file) if (defined $mode);
1347}
1348
1349sub file_to_s {
1350        my $file = shift;
1351        open my $fd,'<',$file or croak "$!: file: $file\n";
1352        local $/;
1353        my $ret = <$fd>;
1354        close $fd or croak $!;
1355        $ret =~ s/\s*$//s;
1356        return $ret;
1357}
1358
1359# '<svn username> = real-name <email address>' mapping based on git-svnimport:
1360sub load_authors {
1361        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1362        my $log = $cmd eq 'log';
1363        while (<$authors>) {
1364                chomp;
1365                next unless /^(.+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
1366                my ($user, $name, $email) = ($1, $2, $3);
1367                if ($log) {
1368                        $Git::SVN::Log::rusers{"$name <$email>"} = $user;
1369                } else {
1370                        $users{$user} = [$name, $email];
1371                }
1372        }
1373        close $authors or croak $!;
1374}
1375
1376# convert GetOpt::Long specs for use by git-config
1377sub read_repo_config {
1378        return unless -d $ENV{GIT_DIR};
1379        my $opts = shift;
1380        my @config_only;
1381        foreach my $o (keys %$opts) {
1382                # if we have mixedCase and a long option-only, then
1383                # it's a config-only variable that we don't need for
1384                # the command-line.
1385                push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
1386                my $v = $opts->{$o};
1387                my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
1388                $key =~ s/-//g;
1389                my $arg = 'git config';
1390                $arg .= ' --int' if ($o =~ /[:=]i$/);
1391                $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
1392                if (ref $v eq 'ARRAY') {
1393                        chomp(my @tmp = `$arg --get-all svn.$key`);
1394                        @$v = @tmp if @tmp;
1395                } else {
1396                        chomp(my $tmp = `$arg --get svn.$key`);
1397                        if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
1398                                $$v = $tmp;
1399                        }
1400                }
1401        }
1402        delete @$opts{@config_only} if @config_only;
1403}
1404
1405sub extract_metadata {
1406        my $id = shift or return (undef, undef, undef);
1407        my ($url, $rev, $uuid) = ($id =~ /^\s*git-svn-id:\s+(.*)\@(\d+)
1408                                                        \s([a-f\d\-]+)$/ix);
1409        if (!defined $rev || !$uuid || !$url) {
1410                # some of the original repositories I made had
1411                # identifiers like this:
1412                ($rev, $uuid) = ($id =~/^\s*git-svn-id:\s(\d+)\@([a-f\d\-]+)/i);
1413        }
1414        return ($url, $rev, $uuid);
1415}
1416
1417sub cmt_metadata {
1418        return extract_metadata((grep(/^git-svn-id: /,
1419                command(qw/cat-file commit/, shift)))[-1]);
1420}
1421
1422sub cmt_sha2rev_batch {
1423        my %s2r;
1424        my ($pid, $in, $out, $ctx) = command_bidi_pipe(qw/cat-file --batch/);
1425        my $list = shift;
1426
1427        foreach my $sha (@{$list}) {
1428                my $first = 1;
1429                my $size = 0;
1430                print $out $sha, "\n";
1431
1432                while (my $line = <$in>) {
1433                        if ($first && $line =~ /^[[:xdigit:]]{40}\smissing$/) {
1434                                last;
1435                        } elsif ($first &&
1436                               $line =~ /^[[:xdigit:]]{40}\scommit\s(\d+)$/) {
1437                                $first = 0;
1438                                $size = $1;
1439                                next;
1440                        } elsif ($line =~ /^(git-svn-id: )/) {
1441                                my (undef, $rev, undef) =
1442                                                      extract_metadata($line);
1443                                $s2r{$sha} = $rev;
1444                        }
1445
1446                        $size -= length($line);
1447                        last if ($size == 0);
1448                }
1449        }
1450
1451        command_close_bidi_pipe($pid, $in, $out, $ctx);
1452
1453        return \%s2r;
1454}
1455
1456sub working_head_info {
1457        my ($head, $refs) = @_;
1458        my @args = ('log', '--no-color', '--first-parent', '--pretty=medium');
1459        my ($fh, $ctx) = command_output_pipe(@args, $head);
1460        my $hash;
1461        my %max;
1462        while (<$fh>) {
1463                if ( m{^commit ($::sha1)$} ) {
1464                        unshift @$refs, $hash if $hash and $refs;
1465                        $hash = $1;
1466                        next;
1467                }
1468                next unless s{^\s*(git-svn-id:)}{$1};
1469                my ($url, $rev, $uuid) = extract_metadata($_);
1470                if (defined $url && defined $rev) {
1471                        next if $max{$url} and $max{$url} < $rev;
1472                        if (my $gs = Git::SVN->find_by_url($url)) {
1473                                my $c = $gs->rev_map_get($rev, $uuid);
1474                                if ($c && $c eq $hash) {
1475                                        close $fh; # break the pipe
1476                                        return ($url, $rev, $uuid, $gs);
1477                                } else {
1478                                        $max{$url} ||= $gs->rev_map_max;
1479                                }
1480                        }
1481                }
1482        }
1483        command_close_pipe($fh, $ctx);
1484        (undef, undef, undef, undef);
1485}
1486
1487sub read_commit_parents {
1488        my ($parents, $c) = @_;
1489        chomp(my $p = command_oneline(qw/rev-list --parents -1/, $c));
1490        $p =~ s/^($c)\s*// or die "rev-list --parents -1 $c failed!\n";
1491        @{$parents->{$c}} = split(/ /, $p);
1492}
1493
1494sub linearize_history {
1495        my ($gs, $refs) = @_;
1496        my %parents;
1497        foreach my $c (@$refs) {
1498                read_commit_parents(\%parents, $c);
1499        }
1500
1501        my @linear_refs;
1502        my %skip = ();
1503        my $last_svn_commit = $gs->last_commit;
1504        foreach my $c (reverse @$refs) {
1505                next if $c eq $last_svn_commit;
1506                last if $skip{$c};
1507
1508                unshift @linear_refs, $c;
1509                $skip{$c} = 1;
1510
1511                # we only want the first parent to diff against for linear
1512                # history, we save the rest to inject when we finalize the
1513                # svn commit
1514                my $fp_a = verify_ref("$c~1");
1515                my $fp_b = shift @{$parents{$c}} if $parents{$c};
1516                if (!$fp_a || !$fp_b) {
1517                        die "Commit $c\n",
1518                            "has no parent commit, and therefore ",
1519                            "nothing to diff against.\n",
1520                            "You should be working from a repository ",
1521                            "originally created by git-svn\n";
1522                }
1523                if ($fp_a ne $fp_b) {
1524                        die "$c~1 = $fp_a, however parsing commit $c ",
1525                            "revealed that:\n$c~1 = $fp_b\nBUG!\n";
1526                }
1527
1528                foreach my $p (@{$parents{$c}}) {
1529                        $skip{$p} = 1;
1530                }
1531        }
1532        (\@linear_refs, \%parents);
1533}
1534
1535sub find_file_type_and_diff_status {
1536        my ($path) = @_;
1537        return ('dir', '') if $path eq '';
1538
1539        my $diff_output =
1540            command_oneline(qw(diff --cached --name-status --), $path) || "";
1541        my $diff_status = (split(' ', $diff_output))[0] || "";
1542
1543        my $ls_tree = command_oneline(qw(ls-tree HEAD), $path) || "";
1544
1545        return (undef, undef) if !$diff_status && !$ls_tree;
1546
1547        if ($diff_status eq "A") {
1548                return ("link", $diff_status) if -l $path;
1549                return ("dir", $diff_status) if -d $path;
1550                return ("file", $diff_status);
1551        }
1552
1553        my $mode = (split(' ', $ls_tree))[0] || "";
1554
1555        return ("link", $diff_status) if $mode eq "120000";
1556        return ("dir", $diff_status) if $mode eq "040000";
1557        return ("file", $diff_status);
1558}
1559
1560sub md5sum {
1561        my $arg = shift;
1562        my $ref = ref $arg;
1563        my $md5 = Digest::MD5->new();
1564        if ($ref eq 'GLOB' || $ref eq 'IO::File' || $ref eq 'File::Temp') {
1565                $md5->addfile($arg) or croak $!;
1566        } elsif ($ref eq 'SCALAR') {
1567                $md5->add($$arg) or croak $!;
1568        } elsif (!$ref) {
1569                $md5->add($arg) or croak $!;
1570        } else {
1571                ::fatal "Can't provide MD5 hash for unknown ref type: '", $ref, "'";
1572        }
1573        return $md5->hexdigest();
1574}
1575
1576sub gc_directory {
1577        if ($can_compress && -f $_ && basename($_) eq "unhandled.log") {
1578                my $out_filename = $_ . ".gz";
1579                open my $in_fh, "<", $_ or die "Unable to open $_: $!\n";
1580                binmode $in_fh;
1581                my $gz = Compress::Zlib::gzopen($out_filename, "ab") or
1582                                die "Unable to open $out_filename: $!\n";
1583
1584                my $res;
1585                while ($res = sysread($in_fh, my $str, 1024)) {
1586                        $gz->gzwrite($str) or
1587                                die "Unable to write: ".$gz->gzerror()."!\n";
1588                }
1589                unlink $_ or die "unlink $File::Find::name: $!\n";
1590        } elsif (-f $_ && basename($_) eq "index") {
1591                unlink $_ or die "unlink $_: $!\n";
1592        }
1593}
1594
1595package Git::SVN;
1596use strict;
1597use warnings;
1598use Fcntl qw/:DEFAULT :seek/;
1599use constant rev_map_fmt => 'NH40';
1600use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
1601            $_repack $_repack_flags $_use_svm_props $_head
1602            $_use_svnsync_props $no_reuse_existing $_minimize_url
1603            $_use_log_author $_add_author_from $_localtime/;
1604use Carp qw/croak/;
1605use File::Path qw/mkpath/;
1606use File::Copy qw/copy/;
1607use IPC::Open3;
1608
1609my ($_gc_nr, $_gc_period);
1610
1611# properties that we do not log:
1612my %SKIP_PROP;
1613BEGIN {
1614        %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
1615                                        svn:special svn:executable
1616                                        svn:entry:committed-rev
1617                                        svn:entry:last-author
1618                                        svn:entry:uuid
1619                                        svn:entry:committed-date/;
1620
1621        # some options are read globally, but can be overridden locally
1622        # per [svn-remote "..."] section.  Command-line options will *NOT*
1623        # override options set in an [svn-remote "..."] section
1624        no strict 'refs';
1625        for my $option (qw/follow_parent no_metadata use_svm_props
1626                           use_svnsync_props/) {
1627                my $key = $option;
1628                $key =~ tr/_//d;
1629                my $prop = "-$option";
1630                *$option = sub {
1631                        my ($self) = @_;
1632                        return $self->{$prop} if exists $self->{$prop};
1633                        my $k = "svn-remote.$self->{repo_id}.$key";
1634                        eval { command_oneline(qw/config --get/, $k) };
1635                        if ($@) {
1636                                $self->{$prop} = ${"Git::SVN::_$option"};
1637                        } else {
1638                                my $v = command_oneline(qw/config --bool/,$k);
1639                                $self->{$prop} = $v eq 'false' ? 0 : 1;
1640                        }
1641                        return $self->{$prop};
1642                }
1643        }
1644}
1645
1646
1647my (%LOCKFILES, %INDEX_FILES);
1648END {
1649        unlink keys %LOCKFILES if %LOCKFILES;
1650        unlink keys %INDEX_FILES if %INDEX_FILES;
1651}
1652
1653sub resolve_local_globs {
1654        my ($url, $fetch, $glob_spec) = @_;
1655        return unless defined $glob_spec;
1656        my $ref = $glob_spec->{ref};
1657        my $path = $glob_spec->{path};
1658        foreach (command(qw#for-each-ref --format=%(refname) refs/#)) {
1659                next unless m#^$ref->{regex}$#;
1660                my $p = $1;
1661                my $pathname = desanitize_refname($path->full_path($p));
1662                my $refname = desanitize_refname($ref->full_path($p));
1663                if (my $existing = $fetch->{$pathname}) {
1664                        if ($existing ne $refname) {
1665                                die "Refspec conflict:\n",
1666                                    "existing: $existing\n",
1667                                    " globbed: $refname\n";
1668                        }
1669                        my $u = (::cmt_metadata("$refname"))[0];
1670                        $u =~ s!^\Q$url\E(/|$)!! or die
1671                          "$refname: '$url' not found in '$u'\n";
1672                        if ($pathname ne $u) {
1673                                warn "W: Refspec glob conflict ",
1674                                     "(ref: $refname):\n",
1675                                     "expected path: $pathname\n",
1676                                     "    real path: $u\n",
1677                                     "Continuing ahead with $u\n";
1678                                next;
1679                        }
1680                } else {
1681                        $fetch->{$pathname} = $refname;
1682                }
1683        }
1684}
1685
1686sub parse_revision_argument {
1687        my ($base, $head) = @_;
1688        if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
1689                return ($base, $head);
1690        }
1691        return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
1692        return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
1693        return ($head, $head) if ($::_revision eq 'HEAD');
1694        return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
1695        return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
1696        die "revision argument: $::_revision not understood by git-svn\n";
1697}
1698
1699sub fetch_all {
1700        my ($repo_id, $remotes) = @_;
1701        if (ref $repo_id) {
1702                my $gs = $repo_id;
1703                $repo_id = undef;
1704                $repo_id = $gs->{repo_id};
1705        }
1706        $remotes ||= read_all_remotes();
1707        my $remote = $remotes->{$repo_id} or
1708                     die "[svn-remote \"$repo_id\"] unknown\n";
1709        my $fetch = $remote->{fetch};
1710        my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
1711        my (@gs, @globs);
1712        my $ra = Git::SVN::Ra->new($url);
1713        my $uuid = $ra->get_uuid;
1714        my $head = $ra->get_latest_revnum;
1715        $ra->get_log("", $head, 0, 1, 0, 1, sub { $head = $_[1] });
1716        my $base = defined $fetch ? $head : 0;
1717
1718        # read the max revs for wildcard expansion (branches/*, tags/*)
1719        foreach my $t (qw/branches tags/) {
1720                defined $remote->{$t} or next;
1721                push @globs, @{$remote->{$t}};
1722
1723                my $max_rev = eval { tmp_config(qw/--int --get/,
1724                                         "svn-remote.$repo_id.${t}-maxRev") };
1725                if (defined $max_rev && ($max_rev < $base)) {
1726                        $base = $max_rev;
1727                } elsif (!defined $max_rev) {
1728                        $base = 0;
1729                }
1730        }
1731
1732        if ($fetch) {
1733                foreach my $p (sort keys %$fetch) {
1734                        my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
1735                        my $lr = $gs->rev_map_max;
1736                        if (defined $lr) {
1737                                $base = $lr if ($lr < $base);
1738                        }
1739                        push @gs, $gs;
1740                }
1741        }
1742
1743        ($base, $head) = parse_revision_argument($base, $head);
1744        $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
1745}
1746
1747sub read_all_remotes {
1748        my $r = {};
1749        my $use_svm_props = eval { command_oneline(qw/config --bool
1750            svn.useSvmProps/) };
1751        $use_svm_props = $use_svm_props eq 'true' if $use_svm_props;
1752        my $svn_refspec = qr{\s*/?(.*?)\s*:\s*(.+?)\s*};
1753        foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
1754                if (m!^(.+)\.fetch=$svn_refspec$!) {
1755                        my ($remote, $local_ref, $remote_ref) = ($1, $2, $3);
1756                        die("svn-remote.$remote: remote ref '$remote_ref' "
1757                            . "must start with 'refs/'\n")
1758                                unless $remote_ref =~ m{^refs/};
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} = $2;
1765                } elsif (m!^(.+)\.(branches|tags)=$svn_refspec$!) {
1766                        my ($remote, $t, $local_ref, $remote_ref) =
1767                                                             ($1, $2, $3, $4);
1768                        die("svn-remote.$remote: remote ref '$remote_ref' ($t) "
1769                            . "must start with 'refs/'\n")
1770                                unless $remote_ref =~ m{^refs/};
1771                        my $rs = {
1772                            t => $t,
1773                            remote => $remote,
1774                            path => Git::SVN::GlobSpec->new($local_ref),
1775                            ref => Git::SVN::GlobSpec->new($remote_ref) };
1776                        if (length($rs->{ref}->{right}) != 0) {
1777                                die "The '*' glob character must be the last ",
1778                                    "character of '$remote_ref'\n";
1779                        }
1780                        push @{ $r->{$remote}->{$t} }, $rs;
1781                }
1782        }
1783
1784        map {
1785                if (defined $r->{$_}->{svm}) {
1786                        my $svm;
1787                        eval {
1788                                my $section = "svn-remote.$_";
1789                                $svm = {
1790                                        source => tmp_config('--get',
1791                                            "$section.svm-source"),
1792                                        replace => tmp_config('--get',
1793                                            "$section.svm-replace"),
1794                                }
1795                        };
1796                        $r->{$_}->{svm} = $svm;
1797                }
1798        } keys %$r;
1799
1800        $r;
1801}
1802
1803sub init_vars {
1804        $_gc_nr = $_gc_period = 1000;
1805        if (defined $_repack || defined $_repack_flags) {
1806               warn "Repack options are obsolete; they have no effect.\n";
1807        }
1808}
1809
1810sub verify_remotes_sanity {
1811        return unless -d $ENV{GIT_DIR};
1812        my %seen;
1813        foreach (command(qw/config -l/)) {
1814                if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
1815                        if ($seen{$1}) {
1816                                die "Remote ref refs/remote/$1 is tracked by",
1817                                    "\n  \"$_\"\nand\n  \"$seen{$1}\"\n",
1818                                    "Please resolve this ambiguity in ",
1819                                    "your git configuration file before ",
1820                                    "continuing\n";
1821                        }
1822                        $seen{$1} = $_;
1823                }
1824        }
1825}
1826
1827sub find_existing_remote {
1828        my ($url, $remotes) = @_;
1829        return undef if $no_reuse_existing;
1830        my $existing;
1831        foreach my $repo_id (keys %$remotes) {
1832                my $u = $remotes->{$repo_id}->{url} or next;
1833                next if $u ne $url;
1834                $existing = $repo_id;
1835                last;
1836        }
1837        $existing;
1838}
1839
1840sub init_remote_config {
1841        my ($self, $url, $no_write) = @_;
1842        $url =~ s!/+$!!; # strip trailing slash
1843        my $r = read_all_remotes();
1844        my $existing = find_existing_remote($url, $r);
1845        if ($existing) {
1846                unless ($no_write) {
1847                        print STDERR "Using existing ",
1848                                     "[svn-remote \"$existing\"]\n";
1849                }
1850                $self->{repo_id} = $existing;
1851        } elsif ($_minimize_url) {
1852                my $min_url = Git::SVN::Ra->new($url)->minimize_url;
1853                $existing = find_existing_remote($min_url, $r);
1854                if ($existing) {
1855                        unless ($no_write) {
1856                                print STDERR "Using existing ",
1857                                             "[svn-remote \"$existing\"]\n";
1858                        }
1859                        $self->{repo_id} = $existing;
1860                }
1861                if ($min_url ne $url) {
1862                        unless ($no_write) {
1863                                print STDERR "Using higher level of URL: ",
1864                                             "$url => $min_url\n";
1865                        }
1866                        my $old_path = $self->{path};
1867                        $self->{path} = $url;
1868                        $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
1869                        if (length $old_path) {
1870                                $self->{path} .= "/$old_path";
1871                        }
1872                        $url = $min_url;
1873                }
1874        }
1875        my $orig_url;
1876        if (!$existing) {
1877                # verify that we aren't overwriting anything:
1878                $orig_url = eval {
1879                        command_oneline('config', '--get',
1880                                        "svn-remote.$self->{repo_id}.url")
1881                };
1882                if ($orig_url && ($orig_url ne $url)) {
1883                        die "svn-remote.$self->{repo_id}.url already set: ",
1884                            "$orig_url\nwanted to set to: $url\n";
1885                }
1886        }
1887        my ($xrepo_id, $xpath) = find_ref($self->refname);
1888        if (!$no_write && defined $xpath) {
1889                die "svn-remote.$xrepo_id.fetch already set to track ",
1890                    "$xpath:", $self->refname, "\n";
1891        }
1892        unless ($no_write) {
1893                command_noisy('config',
1894                              "svn-remote.$self->{repo_id}.url", $url);
1895                $self->{path} =~ s{^/}{};
1896                $self->{path} =~ s{%([0-9A-F]{2})}{chr hex($1)}ieg;
1897                command_noisy('config', '--add',
1898                              "svn-remote.$self->{repo_id}.fetch",
1899                              "$self->{path}:".$self->refname);
1900        }
1901        $self->{url} = $url;
1902}
1903
1904sub find_by_url { # repos_root and, path are optional
1905        my ($class, $full_url, $repos_root, $path) = @_;
1906
1907        return undef unless defined $full_url;
1908        remove_username($full_url);
1909        remove_username($repos_root) if defined $repos_root;
1910        my $remotes = read_all_remotes();
1911        if (defined $full_url && defined $repos_root && !defined $path) {
1912                $path = $full_url;
1913                $path =~ s#^\Q$repos_root\E(?:/|$)##;
1914        }
1915        foreach my $repo_id (keys %$remotes) {
1916                my $u = $remotes->{$repo_id}->{url} or next;
1917                remove_username($u);
1918                next if defined $repos_root && $repos_root ne $u;
1919
1920                my $fetch = $remotes->{$repo_id}->{fetch} || {};
1921                foreach my $t (qw/branches tags/) {
1922                        foreach my $globspec (@{$remotes->{$repo_id}->{$t}}) {
1923                                resolve_local_globs($u, $fetch, $globspec);
1924                        }
1925                }
1926                my $p = $path;
1927                my $rwr = rewrite_root({repo_id => $repo_id});
1928                my $svm = $remotes->{$repo_id}->{svm}
1929                        if defined $remotes->{$repo_id}->{svm};
1930                unless (defined $p) {
1931                        $p = $full_url;
1932                        my $z = $u;
1933                        my $prefix = '';
1934                        if ($rwr) {
1935                                $z = $rwr;
1936                                remove_username($z);
1937                        } elsif (defined $svm) {
1938                                $z = $svm->{source};
1939                                $prefix = $svm->{replace};
1940                                $prefix =~ s#^\Q$u\E(?:/|$)##;
1941                                $prefix =~ s#/$##;
1942                        }
1943                        $p =~ s#^\Q$z\E(?:/|$)#$prefix# or next;
1944                }
1945                foreach my $f (keys %$fetch) {
1946                        next if $f ne $p;
1947                        return Git::SVN->new($fetch->{$f}, $repo_id, $f);
1948                }
1949        }
1950        undef;
1951}
1952
1953sub init {
1954        my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
1955        my $self = _new($class, $repo_id, $ref_id, $path);
1956        if (defined $url) {
1957                $self->init_remote_config($url, $no_write);
1958        }
1959        $self;
1960}
1961
1962sub find_ref {
1963        my ($ref_id) = @_;
1964        foreach (command(qw/config -l/)) {
1965                next unless m!^svn-remote\.(.+)\.fetch=
1966                              \s*/?(.*?)\s*:\s*(.+?)\s*$!x;
1967                my ($repo_id, $path, $ref) = ($1, $2, $3);
1968                if ($ref eq $ref_id) {
1969                        $path = '' if ($path =~ m#^\./?#);
1970                        return ($repo_id, $path);
1971                }
1972        }
1973        (undef, undef, undef);
1974}
1975
1976sub new {
1977        my ($class, $ref_id, $repo_id, $path) = @_;
1978        if (defined $ref_id && !defined $repo_id && !defined $path) {
1979                ($repo_id, $path) = find_ref($ref_id);
1980                if (!defined $repo_id) {
1981                        die "Could not find a \"svn-remote.*.fetch\" key ",
1982                            "in the repository configuration matching: ",
1983                            "$ref_id\n";
1984                }
1985        }
1986        my $self = _new($class, $repo_id, $ref_id, $path);
1987        if (!defined $self->{path} || !length $self->{path}) {
1988                my $fetch = command_oneline('config', '--get',
1989                                            "svn-remote.$repo_id.fetch",
1990                                            ":$ref_id\$") or
1991                     die "Failed to read \"svn-remote.$repo_id.fetch\" ",
1992                         "\":$ref_id\$\" in config\n";
1993                ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
1994        }
1995        $self->{url} = command_oneline('config', '--get',
1996                                       "svn-remote.$repo_id.url") or
1997                  die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
1998        $self->rebuild;
1999        $self;
2000}
2001
2002sub refname {
2003        my ($refname) = $_[0]->{ref_id} ;
2004
2005        # It cannot end with a slash /, we'll throw up on this because
2006        # SVN can't have directories with a slash in their name, either:
2007        if ($refname =~ m{/$}) {
2008                die "ref: '$refname' ends with a trailing slash, this is ",
2009                    "not permitted by git nor Subversion\n";
2010        }
2011
2012        # It cannot have ASCII control character space, tilde ~, caret ^,
2013        # colon :, question-mark ?, asterisk *, space, or open bracket [
2014        # anywhere.
2015        #
2016        # Additionally, % must be escaped because it is used for escaping
2017        # and we want our escaped refname to be reversible
2018        $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
2019
2020        # no slash-separated component can begin with a dot .
2021        # /.* becomes /%2E*
2022        $refname =~ s{/\.}{/%2E}g;
2023
2024        # It cannot have two consecutive dots .. anywhere
2025        # .. becomes %2E%2E
2026        $refname =~ s{\.\.}{%2E%2E}g;
2027
2028        return $refname;
2029}
2030
2031sub desanitize_refname {
2032        my ($refname) = @_;
2033        $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
2034        return $refname;
2035}
2036
2037sub svm_uuid {
2038        my ($self) = @_;
2039        return $self->{svm}->{uuid} if $self->svm;
2040        $self->ra;
2041        unless ($self->{svm}) {
2042                die "SVM UUID not cached, and reading remotely failed\n";
2043        }
2044        $self->{svm}->{uuid};
2045}
2046
2047sub svm {
2048        my ($self) = @_;
2049        return $self->{svm} if $self->{svm};
2050        my $svm;
2051        # see if we have it in our config, first:
2052        eval {
2053                my $section = "svn-remote.$self->{repo_id}";
2054                $svm = {
2055                  source => tmp_config('--get', "$section.svm-source"),
2056                  uuid => tmp_config('--get', "$section.svm-uuid"),
2057                  replace => tmp_config('--get', "$section.svm-replace"),
2058                }
2059        };
2060        if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
2061                $self->{svm} = $svm;
2062        }
2063        $self->{svm};
2064}
2065
2066sub _set_svm_vars {
2067        my ($self, $ra) = @_;
2068        return $ra if $self->svm;
2069
2070        my @err = ( "useSvmProps set, but failed to read SVM properties\n",
2071                    "(svm:source, svm:uuid) ",
2072                    "from the following URLs:\n" );
2073        sub read_svm_props {
2074                my ($self, $ra, $path, $r) = @_;
2075                my $props = ($ra->get_dir($path, $r))[2];
2076                my $src = $props->{'svm:source'};
2077                my $uuid = $props->{'svm:uuid'};
2078                return undef if (!$src || !$uuid);
2079
2080                chomp($src, $uuid);
2081
2082                $uuid =~ m{^[0-9a-f\-]{30,}$}i
2083                    or die "doesn't look right - svm:uuid is '$uuid'\n";
2084
2085                # the '!' is used to mark the repos_root!/relative/path
2086                $src =~ s{/?!/?}{/};
2087                $src =~ s{/+$}{}; # no trailing slashes please
2088                # username is of no interest
2089                $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
2090
2091                my $replace = $ra->{url};
2092                $replace .= "/$path" if length $path;
2093
2094                my $section = "svn-remote.$self->{repo_id}";
2095                tmp_config("$section.svm-source", $src);
2096                tmp_config("$section.svm-replace", $replace);
2097                tmp_config("$section.svm-uuid", $uuid);
2098                $self->{svm} = {
2099                        source => $src,
2100                        uuid => $uuid,
2101                        replace => $replace
2102                };
2103        }
2104
2105        my $r = $ra->get_latest_revnum;
2106        my $path = $self->{path};
2107        my %tried;
2108        while (length $path) {
2109                unless ($tried{"$self->{url}/$path"}) {
2110                        return $ra if $self->read_svm_props($ra, $path, $r);
2111                        $tried{"$self->{url}/$path"} = 1;
2112                }
2113                $path =~ s#/?[^/]+$##;
2114        }
2115        die "Path: '$path' should be ''\n" if $path ne '';
2116        return $ra if $self->read_svm_props($ra, $path, $r);
2117        $tried{"$self->{url}/$path"} = 1;
2118
2119        if ($ra->{repos_root} eq $self->{url}) {
2120                die @err, (map { "  $_\n" } keys %tried), "\n";
2121        }
2122
2123        # nope, make sure we're connected to the repository root:
2124        my $ok;
2125        my @tried_b;
2126        $path = $ra->{svn_path};
2127        $ra = Git::SVN::Ra->new($ra->{repos_root});
2128        while (length $path) {
2129                unless ($tried{"$ra->{url}/$path"}) {
2130                        $ok = $self->read_svm_props($ra, $path, $r);
2131                        last if $ok;
2132                        $tried{"$ra->{url}/$path"} = 1;
2133                }
2134                $path =~ s#/?[^/]+$##;
2135        }
2136        die "Path: '$path' should be ''\n" if $path ne '';
2137        $ok ||= $self->read_svm_props($ra, $path, $r);
2138        $tried{"$ra->{url}/$path"} = 1;
2139        if (!$ok) {
2140                die @err, (map { "  $_\n" } keys %tried), "\n";
2141        }
2142        Git::SVN::Ra->new($self->{url});
2143}
2144
2145sub svnsync {
2146        my ($self) = @_;
2147        return $self->{svnsync} if $self->{svnsync};
2148
2149        if ($self->no_metadata) {
2150                die "Can't have both 'noMetadata' and ",
2151                    "'useSvnsyncProps' options set!\n";
2152        }
2153        if ($self->rewrite_root) {
2154                die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
2155                    "options set!\n";
2156        }
2157
2158        my $svnsync;
2159        # see if we have it in our config, first:
2160        eval {
2161                my $section = "svn-remote.$self->{repo_id}";
2162
2163                my $url = tmp_config('--get', "$section.svnsync-url");
2164                ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2165                   die "doesn't look right - svn:sync-from-url is '$url'\n";
2166
2167                my $uuid = tmp_config('--get', "$section.svnsync-uuid");
2168                ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2169                   die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2170
2171                $svnsync = { url => $url, uuid => $uuid }
2172        };
2173        if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
2174                return $self->{svnsync} = $svnsync;
2175        }
2176
2177        my $err = "useSvnsyncProps set, but failed to read " .
2178                  "svnsync property: svn:sync-from-";
2179        my $rp = $self->ra->rev_proplist(0);
2180
2181        my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
2182        ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2183                   die "doesn't look right - svn:sync-from-url is '$url'\n";
2184
2185        my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
2186        ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2187                   die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2188
2189        my $section = "svn-remote.$self->{repo_id}";
2190        tmp_config('--add', "$section.svnsync-uuid", $uuid);
2191        tmp_config('--add', "$section.svnsync-url", $url);
2192        return $self->{svnsync} = { url => $url, uuid => $uuid };
2193}
2194
2195# this allows us to memoize our SVN::Ra UUID locally and avoid a
2196# remote lookup (useful for 'git svn log').
2197sub ra_uuid {
2198        my ($self) = @_;
2199        unless ($self->{ra_uuid}) {
2200                my $key = "svn-remote.$self->{repo_id}.uuid";
2201                my $uuid = eval { tmp_config('--get', $key) };
2202                if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/i) {
2203                        $self->{ra_uuid} = $uuid;
2204                } else {
2205                        die "ra_uuid called without URL\n" unless $self->{url};
2206                        $self->{ra_uuid} = $self->ra->get_uuid;
2207                        tmp_config('--add', $key, $self->{ra_uuid});
2208                }
2209        }
2210        $self->{ra_uuid};
2211}
2212
2213sub _set_repos_root {
2214        my ($self, $repos_root) = @_;
2215        my $k = "svn-remote.$self->{repo_id}.reposRoot";
2216        $repos_root ||= $self->ra->{repos_root};
2217        tmp_config($k, $repos_root);
2218        $repos_root;
2219}
2220
2221sub repos_root {
2222        my ($self) = @_;
2223        my $k = "svn-remote.$self->{repo_id}.reposRoot";
2224        eval { tmp_config('--get', $k) } || $self->_set_repos_root;
2225}
2226
2227sub ra {
2228        my ($self) = shift;
2229        my $ra = Git::SVN::Ra->new($self->{url});
2230        $self->_set_repos_root($ra->{repos_root});
2231        if ($self->use_svm_props && !$self->{svm}) {
2232                if ($self->no_metadata) {
2233                        die "Can't have both 'noMetadata' and ",
2234                            "'useSvmProps' options set!\n";
2235                } elsif ($self->use_svnsync_props) {
2236                        die "Can't have both 'useSvnsyncProps' and ",
2237                            "'useSvmProps' options set!\n";
2238                }
2239                $ra = $self->_set_svm_vars($ra);
2240                $self->{-want_revprops} = 1;
2241        }
2242        $ra;
2243}
2244
2245# prop_walk(PATH, REV, SUB)
2246# -------------------------
2247# Recursively traverse PATH at revision REV and invoke SUB for each
2248# directory that contains a SVN property.  SUB will be invoked as
2249# follows:  &SUB(gs, path, props);  where `gs' is this instance of
2250# Git::SVN, `path' the path to the directory where the properties
2251# `props' were found.  The `path' will be relative to point of checkout,
2252# that is, if url://repo/trunk is the current Git branch, and that
2253# directory contains a sub-directory `d', SUB will be invoked with `/d/'
2254# as `path' (note the trailing `/').
2255sub prop_walk {
2256        my ($self, $path, $rev, $sub) = @_;
2257
2258        $path =~ s#^/##;
2259        my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev);
2260        $path =~ s#^/*#/#g;
2261        my $p = $path;
2262        # Strip the irrelevant part of the path.
2263        $p =~ s#^/+\Q$self->{path}\E(/|$)#/#;
2264        # Ensure the path is terminated by a `/'.
2265        $p =~ s#/*$#/#;
2266
2267        # The properties contain all the internal SVN stuff nobody
2268        # (usually) cares about.
2269        my $interesting_props = 0;
2270        foreach (keys %{$props}) {
2271                # If it doesn't start with `svn:', it must be a
2272                # user-defined property.
2273                ++$interesting_props and next if $_ !~ /^svn:/;
2274                # FIXME: Fragile, if SVN adds new public properties,
2275                # this needs to be updated.
2276                ++$interesting_props if /^svn:(?:ignore|keywords|executable
2277                                                 |eol-style|mime-type
2278                                                 |externals|needs-lock)$/x;
2279        }
2280        &$sub($self, $p, $props) if $interesting_props;
2281
2282        foreach (sort keys %$dirent) {
2283                next if $dirent->{$_}->{kind} != $SVN::Node::dir;
2284                $self->prop_walk($self->{path} . $p . $_, $rev, $sub);
2285        }
2286}
2287
2288sub last_rev { ($_[0]->last_rev_commit)[0] }
2289sub last_commit { ($_[0]->last_rev_commit)[1] }
2290
2291# returns the newest SVN revision number and newest commit SHA1
2292sub last_rev_commit {
2293        my ($self) = @_;
2294        if (defined $self->{last_rev} && defined $self->{last_commit}) {
2295                return ($self->{last_rev}, $self->{last_commit});
2296        }
2297        my $c = ::verify_ref($self->refname.'^0');
2298        if ($c && !$self->use_svm_props && !$self->no_metadata) {
2299                my $rev = (::cmt_metadata($c))[1];
2300                if (defined $rev) {
2301                        ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
2302                        return ($rev, $c);
2303                }
2304        }
2305        my $map_path = $self->map_path;
2306        unless (-e $map_path) {
2307                ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
2308                return (undef, undef);
2309        }
2310        my ($rev, $commit) = $self->rev_map_max(1);
2311        ($self->{last_rev}, $self->{last_commit}) = ($rev, $commit);
2312        return ($rev, $commit);
2313}
2314
2315sub get_fetch_range {
2316        my ($self, $min, $max) = @_;
2317        $max ||= $self->ra->get_latest_revnum;
2318        $min ||= $self->rev_map_max;
2319        (++$min, $max);
2320}
2321
2322sub tmp_config {
2323        my (@args) = @_;
2324        my $old_def_config = "$ENV{GIT_DIR}/svn/config";
2325        my $config = "$ENV{GIT_DIR}/svn/.metadata";
2326        if (! -f $config && -f $old_def_config) {
2327                rename $old_def_config, $config or
2328                       die "Failed rename $old_def_config => $config: $!\n";
2329        }
2330        my $old_config = $ENV{GIT_CONFIG};
2331        $ENV{GIT_CONFIG} = $config;
2332        $@ = undef;
2333        my @ret = eval {
2334                unless (-f $config) {
2335                        mkfile($config);
2336                        open my $fh, '>', $config or
2337                            die "Can't open $config: $!\n";
2338                        print $fh "; This file is used internally by ",
2339                                  "git-svn\n" or die
2340                                  "Couldn't write to $config: $!\n";
2341                        print $fh "; You should not have to edit it\n" or
2342                              die "Couldn't write to $config: $!\n";
2343                        close $fh or die "Couldn't close $config: $!\n";
2344                }
2345                command('config', @args);
2346        };
2347        my $err = $@;
2348        if (defined $old_config) {
2349                $ENV{GIT_CONFIG} = $old_config;
2350        } else {
2351                delete $ENV{GIT_CONFIG};
2352        }
2353        die $err if $err;
2354        wantarray ? @ret : $ret[0];
2355}
2356
2357sub tmp_index_do {
2358        my ($self, $sub) = @_;
2359        my $old_index = $ENV{GIT_INDEX_FILE};
2360        $ENV{GIT_INDEX_FILE} = $self->{index};
2361        $@ = undef;
2362        my @ret = eval {
2363                my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
2364                mkpath([$dir]) unless -d $dir;
2365                &$sub;
2366        };
2367        my $err = $@;
2368        if (defined $old_index) {
2369                $ENV{GIT_INDEX_FILE} = $old_index;
2370        } else {
2371                delete $ENV{GIT_INDEX_FILE};
2372        }
2373        die $err if $err;
2374        wantarray ? @ret : $ret[0];
2375}
2376
2377sub assert_index_clean {
2378        my ($self, $treeish) = @_;
2379
2380        $self->tmp_index_do(sub {
2381                command_noisy('read-tree', $treeish) unless -e $self->{index};
2382                my $x = command_oneline('write-tree');
2383                my ($y) = (command(qw/cat-file commit/, $treeish) =~
2384                           /^tree ($::sha1)/mo);
2385                return if $y eq $x;
2386
2387                warn "Index mismatch: $y != $x\nrereading $treeish\n";
2388                unlink $self->{index} or die "unlink $self->{index}: $!\n";
2389                command_noisy('read-tree', $treeish);
2390                $x = command_oneline('write-tree');
2391                if ($y ne $x) {
2392                        ::fatal "trees ($treeish) $y != $x\n",
2393                                "Something is seriously wrong...";
2394                }
2395        });
2396}
2397
2398sub get_commit_parents {
2399        my ($self, $log_entry) = @_;
2400        my (%seen, @ret, @tmp);
2401        # legacy support for 'set-tree'; this is only used by set_tree_cb:
2402        if (my $ip = $self->{inject_parents}) {
2403                if (my $commit = delete $ip->{$log_entry->{revision}}) {
2404                        push @tmp, $commit;
2405                }
2406        }
2407        if (my $cur = ::verify_ref($self->refname.'^0')) {
2408                push @tmp, $cur;
2409        }
2410        if (my $ipd = $self->{inject_parents_dcommit}) {
2411                if (my $commit = delete $ipd->{$log_entry->{revision}}) {
2412                        push @tmp, @$commit;
2413                }
2414        }
2415        push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
2416        while (my $p = shift @tmp) {
2417                next if $seen{$p};
2418                $seen{$p} = 1;
2419                push @ret, $p;
2420                # MAXPARENT is defined to 16 in commit-tree.c:
2421                last if @ret >= 16;
2422        }
2423        if (@tmp) {
2424                die "r$log_entry->{revision}: No room for parents:\n\t",
2425                    join("\n\t", @tmp), "\n";
2426        }
2427        @ret;
2428}
2429
2430sub rewrite_root {
2431        my ($self) = @_;
2432        return $self->{-rewrite_root} if exists $self->{-rewrite_root};
2433        my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
2434        my $rwr = eval { command_oneline(qw/config --get/, $k) };
2435        if ($rwr) {
2436                $rwr =~ s#/+$##;
2437                if ($rwr !~ m#^[a-z\+]+://#) {
2438                        die "$rwr is not a valid URL (key: $k)\n";
2439                }
2440        }
2441        $self->{-rewrite_root} = $rwr;
2442}
2443
2444sub metadata_url {
2445        my ($self) = @_;
2446        ($self->rewrite_root || $self->{url}) .
2447           (length $self->{path} ? '/' . $self->{path} : '');
2448}
2449
2450sub full_url {
2451        my ($self) = @_;
2452        $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
2453}
2454
2455
2456sub set_commit_header_env {
2457        my ($log_entry) = @_;
2458        my %env;
2459        foreach my $ned (qw/NAME EMAIL DATE/) {
2460                foreach my $ac (qw/AUTHOR COMMITTER/) {
2461                        $env{"GIT_${ac}_${ned}"} = $ENV{"GIT_${ac}_${ned}"};
2462                }
2463        }
2464
2465        $ENV{GIT_AUTHOR_NAME} = $log_entry->{name};
2466        $ENV{GIT_AUTHOR_EMAIL} = $log_entry->{email};
2467        $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
2468
2469        $ENV{GIT_COMMITTER_NAME} = (defined $log_entry->{commit_name})
2470                                                ? $log_entry->{commit_name}
2471                                                : $log_entry->{name};
2472        $ENV{GIT_COMMITTER_EMAIL} = (defined $log_entry->{commit_email})
2473                                                ? $log_entry->{commit_email}
2474                                                : $log_entry->{email};
2475        \%env;
2476}
2477
2478sub restore_commit_header_env {
2479        my ($env) = @_;
2480        foreach my $ned (qw/NAME EMAIL DATE/) {
2481                foreach my $ac (qw/AUTHOR COMMITTER/) {
2482                        my $k = "GIT_${ac}_${ned}";
2483                        if (defined $env->{$k}) {
2484                                $ENV{$k} = $env->{$k};
2485                        } else {
2486                                delete $ENV{$k};
2487                        }
2488                }
2489        }
2490}
2491
2492sub gc {
2493        command_noisy('gc', '--auto');
2494};
2495
2496sub do_git_commit {
2497        my ($self, $log_entry) = @_;
2498        my $lr = $self->last_rev;
2499        if (defined $lr && $lr >= $log_entry->{revision}) {
2500                die "Last fetched revision of ", $self->refname,
2501                    " was r$lr, but we are about to fetch: ",
2502                    "r$log_entry->{revision}!\n";
2503        }
2504        if (my $c = $self->rev_map_get($log_entry->{revision})) {
2505                croak "$log_entry->{revision} = $c already exists! ",
2506                      "Why are we refetching it?\n";
2507        }
2508        my $old_env = set_commit_header_env($log_entry);
2509        my $tree = $log_entry->{tree};
2510        if (!defined $tree) {
2511                $tree = $self->tmp_index_do(sub {
2512                                            command_oneline('write-tree') });
2513        }
2514        die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
2515
2516        my @exec = ('git', 'commit-tree', $tree);
2517        foreach ($self->get_commit_parents($log_entry)) {
2518                push @exec, '-p', $_;
2519        }
2520        defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
2521                                                                   or croak $!;
2522        binmode $msg_fh;
2523
2524        # we always get UTF-8 from SVN, but we may want our commits in
2525        # a different encoding.
2526        if (my $enc = Git::config('i18n.commitencoding')) {
2527                require Encode;
2528                Encode::from_to($log_entry->{log}, 'UTF-8', $enc);
2529        }
2530        print $msg_fh $log_entry->{log} or croak $!;
2531        restore_commit_header_env($old_env);
2532        unless ($self->no_metadata) {
2533                print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
2534                              or croak $!;
2535        }
2536        $msg_fh->flush == 0 or croak $!;
2537        close $msg_fh or croak $!;
2538        chomp(my $commit = do { local $/; <$out_fh> });
2539        close $out_fh or croak $!;
2540        waitpid $pid, 0;
2541        croak $? if $?;
2542        if ($commit !~ /^$::sha1$/o) {
2543                die "Failed to commit, invalid sha1: $commit\n";
2544        }
2545
2546        $self->rev_map_set($log_entry->{revision}, $commit, 1);
2547
2548        $self->{last_rev} = $log_entry->{revision};
2549        $self->{last_commit} = $commit;
2550        print "r$log_entry->{revision}" unless $::_q > 1;
2551        if (defined $log_entry->{svm_revision}) {
2552                 print " (\@$log_entry->{svm_revision})" unless $::_q > 1;
2553                 $self->rev_map_set($log_entry->{svm_revision}, $commit,
2554                                   0, $self->svm_uuid);
2555        }
2556        print " = $commit ($self->{ref_id})\n" unless $::_q > 1;
2557        if (--$_gc_nr == 0) {
2558                $_gc_nr = $_gc_period;
2559                gc();
2560        }
2561        return $commit;
2562}
2563
2564sub match_paths {
2565        my ($self, $paths, $r) = @_;
2566        return 1 if $self->{path} eq '';
2567        if (my $path = $paths->{"/$self->{path}"}) {
2568                return ($path->{action} eq 'D') ? 0 : 1;
2569        }
2570        $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
2571        if (grep /$self->{path_regex}/, keys %$paths) {
2572                return 1;
2573        }
2574        my $c = '';
2575        foreach (split m#/#, $self->{path}) {
2576                $c .= "/$_";
2577                next unless ($paths->{$c} &&
2578                             ($paths->{$c}->{action} =~ /^[AR]$/));
2579                if ($self->ra->check_path($self->{path}, $r) ==
2580                    $SVN::Node::dir) {
2581                        return 1;
2582                }
2583        }
2584        return 0;
2585}
2586
2587sub find_parent_branch {
2588        my ($self, $paths, $rev) = @_;
2589        return undef unless $self->follow_parent;
2590        unless (defined $paths) {
2591                my $err_handler = $SVN::Error::handler;
2592                $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
2593                $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1,
2594                                   sub { $paths = $_[0] });
2595                $SVN::Error::handler = $err_handler;
2596        }
2597        return undef unless defined $paths;
2598
2599        # look for a parent from another branch:
2600        my @b_path_components = split m#/#, $self->{path};
2601        my @a_path_components;
2602        my $i;
2603        while (@b_path_components) {
2604                $i = $paths->{'/'.join('/', @b_path_components)};
2605                last if $i && defined $i->{copyfrom_path};
2606                unshift(@a_path_components, pop(@b_path_components));
2607        }
2608        return undef unless defined $i && defined $i->{copyfrom_path};
2609        my $branch_from = $i->{copyfrom_path};
2610        if (@a_path_components) {
2611                print STDERR "branch_from: $branch_from => ";
2612                $branch_from .= '/'.join('/', @a_path_components);
2613                print STDERR $branch_from, "\n";
2614        }
2615        my $r = $i->{copyfrom_rev};
2616        my $repos_root = $self->ra->{repos_root};
2617        my $url = $self->ra->{url};
2618        my $new_url = $url . $branch_from;
2619        print STDERR  "Found possible branch point: ",
2620                      "$new_url => ", $self->full_url, ", $r\n";
2621        $branch_from =~ s#^/##;
2622        my $gs = $self->other_gs($new_url, $url,
2623                                 $branch_from, $r, $self->{ref_id});
2624        my ($r0, $parent) = $gs->find_rev_before($r, 1);
2625        {
2626                my ($base, $head);
2627                if (!defined $r0 || !defined $parent) {
2628                        ($base, $head) = parse_revision_argument(0, $r);
2629                } else {
2630                        if ($r0 < $r) {
2631                                $gs->ra->get_log([$gs->{path}], $r0 + 1, $r, 1,
2632                                        0, 1, sub { $base = $_[1] - 1 });
2633                        }
2634                }
2635                if (defined $base && $base <= $r) {
2636                        $gs->fetch($base, $r);
2637                }
2638                ($r0, $parent) = $gs->find_rev_before($r, 1);
2639        }
2640        if (defined $r0 && defined $parent) {
2641                print STDERR "Found branch parent: ($self->{ref_id}) $parent\n";
2642                my $ed;
2643                if ($self->ra->can_do_switch) {
2644                        $self->assert_index_clean($parent);
2645                        print STDERR "Following parent with do_switch\n";
2646                        # do_switch works with svn/trunk >= r22312, but that
2647                        # is not included with SVN 1.4.3 (the latest version
2648                        # at the moment), so we can't rely on it
2649                        $self->{last_rev} = $r0;
2650                        $self->{last_commit} = $parent;
2651                        $ed = SVN::Git::Fetcher->new($self, $gs->{path});
2652                        $gs->ra->gs_do_switch($r0, $rev, $gs,
2653                                              $self->full_url, $ed)
2654                          or die "SVN connection failed somewhere...\n";
2655                } elsif ($self->ra->trees_match($new_url, $r0,
2656                                                $self->full_url, $rev)) {
2657                        print STDERR "Trees match:\n",
2658                                     "  $new_url\@$r0\n",
2659                                     "  ${\$self->full_url}\@$rev\n",
2660                                     "Following parent with no changes\n";
2661                        $self->tmp_index_do(sub {
2662                            command_noisy('read-tree', $parent);
2663                        });
2664                        $self->{last_commit} = $parent;
2665                } else {
2666                        print STDERR "Following parent with do_update\n";
2667                        $ed = SVN::Git::Fetcher->new($self);
2668                        $self->ra->gs_do_update($rev, $rev, $self, $ed)
2669                          or die "SVN connection failed somewhere...\n";
2670                }
2671                print STDERR "Successfully followed parent\n";
2672                return $self->make_log_entry($rev, [$parent], $ed);
2673        }
2674        return undef;
2675}
2676
2677sub do_fetch {
2678        my ($self, $paths, $rev) = @_;
2679        my $ed;
2680        my ($last_rev, @parents);
2681        if (my $lc = $self->last_commit) {
2682                # we can have a branch that was deleted, then re-added
2683                # under the same name but copied from another path, in
2684                # which case we'll have multiple parents (we don't
2685                # want to break the original ref, nor lose copypath info):
2686                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2687                        push @{$log_entry->{parents}}, $lc;
2688                        return $log_entry;
2689                }
2690                $ed = SVN::Git::Fetcher->new($self);
2691                $last_rev = $self->{last_rev};
2692                $ed->{c} = $lc;
2693                @parents = ($lc);
2694        } else {
2695                $last_rev = $rev;
2696                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2697                        return $log_entry;
2698                }
2699                $ed = SVN::Git::Fetcher->new($self);
2700        }
2701        unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
2702                die "SVN connection failed somewhere...\n";
2703        }
2704        $self->make_log_entry($rev, \@parents, $ed);
2705}
2706
2707sub get_untracked {
2708        my ($self, $ed) = @_;
2709        my @out;
2710        my $h = $ed->{empty};
2711        foreach (sort keys %$h) {
2712                my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
2713                push @out, "  $act: " . uri_encode($_);
2714                warn "W: $act: $_\n";
2715        }
2716        foreach my $t (qw/dir_prop file_prop/) {
2717                $h = $ed->{$t} or next;
2718                foreach my $path (sort keys %$h) {
2719                        my $ppath = $path eq '' ? '.' : $path;
2720                        foreach my $prop (sort keys %{$h->{$path}}) {
2721                                next if $SKIP_PROP{$prop};
2722                                my $v = $h->{$path}->{$prop};
2723                                my $t_ppath_prop = "$t: " .
2724                                                    uri_encode($ppath) . ' ' .
2725                                                    uri_encode($prop);
2726                                if (defined $v) {
2727                                        push @out, "  +$t_ppath_prop " .
2728                                                   uri_encode($v);
2729                                } else {
2730                                        push @out, "  -$t_ppath_prop";
2731                                }
2732                        }
2733                }
2734        }
2735        foreach my $t (qw/absent_file absent_directory/) {
2736                $h = $ed->{$t} or next;
2737                foreach my $parent (sort keys %$h) {
2738                        foreach my $path (sort @{$h->{$parent}}) {
2739                                push @out, "  $t: " .
2740                                           uri_encode("$parent/$path");
2741                                warn "W: $t: $parent/$path ",
2742                                     "Insufficient permissions?\n";
2743                        }
2744                }
2745        }
2746        \@out;
2747}
2748
2749# parse_svn_date(DATE)
2750# --------------------
2751# Given a date (in UTC) from Subversion, return a string in the format
2752# "<TZ Offset> <local date/time>" that Git will use.
2753#
2754# By default the parsed date will be in UTC; if $Git::SVN::_localtime
2755# is true we'll convert it to the local timezone instead.
2756sub parse_svn_date {
2757        my $date = shift || return '+0000 1970-01-01 00:00:00';
2758        my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2759                                            (\d\d)\:(\d\d)\:(\d\d)\.\d*Z$/x) or
2760                                         croak "Unable to parse date: $date\n";
2761        my $parsed_date;    # Set next.
2762
2763        if ($Git::SVN::_localtime) {
2764                # Translate the Subversion datetime to an epoch time.
2765                # Begin by switching ourselves to $date's timezone, UTC.
2766                my $old_env_TZ = $ENV{TZ};
2767                $ENV{TZ} = 'UTC';
2768
2769                my $epoch_in_UTC =
2770                    POSIX::strftime('%s', $S, $M, $H, $d, $m - 1, $Y - 1900);
2771
2772                # Determine our local timezone (including DST) at the
2773                # time of $epoch_in_UTC.  $Git::SVN::Log::TZ stored the
2774                # value of TZ, if any, at the time we were run.
2775                if (defined $Git::SVN::Log::TZ) {
2776                        $ENV{TZ} = $Git::SVN::Log::TZ;
2777                } else {
2778                        delete $ENV{TZ};
2779                }
2780
2781                my $our_TZ =
2782                    POSIX::strftime('%Z', $S, $M, $H, $d, $m - 1, $Y - 1900);
2783
2784                # This converts $epoch_in_UTC into our local timezone.
2785                my ($sec, $min, $hour, $mday, $mon, $year,
2786                    $wday, $yday, $isdst) = localtime($epoch_in_UTC);
2787
2788                $parsed_date = sprintf('%s %04d-%02d-%02d %02d:%02d:%02d',
2789                                       $our_TZ, $year + 1900, $mon + 1,
2790                                       $mday, $hour, $min, $sec);
2791
2792                # Reset us to the timezone in effect when we entered
2793                # this routine.
2794                if (defined $old_env_TZ) {
2795                        $ENV{TZ} = $old_env_TZ;
2796                } else {
2797                        delete $ENV{TZ};
2798                }
2799        } else {
2800                $parsed_date = "+0000 $Y-$m-$d $H:$M:$S";
2801        }
2802
2803        return $parsed_date;
2804}
2805
2806sub other_gs {
2807        my ($self, $new_url, $url,
2808            $branch_from, $r, $old_ref_id) = @_;
2809        my $gs = Git::SVN->find_by_url($new_url, $url, $branch_from);
2810        unless ($gs) {
2811                my $ref_id = $old_ref_id;
2812                $ref_id =~ s/\@\d+$//;
2813                $ref_id .= "\@$r";
2814                # just grow a tail if we're not unique enough :x
2815                $ref_id .= '-' while find_ref($ref_id);
2816                print STDERR "Initializing parent: $ref_id\n";
2817                my ($u, $p, $repo_id) = ($new_url, '', $ref_id);
2818                if ($u =~ s#^\Q$url\E(/|$)##) {
2819                        $p = $u;
2820                        $u = $url;
2821                        $repo_id = $self->{repo_id};
2822                }
2823                $gs = Git::SVN->init($u, $p, $repo_id, $ref_id, 1);
2824        }
2825        $gs
2826}
2827
2828sub call_authors_prog {
2829        my ($orig_author) = @_;
2830        my $author = `$::_authors_prog $orig_author`;
2831        if ($? != 0) {
2832                die "$::_authors_prog failed with exit code $?\n"
2833        }
2834        if ($author =~ /^\s*(.+?)\s*<(.*)>\s*$/) {
2835                my ($name, $email) = ($1, $2);
2836                $email = undef if length $2 == 0;
2837                return [$name, $email];
2838        } else {
2839                die "Author: $orig_author: $::_authors_prog returned "
2840                        . "invalid author format: $author\n";
2841        }
2842}
2843
2844sub check_author {
2845        my ($author) = @_;
2846        if (!defined $author || length $author == 0) {
2847                $author = '(no author)';
2848        }
2849        if (!defined $::users{$author}) {
2850                if (defined $::_authors_prog) {
2851                        $::users{$author} = call_authors_prog($author);
2852                } elsif (defined $::_authors) {
2853                        die "Author: $author not defined in $::_authors file\n";
2854                }
2855        }
2856        $author;
2857}
2858
2859sub make_log_entry {
2860        my ($self, $rev, $parents, $ed) = @_;
2861        my $untracked = $self->get_untracked($ed);
2862
2863        open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
2864        print $un "r$rev\n" or croak $!;
2865        print $un $_, "\n" foreach @$untracked;
2866        my %log_entry = ( parents => $parents || [], revision => $rev,
2867                          log => '');
2868
2869        my $headrev;
2870        my $logged = delete $self->{logged_rev_props};
2871        if (!$logged || $self->{-want_revprops}) {
2872                my $rp = $self->ra->rev_proplist($rev);
2873                foreach (sort keys %$rp) {
2874                        my $v = $rp->{$_};
2875                        if (/^svn:(author|date|log)$/) {
2876                                $log_entry{$1} = $v;
2877                        } elsif ($_ eq 'svm:headrev') {
2878                                $headrev = $v;
2879                        } else {
2880                                print $un "  rev_prop: ", uri_encode($_), ' ',
2881                                          uri_encode($v), "\n";
2882                        }
2883                }
2884        } else {
2885                map { $log_entry{$_} = $logged->{$_} } keys %$logged;
2886        }
2887        close $un or croak $!;
2888
2889        $log_entry{date} = parse_svn_date($log_entry{date});
2890        $log_entry{log} .= "\n";
2891        my $author = $log_entry{author} = check_author($log_entry{author});
2892        my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
2893                                                       : ($author, undef);
2894
2895        my ($commit_name, $commit_email) = ($name, $email);
2896        if ($_use_log_author) {
2897                my $name_field;
2898                if ($log_entry{log} =~ /From:\s+(.*\S)\s*\n/i) {
2899                        $name_field = $1;
2900                } elsif ($log_entry{log} =~ /Signed-off-by:\s+(.*\S)\s*\n/i) {
2901                        $name_field = $1;
2902                }
2903                if (!defined $name_field) {
2904                        if (!defined $email) {
2905                                $email = $name;
2906                        }
2907                } elsif ($name_field =~ /(.*?)\s+<(.*)>/) {
2908                        ($name, $email) = ($1, $2);
2909                } elsif ($name_field =~ /(.*)@/) {
2910                        ($name, $email) = ($1, $name_field);
2911                } else {
2912                        ($name, $email) = ($name_field, $name_field);
2913                }
2914        }
2915        if (defined $headrev && $self->use_svm_props) {
2916                if ($self->rewrite_root) {
2917                        die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
2918                            "options set!\n";
2919                }
2920                my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$}i;
2921                # we don't want "SVM: initializing mirror for junk" ...
2922                return undef if $r == 0;
2923                my $svm = $self->svm;
2924                if ($uuid ne $svm->{uuid}) {
2925                        die "UUID mismatch on SVM path:\n",
2926                            "expected: $svm->{uuid}\n",
2927                            "     got: $uuid\n";
2928                }
2929                my $full_url = $self->full_url;
2930                $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
2931                             die "Failed to replace '$svm->{replace}' with ",
2932                                 "'$svm->{source}' in $full_url\n";
2933                # throw away username for storing in records
2934                remove_username($full_url);
2935                $log_entry{metadata} = "$full_url\@$r $uuid";
2936                $log_entry{svm_revision} = $r;
2937                $email ||= "$author\@$uuid";
2938                $commit_email ||= "$author\@$uuid";
2939        } elsif ($self->use_svnsync_props) {
2940                my $full_url = $self->svnsync->{url};
2941                $full_url .= "/$self->{path}" if length $self->{path};
2942                remove_username($full_url);
2943                my $uuid = $self->svnsync->{uuid};
2944                $log_entry{metadata} = "$full_url\@$rev $uuid";
2945                $email ||= "$author\@$uuid";
2946                $commit_email ||= "$author\@$uuid";
2947        } else {
2948                my $url = $self->metadata_url;
2949                remove_username($url);
2950                $log_entry{metadata} = "$url\@$rev " .
2951                                       $self->ra->get_uuid;
2952                $email ||= "$author\@" . $self->ra->get_uuid;
2953                $commit_email ||= "$author\@" . $self->ra->get_uuid;
2954        }
2955        $log_entry{name} = $name;
2956        $log_entry{email} = $email;
2957        $log_entry{commit_name} = $commit_name;
2958        $log_entry{commit_email} = $commit_email;
2959        \%log_entry;
2960}
2961
2962sub fetch {
2963        my ($self, $min_rev, $max_rev, @parents) = @_;
2964        my ($last_rev, $last_commit) = $self->last_rev_commit;
2965        my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
2966        $self->ra->gs_fetch_loop_common($base, $head, [$self]);
2967}
2968
2969sub set_tree_cb {
2970        my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
2971        $self->{inject_parents} = { $rev => $tree };
2972        $self->fetch(undef, undef);
2973}
2974
2975sub set_tree {
2976        my ($self, $tree) = (shift, shift);
2977        my $log_entry = ::get_commit_entry($tree);
2978        unless ($self->{last_rev}) {
2979                ::fatal("Must have an existing revision to commit");
2980        }
2981        my %ed_opts = ( r => $self->{last_rev},
2982                        log => $log_entry->{log},
2983                        ra => $self->ra,
2984                        tree_a => $self->{last_commit},
2985                        tree_b => $tree,
2986                        editor_cb => sub {
2987                               $self->set_tree_cb($log_entry, $tree, @_) },
2988                        svn_path => $self->{path} );
2989        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
2990                print "No changes\nr$self->{last_rev} = $tree\n";
2991        }
2992}
2993
2994sub rebuild_from_rev_db {
2995        my ($self, $path) = @_;
2996        my $r = -1;
2997        open my $fh, '<', $path or croak "open: $!";
2998        binmode $fh or croak "binmode: $!";
2999        while (<$fh>) {
3000                length($_) == 41 or croak "inconsistent size in ($_) != 41";
3001                chomp($_);
3002                ++$r;
3003                next if $_ eq ('0' x 40);
3004                $self->rev_map_set($r, $_);
3005                print "r$r = $_\n";
3006        }
3007        close $fh or croak "close: $!";
3008        unlink $path or croak "unlink: $!";
3009}
3010
3011sub rebuild {
3012        my ($self) = @_;
3013        my $map_path = $self->map_path;
3014        my $partial = (-e $map_path && ! -z $map_path);
3015        return unless ::verify_ref($self->refname.'^0');
3016        if (!$partial && ($self->use_svm_props || $self->no_metadata)) {
3017                my $rev_db = $self->rev_db_path;
3018                $self->rebuild_from_rev_db($rev_db);
3019                if ($self->use_svm_props) {
3020                        my $svm_rev_db = $self->rev_db_path($self->svm_uuid);
3021                        $self->rebuild_from_rev_db($svm_rev_db);
3022                }
3023                $self->unlink_rev_db_symlink;
3024                return;
3025        }
3026        print "Rebuilding $map_path ...\n" if (!$partial);
3027        my ($base_rev, $head) = ($partial ? $self->rev_map_max_norebuild(1) :
3028                (undef, undef));
3029        my ($log, $ctx) =
3030            command_output_pipe(qw/rev-list --pretty=raw --no-color --reverse/,
3031                                ($head ? "$head.." : "") . $self->refname,
3032                                '--');
3033        my $metadata_url = $self->metadata_url;
3034        remove_username($metadata_url);
3035        my $svn_uuid = $self->ra_uuid;
3036        my $c;
3037        while (<$log>) {
3038                if ( m{^commit ($::sha1)$} ) {
3039                        $c = $1;
3040                        next;
3041                }
3042                next unless s{^\s*(git-svn-id:)}{$1};
3043                my ($url, $rev, $uuid) = ::extract_metadata($_);
3044                remove_username($url);
3045
3046                # ignore merges (from set-tree)
3047                next if (!defined $rev || !$uuid);
3048
3049                # if we merged or otherwise started elsewhere, this is
3050                # how we break out of it
3051                if (($uuid ne $svn_uuid) ||
3052                    ($metadata_url && $url && ($url ne $metadata_url))) {
3053                        next;
3054                }
3055                if ($partial && $head) {
3056                        print "Partial-rebuilding $map_path ...\n";
3057                        print "Currently at $base_rev = $head\n";
3058                        $head = undef;
3059                }
3060
3061                $self->rev_map_set($rev, $c);
3062                print "r$rev = $c\n";
3063        }
3064        command_close_pipe($log, $ctx);
3065        print "Done rebuilding $map_path\n" if (!$partial || !$head);
3066        my $rev_db_path = $self->rev_db_path;
3067        if (-f $self->rev_db_path) {
3068                unlink $self->rev_db_path or croak "unlink: $!";
3069        }
3070        $self->unlink_rev_db_symlink;
3071}
3072
3073# rev_map:
3074# Tie::File seems to be prone to offset errors if revisions get sparse,
3075# it's not that fast, either.  Tie::File is also not in Perl 5.6.  So
3076# one of my favorite modules is out :<  Next up would be one of the DBM
3077# modules, but I'm not sure which is most portable...
3078#
3079# This is the replacement for the rev_db format, which was too big
3080# and inefficient for large repositories with a lot of sparse history
3081# (mainly tags)
3082#
3083# The format is this:
3084#   - 24 bytes for every record,
3085#     * 4 bytes for the integer representing an SVN revision number
3086#     * 20 bytes representing the sha1 of a git commit
3087#   - No empty padding records like the old format
3088#     (except the last record, which can be overwritten)
3089#   - new records are written append-only since SVN revision numbers
3090#     increase monotonically
3091#   - lookups on SVN revision number are done via a binary search
3092#   - Piping the file to xxd -c24 is a good way of dumping it for
3093#     viewing or editing (piped back through xxd -r), should the need
3094#     ever arise.
3095#   - The last record can be padding revision with an all-zero sha1
3096#     This is used to optimize fetch performance when using multiple
3097#     "fetch" directives in .git/config
3098#
3099# These files are disposable unless noMetadata or useSvmProps is set
3100
3101sub _rev_map_set {
3102        my ($fh, $rev, $commit) = @_;
3103
3104        binmode $fh or croak "binmode: $!";
3105        my $size = (stat($fh))[7];
3106        ($size % 24) == 0 or croak "inconsistent size: $size";
3107
3108        my $wr_offset = 0;
3109        if ($size > 0) {
3110                sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3111                my $read = sysread($fh, my $buf, 24) or croak "read: $!";
3112                $read == 24 or croak "read only $read bytes (!= 24)";
3113                my ($last_rev, $last_commit) = unpack(rev_map_fmt, $buf);
3114                if ($last_commit eq ('0' x40)) {
3115                        if ($size >= 48) {
3116                                sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3117                                $read = sysread($fh, $buf, 24) or
3118                                    croak "read: $!";
3119                                $read == 24 or
3120                                    croak "read only $read bytes (!= 24)";
3121                                ($last_rev, $last_commit) =
3122                                    unpack(rev_map_fmt, $buf);
3123                                if ($last_commit eq ('0' x40)) {
3124                                        croak "inconsistent .rev_map\n";
3125                                }
3126                        }
3127                        if ($last_rev >= $rev) {
3128                                croak "last_rev is higher!: $last_rev >= $rev";
3129                        }
3130                        $wr_offset = -24;
3131                }
3132        }
3133        sysseek($fh, $wr_offset, SEEK_END) or croak "seek: $!";
3134        syswrite($fh, pack(rev_map_fmt, $rev, $commit), 24) == 24 or
3135          croak "write: $!";
3136}
3137
3138sub _rev_map_reset {
3139        my ($fh, $rev, $commit) = @_;
3140        my $c = _rev_map_get($fh, $rev);
3141        $c eq $commit or die "_rev_map_reset(@_) commit $c does not match!\n";
3142        my $offset = sysseek($fh, 0, SEEK_CUR) or croak "seek: $!";
3143        truncate $fh, $offset or croak "truncate: $!";
3144}
3145
3146sub mkfile {
3147        my ($path) = @_;
3148        unless (-e $path) {
3149                my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
3150                mkpath([$dir]) unless -d $dir;
3151                open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
3152                close $fh or die "Couldn't close (create) $path: $!\n";
3153        }
3154}
3155
3156sub rev_map_set {
3157        my ($self, $rev, $commit, $update_ref, $uuid) = @_;
3158        length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
3159        my $db = $self->map_path($uuid);
3160        my $db_lock = "$db.lock";
3161        my $sig;
3162        $update_ref ||= 0;
3163        if ($update_ref) {
3164                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3165                            $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
3166        }
3167        mkfile($db);
3168
3169        $LOCKFILES{$db_lock} = 1;
3170        my $sync;
3171        # both of these options make our .rev_db file very, very important
3172        # and we can't afford to lose it because rebuild() won't work
3173        if ($self->use_svm_props || $self->no_metadata) {
3174                $sync = 1;
3175                copy($db, $db_lock) or die "rev_map_set(@_): ",
3176                                           "Failed to copy: ",
3177                                           "$db => $db_lock ($!)\n";
3178        } else {
3179                rename $db, $db_lock or die "rev_map_set(@_): ",
3180                                            "Failed to rename: ",
3181                                            "$db => $db_lock ($!)\n";
3182        }
3183
3184        sysopen(my $fh, $db_lock, O_RDWR | O_CREAT)
3185             or croak "Couldn't open $db_lock: $!\n";
3186        $update_ref eq 'reset' ? _rev_map_reset($fh, $rev, $commit) :
3187                                 _rev_map_set($fh, $rev, $commit);
3188        if ($sync) {
3189                $fh->flush or die "Couldn't flush $db_lock: $!\n";
3190                $fh->sync or die "Couldn't sync $db_lock: $!\n";
3191        }
3192        close $fh or croak $!;
3193        if ($update_ref) {
3194                $_head = $self;
3195                my $note = "";
3196                $note = " ($update_ref)" if ($update_ref !~ /^\d*$/);
3197                command_noisy('update-ref', '-m', "r$rev$note",
3198                              $self->refname, $commit);
3199        }
3200        rename $db_lock, $db or die "rev_map_set(@_): ", "Failed to rename: ",
3201                                    "$db_lock => $db ($!)\n";
3202        delete $LOCKFILES{$db_lock};
3203        if ($update_ref) {
3204                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3205                            $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
3206                kill $sig, $$ if defined $sig;
3207        }
3208}
3209
3210# If want_commit, this will return an array of (rev, commit) where
3211# commit _must_ be a valid commit in the archive.
3212# Otherwise, it'll return the max revision (whether or not the
3213# commit is valid or just a 0x40 placeholder).
3214sub rev_map_max {
3215        my ($self, $want_commit) = @_;
3216        $self->rebuild;
3217        my ($r, $c) = $self->rev_map_max_norebuild($want_commit);
3218        $want_commit ? ($r, $c) : $r;
3219}
3220
3221sub rev_map_max_norebuild {
3222        my ($self, $want_commit) = @_;
3223        my $map_path = $self->map_path;
3224        stat $map_path or return $want_commit ? (0, undef) : 0;
3225        sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3226        binmode $fh or croak "binmode: $!";
3227        my $size = (stat($fh))[7];
3228        ($size % 24) == 0 or croak "inconsistent size: $size";
3229
3230        if ($size == 0) {
3231                close $fh or croak "close: $!";
3232                return $want_commit ? (0, undef) : 0;
3233        }
3234
3235        sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3236        sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3237        my ($r, $c) = unpack(rev_map_fmt, $buf);
3238        if ($want_commit && $c eq ('0' x40)) {
3239                if ($size < 48) {
3240                        return $want_commit ? (0, undef) : 0;
3241                }
3242                sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3243                sysread($fh, $buf, 24) == 24 or croak "read: $!";
3244                ($r, $c) = unpack(rev_map_fmt, $buf);
3245                if ($c eq ('0'x40)) {
3246                        croak "Penultimate record is all-zeroes in $map_path";
3247                }
3248        }
3249        close $fh or croak "close: $!";
3250        $want_commit ? ($r, $c) : $r;
3251}
3252
3253sub rev_map_get {
3254        my ($self, $rev, $uuid) = @_;
3255        my $map_path = $self->map_path($uuid);
3256        return undef unless -e $map_path;
3257
3258        sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3259        my $c = _rev_map_get($fh, $rev);
3260        close($fh) or croak "close: $!";
3261        $c
3262}
3263
3264sub _rev_map_get {
3265        my ($fh, $rev) = @_;
3266
3267        binmode $fh or croak "binmode: $!";
3268        my $size = (stat($fh))[7];
3269        ($size % 24) == 0 or croak "inconsistent size: $size";
3270
3271        if ($size == 0) {
3272                return undef;
3273        }
3274
3275        my ($l, $u) = (0, $size - 24);
3276        my ($r, $c, $buf);
3277
3278        while ($l <= $u) {
3279                my $i = int(($l/24 + $u/24) / 2) * 24;
3280                sysseek($fh, $i, SEEK_SET) or croak "seek: $!";
3281                sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3282                my ($r, $c) = unpack(rev_map_fmt, $buf);
3283
3284                if ($r < $rev) {
3285                        $l = $i + 24;
3286                } elsif ($r > $rev) {
3287                        $u = $i - 24;
3288                } else { # $r == $rev
3289                        return $c eq ('0' x 40) ? undef : $c;
3290                }
3291        }
3292        undef;
3293}
3294
3295# Finds the first svn revision that exists on (if $eq_ok is true) or
3296# before $rev for the current branch.  It will not search any lower
3297# than $min_rev.  Returns the git commit hash and svn revision number
3298# if found, else (undef, undef).
3299sub find_rev_before {
3300        my ($self, $rev, $eq_ok, $min_rev) = @_;
3301        --$rev unless $eq_ok;
3302        $min_rev ||= 1;
3303        my $max_rev = $self->rev_map_max;
3304        $rev = $max_rev if ($rev > $max_rev);
3305        while ($rev >= $min_rev) {
3306                if (my $c = $self->rev_map_get($rev)) {
3307                        return ($rev, $c);
3308                }
3309                --$rev;
3310        }
3311        return (undef, undef);
3312}
3313
3314# Finds the first svn revision that exists on (if $eq_ok is true) or
3315# after $rev for the current branch.  It will not search any higher
3316# than $max_rev.  Returns the git commit hash and svn revision number
3317# if found, else (undef, undef).
3318sub find_rev_after {
3319        my ($self, $rev, $eq_ok, $max_rev) = @_;
3320        ++$rev unless $eq_ok;
3321        $max_rev ||= $self->rev_map_max;
3322        while ($rev <= $max_rev) {
3323                if (my $c = $self->rev_map_get($rev)) {
3324                        return ($rev, $c);
3325                }
3326                ++$rev;
3327        }
3328        return (undef, undef);
3329}
3330
3331sub _new {
3332        my ($class, $repo_id, $ref_id, $path) = @_;
3333        unless (defined $repo_id && length $repo_id) {
3334                $repo_id = $Git::SVN::default_repo_id;
3335        }
3336        unless (defined $ref_id && length $ref_id) {
3337                $_prefix = '' unless defined($_prefix);
3338                $_[2] = $ref_id =
3339                             "refs/remotes/$_prefix$Git::SVN::default_ref_id";
3340        }
3341        $_[1] = $repo_id;
3342        my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
3343
3344        # Older repos imported by us used $GIT_DIR/svn/foo instead of
3345        # $GIT_DIR/svn/refs/remotes/foo when tracking refs/remotes/foo
3346        if ($ref_id =~ m{^refs/remotes/(.*)}) {
3347                my $old_dir = "$ENV{GIT_DIR}/svn/$1";
3348                if (-d $old_dir && ! -d $dir) {
3349                        $dir = $old_dir;
3350                }
3351        }
3352
3353        $_[3] = $path = '' unless (defined $path);
3354        mkpath([$dir]);
3355        bless {
3356                ref_id => $ref_id, dir => $dir, index => "$dir/index",
3357                path => $path, config => "$ENV{GIT_DIR}/svn/config",
3358                map_root => "$dir/.rev_map", repo_id => $repo_id }, $class;
3359}
3360
3361# for read-only access of old .rev_db formats
3362sub unlink_rev_db_symlink {
3363        my ($self) = @_;
3364        my $link = $self->rev_db_path;
3365        $link =~ s/\.[\w-]+$// or croak "missing UUID at the end of $link";
3366        if (-l $link) {
3367                unlink $link or croak "unlink: $link failed!";
3368        }
3369}
3370
3371sub rev_db_path {
3372        my ($self, $uuid) = @_;
3373        my $db_path = $self->map_path($uuid);
3374        $db_path =~ s{/\.rev_map\.}{/\.rev_db\.}
3375            or croak "map_path: $db_path does not contain '/.rev_map.' !";
3376        $db_path;
3377}
3378
3379# the new replacement for .rev_db
3380sub map_path {
3381        my ($self, $uuid) = @_;
3382        $uuid ||= $self->ra_uuid;
3383        "$self->{map_root}.$uuid";
3384}
3385
3386sub uri_encode {
3387        my ($f) = @_;
3388        $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
3389        $f
3390}
3391
3392sub remove_username {
3393        $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
3394}
3395
3396package Git::SVN::Prompt;
3397use strict;
3398use warnings;
3399require SVN::Core;
3400use vars qw/$_no_auth_cache $_username/;
3401
3402sub simple {
3403        my ($cred, $realm, $default_username, $may_save, $pool) = @_;
3404        $may_save = undef if $_no_auth_cache;
3405        $default_username = $_username if defined $_username;
3406        if (defined $default_username && length $default_username) {
3407                if (defined $realm && length $realm) {
3408                        print STDERR "Authentication realm: $realm\n";
3409                        STDERR->flush;
3410                }
3411                $cred->username($default_username);
3412        } else {
3413                username($cred, $realm, $may_save, $pool);
3414        }
3415        $cred->password(_read_password("Password for '" .
3416                                       $cred->username . "': ", $realm));
3417        $cred->may_save($may_save);
3418        $SVN::_Core::SVN_NO_ERROR;
3419}
3420
3421sub ssl_server_trust {
3422        my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
3423        $may_save = undef if $_no_auth_cache;
3424        print STDERR "Error validating server certificate for '$realm':\n";
3425        {
3426                no warnings 'once';
3427                # All variables SVN::Auth::SSL::* are used only once,
3428                # so we're shutting up Perl warnings about this.
3429                if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
3430                        print STDERR " - The certificate is not issued ",
3431                            "by a trusted authority. Use the\n",
3432                            "   fingerprint to validate ",
3433                            "the certificate manually!\n";
3434                }
3435                if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
3436                        print STDERR " - The certificate hostname ",
3437                            "does not match.\n";
3438                }
3439                if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
3440                        print STDERR " - The certificate is not yet valid.\n";
3441                }
3442                if ($failures & $SVN::Auth::SSL::EXPIRED) {
3443                        print STDERR " - The certificate has expired.\n";
3444                }
3445                if ($failures & $SVN::Auth::SSL::OTHER) {
3446                        print STDERR " - The certificate has ",
3447                            "an unknown error.\n";
3448                }
3449        } # no warnings 'once'
3450        printf STDERR
3451                "Certificate information:\n".
3452                " - Hostname: %s\n".
3453                " - Valid: from %s until %s\n".
3454                " - Issuer: %s\n".
3455                " - Fingerprint: %s\n",
3456                map $cert_info->$_, qw(hostname valid_from valid_until
3457                                       issuer_dname fingerprint);
3458        my $choice;
3459prompt:
3460        print STDERR $may_save ?
3461              "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
3462              "(R)eject or accept (t)emporarily? ";
3463        STDERR->flush;
3464        $choice = lc(substr(<STDIN> || 'R', 0, 1));
3465        if ($choice =~ /^t$/i) {
3466                $cred->may_save(undef);
3467        } elsif ($choice =~ /^r$/i) {
3468                return -1;
3469        } elsif ($may_save && $choice =~ /^p$/i) {
3470                $cred->may_save($may_save);
3471        } else {
3472                goto prompt;
3473        }
3474        $cred->accepted_failures($failures);
3475        $SVN::_Core::SVN_NO_ERROR;
3476}
3477
3478sub ssl_client_cert {
3479        my ($cred, $realm, $may_save, $pool) = @_;
3480        $may_save = undef if $_no_auth_cache;
3481        print STDERR "Client certificate filename: ";
3482        STDERR->flush;
3483        chomp(my $filename = <STDIN>);
3484        $cred->cert_file($filename);
3485        $cred->may_save($may_save);
3486        $SVN::_Core::SVN_NO_ERROR;
3487}
3488
3489sub ssl_client_cert_pw {
3490        my ($cred, $realm, $may_save, $pool) = @_;
3491        $may_save = undef if $_no_auth_cache;
3492        $cred->password(_read_password("Password: ", $realm));
3493        $cred->may_save($may_save);
3494        $SVN::_Core::SVN_NO_ERROR;
3495}
3496
3497sub username {
3498        my ($cred, $realm, $may_save, $pool) = @_;
3499        $may_save = undef if $_no_auth_cache;
3500        if (defined $realm && length $realm) {
3501                print STDERR "Authentication realm: $realm\n";
3502        }
3503        my $username;
3504        if (defined $_username) {
3505                $username = $_username;
3506        } else {
3507                print STDERR "Username: ";
3508                STDERR->flush;
3509                chomp($username = <STDIN>);
3510        }
3511        $cred->username($username);
3512        $cred->may_save($may_save);
3513        $SVN::_Core::SVN_NO_ERROR;
3514}
3515
3516sub _read_password {
3517        my ($prompt, $realm) = @_;
3518        print STDERR $prompt;
3519        STDERR->flush;
3520        require Term::ReadKey;
3521        Term::ReadKey::ReadMode('noecho');
3522        my $password = '';
3523        while (defined(my $key = Term::ReadKey::ReadKey(0))) {
3524                last if $key =~ /[\012\015]/; # \n\r
3525                $password .= $key;
3526        }
3527        Term::ReadKey::ReadMode('restore');
3528        print STDERR "\n";
3529        STDERR->flush;
3530        $password;
3531}
3532
3533package SVN::Git::Fetcher;
3534use vars qw/@ISA/;
3535use strict;
3536use warnings;
3537use Carp qw/croak/;
3538use File::Temp qw/tempfile/;
3539use IO::File qw//;
3540use vars qw/$_ignore_regex/;
3541
3542# file baton members: path, mode_a, mode_b, pool, fh, blob, base
3543sub new {
3544        my ($class, $git_svn, $switch_path) = @_;
3545        my $self = SVN::Delta::Editor->new;
3546        bless $self, $class;
3547        if (exists $git_svn->{last_commit}) {
3548                $self->{c} = $git_svn->{last_commit};
3549                $self->{empty_symlinks} =
3550                                  _mark_empty_symlinks($git_svn, $switch_path);
3551        }
3552        $self->{ignore_regex} = eval { command_oneline('config', '--get',
3553                             "svn-remote.$git_svn->{repo_id}.ignore-paths") };
3554        $self->{empty} = {};
3555        $self->{dir_prop} = {};
3556        $self->{file_prop} = {};
3557        $self->{absent_dir} = {};
3558        $self->{absent_file} = {};
3559        $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
3560        $self;
3561}
3562
3563# this uses the Ra object, so it must be called before do_{switch,update},
3564# not inside them (when the Git::SVN::Fetcher object is passed) to
3565# do_{switch,update}
3566sub _mark_empty_symlinks {
3567        my ($git_svn, $switch_path) = @_;
3568        my $bool = Git::config_bool('svn.brokenSymlinkWorkaround');
3569        return {} if (!defined($bool)) || (defined($bool) && ! $bool);
3570
3571        my %ret;
3572        my ($rev, $cmt) = $git_svn->last_rev_commit;
3573        return {} unless ($rev && $cmt);
3574
3575        # allow the warning to be printed for each revision we fetch to
3576        # ensure the user sees it.  The user can also disable the workaround
3577        # on the repository even while git svn is running and the next
3578        # revision fetched will skip this expensive function.
3579        my $printed_warning;
3580        chomp(my $empty_blob = `git hash-object -t blob --stdin < /dev/null`);
3581        my ($ls, $ctx) = command_output_pipe(qw/ls-tree -r -z/, $cmt);
3582        local $/ = "\0";
3583        my $pfx = defined($switch_path) ? $switch_path : $git_svn->{path};
3584        $pfx .= '/' if length($pfx);
3585        while (<$ls>) {
3586                chomp;
3587                s/\A100644 blob $empty_blob\t//o or next;
3588                unless ($printed_warning) {
3589                        print STDERR "Scanning for empty symlinks, ",
3590                                     "this may take a while if you have ",
3591                                     "many empty files\n",
3592                                     "You may disable this with `",
3593                                     "git config svn.brokenSymlinkWorkaround ",
3594                                     "false'.\n",
3595                                     "This may be done in a different ",
3596                                     "terminal without restarting ",
3597                                     "git svn\n";
3598                        $printed_warning = 1;
3599                }
3600                my $path = $_;
3601                my (undef, $props) =
3602                               $git_svn->ra->get_file($pfx.$path, $rev, undef);
3603                if ($props->{'svn:special'}) {
3604                        $ret{$path} = 1;
3605                }
3606        }
3607        command_close_pipe($ls, $ctx);
3608        \%ret;
3609}
3610
3611# returns true if a given path is inside a ".git" directory
3612sub in_dot_git {
3613        $_[0] =~ m{(?:^|/)\.git(?:/|$)};
3614}
3615
3616# return value: 0 -- don't ignore, 1 -- ignore
3617sub is_path_ignored {
3618        my ($self, $path) = @_;
3619        return 1 if in_dot_git($path);
3620        return 1 if defined($self->{ignore_regex}) &&
3621                    $path =~ m!$self->{ignore_regex}!;
3622        return 0 unless defined($_ignore_regex);
3623        return 1 if $path =~ m!$_ignore_regex!o;
3624        return 0;
3625}
3626
3627sub set_path_strip {
3628        my ($self, $path) = @_;
3629        $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
3630}
3631
3632sub open_root {
3633        { path => '' };
3634}
3635
3636sub open_directory {
3637        my ($self, $path, $pb, $rev) = @_;
3638        { path => $path };
3639}
3640
3641sub git_path {
3642        my ($self, $path) = @_;
3643        if ($self->{path_strip}) {
3644                $path =~ s!$self->{path_strip}!! or
3645                  die "Failed to strip path '$path' ($self->{path_strip})\n";
3646        }
3647        $path;
3648}
3649
3650sub delete_entry {
3651        my ($self, $path, $rev, $pb) = @_;
3652        return undef if $self->is_path_ignored($path);
3653
3654        my $gpath = $self->git_path($path);
3655        return undef if ($gpath eq '');
3656
3657        # remove entire directories.
3658        my ($tree) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
3659                         =~ /\A040000 tree ([a-f\d]{40})\t\Q$gpath\E\0/);
3660        if ($tree) {
3661                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3662                                                     -r --name-only -z/,
3663                                                     $tree);
3664                local $/ = "\0";
3665                while (<$ls>) {
3666                        chomp;
3667                        my $rmpath = "$gpath/$_";
3668                        $self->{gii}->remove($rmpath);
3669                        print "\tD\t$rmpath\n" unless $::_q;
3670                }
3671                print "\tD\t$gpath/\n" unless $::_q;
3672                command_close_pipe($ls, $ctx);
3673                $self->{empty}->{$path} = 0
3674        } else {
3675                $self->{gii}->remove($gpath);
3676                print "\tD\t$gpath\n" unless $::_q;
3677        }
3678        undef;
3679}
3680
3681sub open_file {
3682        my ($self, $path, $pb, $rev) = @_;
3683        my ($mode, $blob);
3684
3685        goto out if $self->is_path_ignored($path);
3686
3687        my $gpath = $self->git_path($path);
3688        ($mode, $blob) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
3689                             =~ /\A(\d{6}) blob ([a-f\d]{40})\t\Q$gpath\E\0/);
3690        unless (defined $mode && defined $blob) {
3691                die "$path was not found in commit $self->{c} (r$rev)\n";
3692        }
3693        if ($mode eq '100644' && $self->{empty_symlinks}->{$path}) {
3694                $mode = '120000';
3695        }
3696out:
3697        { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
3698          pool => SVN::Pool->new, action => 'M' };
3699}
3700
3701sub add_file {
3702        my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
3703        my $mode;
3704
3705        if (!$self->is_path_ignored($path)) {
3706                my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3707                delete $self->{empty}->{$dir};
3708                $mode = '100644';
3709        }
3710        { path => $path, mode_a => $mode, mode_b => $mode,
3711          pool => SVN::Pool->new, action => 'A' };
3712}
3713
3714sub add_directory {
3715        my ($self, $path, $cp_path, $cp_rev) = @_;
3716        goto out if $self->is_path_ignored($path);
3717        my $gpath = $self->git_path($path);
3718        if ($gpath eq '') {
3719                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3720                                                     -r --name-only -z/,
3721                                                     $self->{c});
3722                local $/ = "\0";
3723                while (<$ls>) {
3724                        chomp;
3725                        $self->{gii}->remove($_);
3726                        print "\tD\t$_\n" unless $::_q;
3727                }
3728                command_close_pipe($ls, $ctx);
3729                $self->{empty}->{$path} = 0;
3730        }
3731        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3732        delete $self->{empty}->{$dir};
3733        $self->{empty}->{$path} = 1;
3734out:
3735        { path => $path };
3736}
3737
3738sub change_dir_prop {
3739        my ($self, $db, $prop, $value) = @_;
3740        return undef if $self->is_path_ignored($db->{path});
3741        $self->{dir_prop}->{$db->{path}} ||= {};
3742        $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
3743        undef;
3744}
3745
3746sub absent_directory {
3747        my ($self, $path, $pb) = @_;
3748        return undef if $self->is_path_ignored($path);
3749        $self->{absent_dir}->{$pb->{path}} ||= [];
3750        push @{$self->{absent_dir}->{$pb->{path}}}, $path;
3751        undef;
3752}
3753
3754sub absent_file {
3755        my ($self, $path, $pb) = @_;
3756        return undef if $self->is_path_ignored($path);
3757        $self->{absent_file}->{$pb->{path}} ||= [];
3758        push @{$self->{absent_file}->{$pb->{path}}}, $path;
3759        undef;
3760}
3761
3762sub change_file_prop {
3763        my ($self, $fb, $prop, $value) = @_;
3764        return undef if $self->is_path_ignored($fb->{path});
3765        if ($prop eq 'svn:executable') {
3766                if ($fb->{mode_b} != 120000) {
3767                        $fb->{mode_b} = defined $value ? 100755 : 100644;
3768                }
3769        } elsif ($prop eq 'svn:special') {
3770                $fb->{mode_b} = defined $value ? 120000 : 100644;
3771        } else {
3772                $self->{file_prop}->{$fb->{path}} ||= {};
3773                $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
3774        }
3775        undef;
3776}
3777
3778sub apply_textdelta {
3779        my ($self, $fb, $exp) = @_;
3780        return undef if $self->is_path_ignored($fb->{path});
3781        my $fh = $::_repository->temp_acquire('svn_delta');
3782        # $fh gets auto-closed() by SVN::TxDelta::apply(),
3783        # (but $base does not,) so dup() it for reading in close_file
3784        open my $dup, '<&', $fh or croak $!;
3785        my $base = $::_repository->temp_acquire('git_blob');
3786
3787        if ($fb->{blob}) {
3788                my ($base_is_link, $size);
3789
3790                if ($fb->{mode_a} eq '120000' &&
3791                    ! $self->{empty_symlinks}->{$fb->{path}}) {
3792                        print $base 'link ' or die "print $!\n";
3793                        $base_is_link = 1;
3794                }
3795        retry:
3796                $size = $::_repository->cat_blob($fb->{blob}, $base);
3797                die "Failed to read object $fb->{blob}" if ($size < 0);
3798
3799                if (defined $exp) {
3800                        seek $base, 0, 0 or croak $!;
3801                        my $got = ::md5sum($base);
3802                        if ($got ne $exp) {
3803                                my $err = "Checksum mismatch: ".
3804                                       "$fb->{path} $fb->{blob}\n" .
3805                                       "expected: $exp\n" .
3806                                       "     got: $got\n";
3807                                if ($base_is_link) {
3808                                        warn $err,
3809                                             "Retrying... (possibly ",
3810                                             "a bad symlink from SVN)\n";
3811                                        $::_repository->temp_reset($base);
3812                                        $base_is_link = 0;
3813                                        goto retry;
3814                                }
3815                                die $err;
3816                        }
3817                }
3818        }
3819        seek $base, 0, 0 or croak $!;
3820        $fb->{fh} = $fh;
3821        $fb->{base} = $base;
3822        [ SVN::TxDelta::apply($base, $dup, undef, $fb->{path}, $fb->{pool}) ];
3823}
3824
3825sub close_file {
3826        my ($self, $fb, $exp) = @_;
3827        return undef if $self->is_path_ignored($fb->{path});
3828
3829        my $hash;
3830        my $path = $self->git_path($fb->{path});
3831        if (my $fh = $fb->{fh}) {
3832                if (defined $exp) {
3833                        seek($fh, 0, 0) or croak $!;
3834                        my $got = ::md5sum($fh);
3835                        if ($got ne $exp) {
3836                                die "Checksum mismatch: $path\n",
3837                                    "expected: $exp\n    got: $got\n";
3838                        }
3839                }
3840                if ($fb->{mode_b} == 120000) {
3841                        sysseek($fh, 0, 0) or croak $!;
3842                        my $rd = sysread($fh, my $buf, 5);
3843
3844                        if (!defined $rd) {
3845                                croak "sysread: $!\n";
3846                        } elsif ($rd == 0) {
3847                                warn "$path has mode 120000",
3848                                     " but it points to nothing\n",
3849                                     "converting to an empty file with mode",
3850                                     " 100644\n";
3851                                $fb->{mode_b} = '100644';
3852                        } elsif ($buf ne 'link ') {
3853                                warn "$path has mode 120000",
3854                                     " but is not a link\n";
3855                        } else {
3856                                my $tmp_fh = $::_repository->temp_acquire(
3857                                        'svn_hash');
3858                                my $res;
3859                                while ($res = sysread($fh, my $str, 1024)) {
3860                                        my $out = syswrite($tmp_fh, $str, $res);
3861                                        defined($out) && $out == $res
3862                                                or croak("write ",
3863                                                        Git::temp_path($tmp_fh),
3864                                                        ": $!\n");
3865                                }
3866                                defined $res or croak $!;
3867
3868                                ($fh, $tmp_fh) = ($tmp_fh, $fh);
3869                                Git::temp_release($tmp_fh, 1);
3870                        }
3871                }
3872
3873                $hash = $::_repository->hash_and_insert_object(
3874                                Git::temp_path($fh));
3875                $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
3876
3877                Git::temp_release($fb->{base}, 1);
3878                Git::temp_release($fh, 1);
3879        } else {
3880                $hash = $fb->{blob} or die "no blob information\n";
3881        }
3882        $fb->{pool}->clear;
3883        $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
3884        print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
3885        undef;
3886}
3887
3888sub abort_edit {
3889        my $self = shift;
3890        $self->{nr} = $self->{gii}->{nr};
3891        delete $self->{gii};
3892        $self->SUPER::abort_edit(@_);
3893}
3894
3895sub close_edit {
3896        my $self = shift;
3897        $self->{git_commit_ok} = 1;
3898        $self->{nr} = $self->{gii}->{nr};
3899        delete $self->{gii};
3900        $self->SUPER::close_edit(@_);
3901}
3902
3903package SVN::Git::Editor;
3904use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
3905use strict;
3906use warnings;
3907use Carp qw/croak/;
3908use IO::File;
3909
3910sub new {
3911        my ($class, $opts) = @_;
3912        foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
3913                die "$_ required!\n" unless (defined $opts->{$_});
3914        }
3915
3916        my $pool = SVN::Pool->new;
3917        my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
3918        my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
3919                                     $opts->{r}, $mods);
3920
3921        # $opts->{ra} functions should not be used after this:
3922        my @ce  = $opts->{ra}->get_commit_editor($opts->{log},
3923                                                $opts->{editor_cb}, $pool);
3924        my $self = SVN::Delta::Editor->new(@ce, $pool);
3925        bless $self, $class;
3926        foreach (qw/svn_path r tree_a tree_b/) {
3927                $self->{$_} = $opts->{$_};
3928        }
3929        $self->{url} = $opts->{ra}->{url};
3930        $self->{mods} = $mods;
3931        $self->{types} = $types;
3932        $self->{pool} = $pool;
3933        $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
3934        $self->{rm} = { };
3935        $self->{path_prefix} = length $self->{svn_path} ?
3936                               "$self->{svn_path}/" : '';
3937        $self->{config} = $opts->{config};
3938        return $self;
3939}
3940
3941sub generate_diff {
3942        my ($tree_a, $tree_b) = @_;
3943        my @diff_tree = qw(diff-tree -z -r);
3944        if ($_cp_similarity) {
3945                push @diff_tree, "-C$_cp_similarity";
3946        } else {
3947                push @diff_tree, '-C';
3948        }
3949        push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
3950        push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
3951        push @diff_tree, $tree_a, $tree_b;
3952        my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
3953        local $/ = "\0";
3954        my $state = 'meta';
3955        my @mods;
3956        while (<$diff_fh>) {
3957                chomp $_; # this gets rid of the trailing "\0"
3958                if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
3959                                        ($::sha1)\s($::sha1)\s
3960                                        ([MTCRAD])\d*$/xo) {
3961                        push @mods, {   mode_a => $1, mode_b => $2,
3962                                        sha1_a => $3, sha1_b => $4,
3963                                        chg => $5 };
3964                        if ($5 =~ /^(?:C|R)$/) {
3965                                $state = 'file_a';
3966                        } else {
3967                                $state = 'file_b';
3968                        }
3969                } elsif ($state eq 'file_a') {
3970                        my $x = $mods[$#mods] or croak "Empty array\n";
3971                        if ($x->{chg} !~ /^(?:C|R)$/) {
3972                                croak "Error parsing $_, $x->{chg}\n";
3973                        }
3974                        $x->{file_a} = $_;
3975                        $state = 'file_b';
3976                } elsif ($state eq 'file_b') {
3977                        my $x = $mods[$#mods] or croak "Empty array\n";
3978                        if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
3979                                croak "Error parsing $_, $x->{chg}\n";
3980                        }
3981                        if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
3982                                croak "Error parsing $_, $x->{chg}\n";
3983                        }
3984                        $x->{file_b} = $_;
3985                        $state = 'meta';
3986                } else {
3987                        croak "Error parsing $_\n";
3988                }
3989        }
3990        command_close_pipe($diff_fh, $ctx);
3991        \@mods;
3992}
3993
3994sub check_diff_paths {
3995        my ($ra, $pfx, $rev, $mods) = @_;
3996        my %types;
3997        $pfx .= '/' if length $pfx;
3998
3999        sub type_diff_paths {
4000                my ($ra, $types, $path, $rev) = @_;
4001                my @p = split m#/+#, $path;
4002                my $c = shift @p;
4003                unless (defined $types->{$c}) {
4004                        $types->{$c} = $ra->check_path($c, $rev);
4005                }
4006                while (@p) {
4007                        $c .= '/' . shift @p;
4008                        next if defined $types->{$c};
4009                        $types->{$c} = $ra->check_path($c, $rev);
4010                }
4011        }
4012
4013        foreach my $m (@$mods) {
4014                foreach my $f (qw/file_a file_b/) {
4015                        next unless defined $m->{$f};
4016                        my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
4017                        if (length $pfx.$dir && ! defined $types{$dir}) {
4018                                type_diff_paths($ra, \%types, $pfx.$dir, $rev);
4019                        }
4020                }
4021        }
4022        \%types;
4023}
4024
4025sub split_path {
4026        return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
4027}
4028
4029sub repo_path {
4030        my ($self, $path) = @_;
4031        $self->{path_prefix}.(defined $path ? $path : '');
4032}
4033
4034sub url_path {
4035        my ($self, $path) = @_;
4036        if ($self->{url} =~ m#^https?://#) {
4037                $path =~ s!([^~a-zA-Z0-9_./-])!uc sprintf("%%%02x",ord($1))!eg;
4038        }
4039        $self->{url} . '/' . $self->repo_path($path);
4040}
4041
4042sub rmdirs {
4043        my ($self) = @_;
4044        my $rm = $self->{rm};
4045        delete $rm->{''}; # we never delete the url we're tracking
4046        return unless %$rm;
4047
4048        foreach (keys %$rm) {
4049                my @d = split m#/#, $_;
4050                my $c = shift @d;
4051                $rm->{$c} = 1;
4052                while (@d) {
4053                        $c .= '/' . shift @d;
4054                        $rm->{$c} = 1;
4055                }
4056        }
4057        delete $rm->{$self->{svn_path}};
4058        delete $rm->{''}; # we never delete the url we're tracking
4059        return unless %$rm;
4060
4061        my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
4062                                             $self->{tree_b});
4063        local $/ = "\0";
4064        while (<$fh>) {
4065                chomp;
4066                my @dn = split m#/#, $_;
4067                while (pop @dn) {
4068                        delete $rm->{join '/', @dn};
4069                }
4070                unless (%$rm) {
4071                        close $fh;
4072                        return;
4073                }
4074        }
4075        command_close_pipe($fh, $ctx);
4076
4077        my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
4078        foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
4079                $self->close_directory($bat->{$d}, $p);
4080                my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
4081                print "\tD+\t$d/\n" unless $::_q;
4082                $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
4083                delete $bat->{$d};
4084        }
4085}
4086
4087sub open_or_add_dir {
4088        my ($self, $full_path, $baton) = @_;
4089        my $t = $self->{types}->{$full_path};
4090        if (!defined $t) {
4091                die "$full_path not known in r$self->{r} or we have a bug!\n";
4092        }
4093        {
4094                no warnings 'once';
4095                # SVN::Node::none and SVN::Node::file are used only once,
4096                # so we're shutting up Perl's warnings about them.
4097                if ($t == $SVN::Node::none) {
4098                        return $self->add_directory($full_path, $baton,
4099                            undef, -1, $self->{pool});
4100                } elsif ($t == $SVN::Node::dir) {
4101                        return $self->open_directory($full_path, $baton,
4102                            $self->{r}, $self->{pool});
4103                } # no warnings 'once'
4104                print STDERR "$full_path already exists in repository at ",
4105                    "r$self->{r} and it is not a directory (",
4106                    ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
4107        } # no warnings 'once'
4108        exit 1;
4109}
4110
4111sub ensure_path {
4112        my ($self, $path) = @_;
4113        my $bat = $self->{bat};
4114        my $repo_path = $self->repo_path($path);
4115        return $bat->{''} unless (length $repo_path);
4116        my @p = split m#/+#, $repo_path;
4117        my $c = shift @p;
4118        $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
4119        while (@p) {
4120                my $c0 = $c;
4121                $c .= '/' . shift @p;
4122                $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
4123        }
4124        return $bat->{$c};
4125}
4126
4127# Subroutine to convert a globbing pattern to a regular expression.
4128# From perl cookbook.
4129sub glob2pat {
4130        my $globstr = shift;
4131        my %patmap = ('*' => '.*', '?' => '.', '[' => '[', ']' => ']');
4132        $globstr =~ s{(.)} { $patmap{$1} || "\Q$1" }ge;
4133        return '^' . $globstr . '$';
4134}
4135
4136sub check_autoprop {
4137        my ($self, $pattern, $properties, $file, $fbat) = @_;
4138        # Convert the globbing pattern to a regular expression.
4139        my $regex = glob2pat($pattern);
4140        # Check if the pattern matches the file name.
4141        if($file =~ m/($regex)/) {
4142                # Parse the list of properties to set.
4143                my @props = split(/;/, $properties);
4144                foreach my $prop (@props) {
4145                        # Parse 'name=value' syntax and set the property.
4146                        if ($prop =~ /([^=]+)=(.*)/) {
4147                                my ($n,$v) = ($1,$2);
4148                                for ($n, $v) {
4149                                        s/^\s+//; s/\s+$//;
4150                                }
4151                                $self->change_file_prop($fbat, $n, $v);
4152                        }
4153                }
4154        }
4155}
4156
4157sub apply_autoprops {
4158        my ($self, $file, $fbat) = @_;
4159        my $conf_t = ${$self->{config}}{'config'};
4160        no warnings 'once';
4161        # Check [miscellany]/enable-auto-props in svn configuration.
4162        if (SVN::_Core::svn_config_get_bool(
4163                $conf_t,
4164                $SVN::_Core::SVN_CONFIG_SECTION_MISCELLANY,
4165                $SVN::_Core::SVN_CONFIG_OPTION_ENABLE_AUTO_PROPS,
4166                0)) {
4167                # Auto-props are enabled.  Enumerate them to look for matches.
4168                my $callback = sub {
4169                        $self->check_autoprop($_[0], $_[1], $file, $fbat);
4170                };
4171                SVN::_Core::svn_config_enumerate(
4172                        $conf_t,
4173                        $SVN::_Core::SVN_CONFIG_SECTION_AUTO_PROPS,
4174                        $callback);
4175        }
4176}
4177
4178sub A {
4179        my ($self, $m) = @_;
4180        my ($dir, $file) = split_path($m->{file_b});
4181        my $pbat = $self->ensure_path($dir);
4182        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4183                                        undef, -1);
4184        print "\tA\t$m->{file_b}\n" unless $::_q;
4185        $self->apply_autoprops($file, $fbat);
4186        $self->chg_file($fbat, $m);
4187        $self->close_file($fbat,undef,$self->{pool});
4188}
4189
4190sub C {
4191        my ($self, $m) = @_;
4192        my ($dir, $file) = split_path($m->{file_b});
4193        my $pbat = $self->ensure_path($dir);
4194        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4195                                $self->url_path($m->{file_a}), $self->{r});
4196        print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
4197        $self->chg_file($fbat, $m);
4198        $self->close_file($fbat,undef,$self->{pool});
4199}
4200
4201sub delete_entry {
4202        my ($self, $path, $pbat) = @_;
4203        my $rpath = $self->repo_path($path);
4204        my ($dir, $file) = split_path($rpath);
4205        $self->{rm}->{$dir} = 1;
4206        $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
4207}
4208
4209sub R {
4210        my ($self, $m) = @_;
4211        my ($dir, $file) = split_path($m->{file_b});
4212        my $pbat = $self->ensure_path($dir);
4213        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4214                                $self->url_path($m->{file_a}), $self->{r});
4215        print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
4216        $self->apply_autoprops($file, $fbat);
4217        $self->chg_file($fbat, $m);
4218        $self->close_file($fbat,undef,$self->{pool});
4219
4220        ($dir, $file) = split_path($m->{file_a});
4221        $pbat = $self->ensure_path($dir);
4222        $self->delete_entry($m->{file_a}, $pbat);
4223}
4224
4225sub M {
4226        my ($self, $m) = @_;
4227        my ($dir, $file) = split_path($m->{file_b});
4228        my $pbat = $self->ensure_path($dir);
4229        my $fbat = $self->open_file($self->repo_path($m->{file_b}),
4230                                $pbat,$self->{r},$self->{pool});
4231        print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
4232        $self->chg_file($fbat, $m);
4233        $self->close_file($fbat,undef,$self->{pool});
4234}
4235
4236sub T { shift->M(@_) }
4237
4238sub change_file_prop {
4239        my ($self, $fbat, $pname, $pval) = @_;
4240        $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
4241}
4242
4243sub _chg_file_get_blob ($$$$) {
4244        my ($self, $fbat, $m, $which) = @_;
4245        my $fh = $::_repository->temp_acquire("git_blob_$which");
4246        if ($m->{"mode_$which"} =~ /^120/) {
4247                print $fh 'link ' or croak $!;
4248                $self->change_file_prop($fbat,'svn:special','*');
4249        } elsif ($m->{mode_a} =~ /^120/ && $m->{"mode_$which"} !~ /^120/) {
4250                $self->change_file_prop($fbat,'svn:special',undef);
4251        }
4252        my $blob = $m->{"sha1_$which"};
4253        return ($fh,) if ($blob =~ /^0{40}$/);
4254        my $size = $::_repository->cat_blob($blob, $fh);
4255        croak "Failed to read object $blob" if ($size < 0);
4256        $fh->flush == 0 or croak $!;
4257        seek $fh, 0, 0 or croak $!;
4258
4259        my $exp = ::md5sum($fh);
4260        seek $fh, 0, 0 or croak $!;
4261        return ($fh, $exp);
4262}
4263
4264sub chg_file {
4265        my ($self, $fbat, $m) = @_;
4266        if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
4267                $self->change_file_prop($fbat,'svn:executable','*');
4268        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
4269                $self->change_file_prop($fbat,'svn:executable',undef);
4270        }
4271        my ($fh_a, $exp_a) = _chg_file_get_blob $self, $fbat, $m, 'a';
4272        my ($fh_b, $exp_b) = _chg_file_get_blob $self, $fbat, $m, 'b';
4273        my $pool = SVN::Pool->new;
4274        my $atd = $self->apply_textdelta($fbat, $exp_a, $pool);
4275        if (-s $fh_a) {
4276                my $txstream = SVN::TxDelta::new ($fh_a, $fh_b, $pool);
4277                my $res = SVN::TxDelta::send_txstream($txstream, @$atd, $pool);
4278                if (defined $res) {
4279                        die "Unexpected result from send_txstream: $res\n",
4280                            "(SVN::Core::VERSION: $SVN::Core::VERSION)\n";
4281                }
4282        } else {
4283                my $got = SVN::TxDelta::send_stream($fh_b, @$atd, $pool);
4284                die "Checksum mismatch\nexpected: $exp_b\ngot: $got\n"
4285                    if ($got ne $exp_b);
4286        }
4287        Git::temp_release($fh_b, 1);
4288        Git::temp_release($fh_a, 1);
4289        $pool->clear;
4290}
4291
4292sub D {
4293        my ($self, $m) = @_;
4294        my ($dir, $file) = split_path($m->{file_b});
4295        my $pbat = $self->ensure_path($dir);
4296        print "\tD\t$m->{file_b}\n" unless $::_q;
4297        $self->delete_entry($m->{file_b}, $pbat);
4298}
4299
4300sub close_edit {
4301        my ($self) = @_;
4302        my ($p,$bat) = ($self->{pool}, $self->{bat});
4303        foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
4304                next if $_ eq '';
4305                $self->close_directory($bat->{$_}, $p);
4306        }
4307        $self->close_directory($bat->{''}, $p);
4308        $self->SUPER::close_edit($p);
4309        $p->clear;
4310}
4311
4312sub abort_edit {
4313        my ($self) = @_;
4314        $self->SUPER::abort_edit($self->{pool});
4315}
4316
4317sub DESTROY {
4318        my $self = shift;
4319        $self->SUPER::DESTROY(@_);
4320        $self->{pool}->clear;
4321}
4322
4323# this drives the editor
4324sub apply_diff {
4325        my ($self) = @_;
4326        my $mods = $self->{mods};
4327        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
4328        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
4329                my $f = $m->{chg};
4330                if (defined $o{$f}) {
4331                        $self->$f($m);
4332                } else {
4333                        fatal("Invalid change type: $f");
4334                }
4335        }
4336        $self->rmdirs if $_rmdir;
4337        if (@$mods == 0) {
4338                $self->abort_edit;
4339        } else {
4340                $self->close_edit;
4341        }
4342        return scalar @$mods;
4343}
4344
4345package Git::SVN::Ra;
4346use vars qw/@ISA $config_dir $_log_window_size/;
4347use strict;
4348use warnings;
4349my ($ra_invalid, $can_do_switch, %ignored_err, $RA);
4350
4351BEGIN {
4352        # enforce temporary pool usage for some simple functions
4353        no strict 'refs';
4354        for my $f (qw/rev_proplist get_latest_revnum get_uuid get_repos_root
4355                      get_file/) {
4356                my $SUPER = "SUPER::$f";
4357                *$f = sub {
4358                        my $self = shift;
4359                        my $pool = SVN::Pool->new;
4360                        my @ret = $self->$SUPER(@_,$pool);
4361                        $pool->clear;
4362                        wantarray ? @ret : $ret[0];
4363                };
4364        }
4365}
4366
4367sub _auth_providers () {
4368        [
4369          SVN::Client::get_simple_provider(),
4370          SVN::Client::get_ssl_server_trust_file_provider(),
4371          SVN::Client::get_simple_prompt_provider(
4372            \&Git::SVN::Prompt::simple, 2),
4373          SVN::Client::get_ssl_client_cert_file_provider(),
4374          SVN::Client::get_ssl_client_cert_prompt_provider(
4375            \&Git::SVN::Prompt::ssl_client_cert, 2),
4376          SVN::Client::get_ssl_client_cert_pw_file_provider(),
4377          SVN::Client::get_ssl_client_cert_pw_prompt_provider(
4378            \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
4379          SVN::Client::get_username_provider(),
4380          SVN::Client::get_ssl_server_trust_prompt_provider(
4381            \&Git::SVN::Prompt::ssl_server_trust),
4382          SVN::Client::get_username_prompt_provider(
4383            \&Git::SVN::Prompt::username, 2)
4384        ]
4385}
4386
4387sub escape_uri_only {
4388        my ($uri) = @_;
4389        my @tmp;
4390        foreach (split m{/}, $uri) {
4391                s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
4392                push @tmp, $_;
4393        }
4394        join('/', @tmp);
4395}
4396
4397sub escape_url {
4398        my ($url) = @_;
4399        if ($url =~ m#^(https?)://([^/]+)(.*)$#) {
4400                my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
4401                $url = "$scheme://$domain$uri";
4402        }
4403        $url;
4404}
4405
4406sub new {
4407        my ($class, $url) = @_;
4408        $url =~ s!/+$!!;
4409        return $RA if ($RA && $RA->{url} eq $url);
4410
4411        SVN::_Core::svn_config_ensure($config_dir, undef);
4412        my ($baton, $callbacks) = SVN::Core::auth_open_helper(_auth_providers);
4413        my $config = SVN::Core::config_get_config($config_dir);
4414        $RA = undef;
4415        my $dont_store_passwords = 1;
4416        my $conf_t = ${$config}{'config'};
4417        {
4418                no warnings 'once';
4419                # The usage of $SVN::_Core::SVN_CONFIG_* variables
4420                # produces warnings that variables are used only once.
4421                # I had not found the better way to shut them up, so
4422                # the warnings of type 'once' are disabled in this block.
4423                if (SVN::_Core::svn_config_get_bool($conf_t,
4424                    $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4425                    $SVN::_Core::SVN_CONFIG_OPTION_STORE_PASSWORDS,
4426                    1) == 0) {
4427                        SVN::_Core::svn_auth_set_parameter($baton,
4428                            $SVN::_Core::SVN_AUTH_PARAM_DONT_STORE_PASSWORDS,
4429                            bless (\$dont_store_passwords, "_p_void"));
4430                }
4431                if (SVN::_Core::svn_config_get_bool($conf_t,
4432                    $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4433                    $SVN::_Core::SVN_CONFIG_OPTION_STORE_AUTH_CREDS,
4434                    1) == 0) {
4435                        $Git::SVN::Prompt::_no_auth_cache = 1;
4436                }
4437        } # no warnings 'once'
4438        my $self = SVN::Ra->new(url => escape_url($url), auth => $baton,
4439                              config => $config,
4440                              pool => SVN::Pool->new,
4441                              auth_provider_callbacks => $callbacks);
4442        $self->{url} = $url;
4443        $self->{svn_path} = $url;
4444        $self->{repos_root} = $self->get_repos_root;
4445        $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
4446        $self->{cache} = { check_path => { r => 0, data => {} },
4447                           get_dir => { r => 0, data => {} } };
4448        $RA = bless $self, $class;
4449}
4450
4451sub check_path {
4452        my ($self, $path, $r) = @_;
4453        my $cache = $self->{cache}->{check_path};
4454        if ($r == $cache->{r} && exists $cache->{data}->{$path}) {
4455                return $cache->{data}->{$path};
4456        }
4457        my $pool = SVN::Pool->new;
4458        my $t = $self->SUPER::check_path($path, $r, $pool);
4459        $pool->clear;
4460        if ($r != $cache->{r}) {
4461                %{$cache->{data}} = ();
4462                $cache->{r} = $r;
4463        }
4464        $cache->{data}->{$path} = $t;
4465}
4466
4467sub get_dir {
4468        my ($self, $dir, $r) = @_;
4469        my $cache = $self->{cache}->{get_dir};
4470        if ($r == $cache->{r}) {
4471                if (my $x = $cache->{data}->{$dir}) {
4472                        return wantarray ? @$x : $x->[0];
4473                }
4474        }
4475        my $pool = SVN::Pool->new;
4476        my ($d, undef, $props) = $self->SUPER::get_dir($dir, $r, $pool);
4477        my %dirents = map { $_ => { kind => $d->{$_}->kind } } keys %$d;
4478        $pool->clear;
4479        if ($r != $cache->{r}) {
4480                %{$cache->{data}} = ();
4481                $cache->{r} = $r;
4482        }
4483        $cache->{data}->{$dir} = [ \%dirents, $r, $props ];
4484        wantarray ? (\%dirents, $r, $props) : \%dirents;
4485}
4486
4487sub DESTROY {
4488        # do not call the real DESTROY since we store ourselves in $RA
4489}
4490
4491# get_log(paths, start, end, limit,
4492#         discover_changed_paths, strict_node_history, receiver)
4493sub get_log {
4494        my ($self, @args) = @_;
4495        my $pool = SVN::Pool->new;
4496
4497        # svn_log_changed_path_t objects passed to get_log are likely to be
4498        # overwritten even if only the refs are copied to an external variable,
4499        # so we should dup the structures in their entirety.  Using an
4500        # externally passed pool (instead of our temporary and quickly cleared
4501        # pool in Git::SVN::Ra) does not help matters at all...
4502        my $receiver = pop @args;
4503        my $prefix = "/".$self->{svn_path};
4504        $prefix =~ s#/+($)##;
4505        my $prefix_regex = qr#^\Q$prefix\E#;
4506        push(@args, sub {
4507                my ($paths) = $_[0];
4508                return &$receiver(@_) unless $paths;
4509                $_[0] = ();
4510                foreach my $p (keys %$paths) {
4511                        my $i = $paths->{$p};
4512                        # Make path relative to our url, not repos_root
4513                        $p =~ s/$prefix_regex//;
4514                        my %s = map { $_ => $i->$_; }
4515                                qw/copyfrom_path copyfrom_rev action/;
4516                        if ($s{'copyfrom_path'}) {
4517                                $s{'copyfrom_path'} =~ s/$prefix_regex//;
4518                        }
4519                        $_[0]{$p} = \%s;
4520                }
4521                &$receiver(@_);
4522        });
4523
4524
4525        # the limit parameter was not supported in SVN 1.1.x, so we
4526        # drop it.  Therefore, the receiver callback passed to it
4527        # is made aware of this limitation by being wrapped if
4528        # the limit passed to is being wrapped.
4529        if ($SVN::Core::VERSION le '1.2.0') {
4530                my $limit = splice(@args, 3, 1);
4531                if ($limit > 0) {
4532                        my $receiver = pop @args;
4533                        push(@args, sub { &$receiver(@_) if (--$limit >= 0) });
4534                }
4535        }
4536        my $ret = $self->SUPER::get_log(@args, $pool);
4537        $pool->clear;
4538        $ret;
4539}
4540
4541sub trees_match {
4542        my ($self, $url1, $rev1, $url2, $rev2) = @_;
4543        my $ctx = SVN::Client->new(auth => _auth_providers);
4544        my $out = IO::File->new_tmpfile;
4545
4546        # older SVN (1.1.x) doesn't take $pool as the last parameter for
4547        # $ctx->diff(), so we'll create a default one
4548        my $pool = SVN::Pool->new_default_sub;
4549
4550        $ra_invalid = 1; # this will open a new SVN::Ra connection to $url1
4551        $ctx->diff([], $url1, $rev1, $url2, $rev2, 1, 1, 0, $out, $out);
4552        $out->flush;
4553        my $ret = (($out->stat)[7] == 0);
4554        close $out or croak $!;
4555
4556        $ret;
4557}
4558
4559sub get_commit_editor {
4560        my ($self, $log, $cb, $pool) = @_;
4561        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
4562        $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
4563}
4564
4565sub gs_do_update {
4566        my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
4567        my $new = ($rev_a == $rev_b);
4568        my $path = $gs->{path};
4569
4570        if ($new && -e $gs->{index}) {
4571                unlink $gs->{index} or die
4572                  "Couldn't unlink index: $gs->{index}: $!\n";
4573        }
4574        my $pool = SVN::Pool->new;
4575        $editor->set_path_strip($path);
4576        my (@pc) = split m#/#, $path;
4577        my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
4578                                        1, $editor, $pool);
4579        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
4580
4581        # Since we can't rely on svn_ra_reparent being available, we'll
4582        # just have to do some magic with set_path to make it so
4583        # we only want a partial path.
4584        my $sp = '';
4585        my $final = join('/', @pc);
4586        while (@pc) {
4587                $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
4588                $sp .= '/' if length $sp;
4589                $sp .= shift @pc;
4590        }
4591        die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
4592
4593        $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
4594
4595        $reporter->finish_report($pool);
4596        $pool->clear;
4597        $editor->{git_commit_ok};
4598}
4599
4600# this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
4601# svn_ra_reparent didn't work before 1.4)
4602sub gs_do_switch {
4603        my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
4604        my $path = $gs->{path};
4605        my $pool = SVN::Pool->new;
4606
4607        my $full_url = $self->{url};
4608        my $old_url = $full_url;
4609        $full_url .= '/' . $path if length $path;
4610        my ($ra, $reparented);
4611
4612        if ($old_url =~ m#^svn(\+ssh)?://# ||
4613            ($full_url =~ m#^https?://# &&
4614             escape_url($full_url) ne $full_url)) {
4615                $_[0] = undef;
4616                $self = undef;
4617                $RA = undef;
4618                $ra = Git::SVN::Ra->new($full_url);
4619                $ra_invalid = 1;
4620        } elsif ($old_url ne $full_url) {
4621                SVN::_Ra::svn_ra_reparent($self->{session}, $full_url, $pool);
4622                $self->{url} = $full_url;
4623                $reparented = 1;
4624        }
4625
4626        $ra ||= $self;
4627        $url_b = escape_url($url_b);
4628        my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
4629        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
4630        $reporter->set_path('', $rev_a, 0, @lock, $pool);
4631        $reporter->finish_report($pool);
4632
4633        if ($reparented) {
4634                SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
4635                $self->{url} = $old_url;
4636        }
4637
4638        $pool->clear;
4639        $editor->{git_commit_ok};
4640}
4641
4642sub longest_common_path {
4643        my ($gsv, $globs) = @_;
4644        my %common;
4645        my $common_max = scalar @$gsv;
4646
4647        foreach my $gs (@$gsv) {
4648                my @tmp = split m#/#, $gs->{path};
4649                my $p = '';
4650                foreach (@tmp) {
4651                        $p .= length($p) ? "/$_" : $_;
4652                        $common{$p} ||= 0;
4653                        $common{$p}++;
4654                }
4655        }
4656        $globs ||= [];
4657        $common_max += scalar @$globs;
4658        foreach my $glob (@$globs) {
4659                my @tmp = split m#/#, $glob->{path}->{left};
4660                my $p = '';
4661                foreach (@tmp) {
4662                        $p .= length($p) ? "/$_" : $_;
4663                        $common{$p} ||= 0;
4664                        $common{$p}++;
4665                }
4666        }
4667
4668        my $longest_path = '';
4669        foreach (sort {length $b <=> length $a} keys %common) {
4670                if ($common{$_} == $common_max) {
4671                        $longest_path = $_;
4672                        last;
4673                }
4674        }
4675        $longest_path;
4676}
4677
4678sub gs_fetch_loop_common {
4679        my ($self, $base, $head, $gsv, $globs) = @_;
4680        return if ($base > $head);
4681        my $inc = $_log_window_size;
4682        my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
4683        my $longest_path = longest_common_path($gsv, $globs);
4684        my $ra_url = $self->{url};
4685        my $find_trailing_edge;
4686        while (1) {
4687                my %revs;
4688                my $err;
4689                my $err_handler = $SVN::Error::handler;
4690                $SVN::Error::handler = sub {
4691                        ($err) = @_;
4692                        skip_unknown_revs($err);
4693                };
4694                sub _cb {
4695                        my ($paths, $r, $author, $date, $log) = @_;
4696                        [ $paths,
4697                          { author => $author, date => $date, log => $log } ];
4698                }
4699                $self->get_log([$longest_path], $min, $max, 0, 1, 1,
4700                               sub { $revs{$_[1]} = _cb(@_) });
4701                if ($err) {
4702                        print "Checked through r$max\r";
4703                } else {
4704                        $find_trailing_edge = 1;
4705                }
4706                if ($err and $find_trailing_edge) {
4707                        print STDERR "Path '$longest_path' ",
4708                                     "was probably deleted:\n",
4709                                     $err->expanded_message,
4710                                     "\nWill attempt to follow ",
4711                                     "revisions r$min .. r$max ",
4712                                     "committed before the deletion\n";
4713                        my $hi = $max;
4714                        while (--$hi >= $min) {
4715                                my $ok;
4716                                $self->get_log([$longest_path], $min, $hi,
4717                                               0, 1, 1, sub {
4718                                               $ok = $_[1];
4719                                               $revs{$_[1]} = _cb(@_) });
4720                                if ($ok) {
4721                                        print STDERR "r$min .. r$ok OK\n";
4722                                        last;
4723                                }
4724                        }
4725                        $find_trailing_edge = 0;
4726                }
4727                $SVN::Error::handler = $err_handler;
4728
4729                my %exists = map { $_->{path} => $_ } @$gsv;
4730                foreach my $r (sort {$a <=> $b} keys %revs) {
4731                        my ($paths, $logged) = @{$revs{$r}};
4732
4733                        foreach my $gs ($self->match_globs(\%exists, $paths,
4734                                                           $globs, $r)) {
4735                                if ($gs->rev_map_max >= $r) {
4736                                        next;
4737                                }
4738                                next unless $gs->match_paths($paths, $r);
4739                                $gs->{logged_rev_props} = $logged;
4740                                if (my $last_commit = $gs->last_commit) {
4741                                        $gs->assert_index_clean($last_commit);
4742                                }
4743                                my $log_entry = $gs->do_fetch($paths, $r);
4744                                if ($log_entry) {
4745                                        $gs->do_git_commit($log_entry);
4746                                }
4747                                $INDEX_FILES{$gs->{index}} = 1;
4748                        }
4749                        foreach my $g (@$globs) {
4750                                my $k = "svn-remote.$g->{remote}." .
4751                                        "$g->{t}-maxRev";
4752                                Git::SVN::tmp_config($k, $r);
4753                        }
4754                        if ($ra_invalid) {
4755                                $_[0] = undef;
4756                                $self = undef;
4757                                $RA = undef;
4758                                $self = Git::SVN::Ra->new($ra_url);
4759                                $ra_invalid = undef;
4760                        }
4761                }
4762                # pre-fill the .rev_db since it'll eventually get filled in
4763                # with '0' x40 if something new gets committed
4764                foreach my $gs (@$gsv) {
4765                        next if $gs->rev_map_max >= $max;
4766                        next if defined $gs->rev_map_get($max);
4767                        $gs->rev_map_set($max, 0 x40);
4768                }
4769                foreach my $g (@$globs) {
4770                        my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
4771                        Git::SVN::tmp_config($k, $max);
4772                }
4773                last if $max >= $head;
4774                $min = $max + 1;
4775                $max += $inc;
4776                $max = $head if ($max > $head);
4777        }
4778        Git::SVN::gc();
4779}
4780
4781sub get_dir_globbed {
4782        my ($self, $left, $depth, $r) = @_;
4783
4784        my @x = eval { $self->get_dir($left, $r) };
4785        return unless scalar @x == 3;
4786        my $dirents = $x[0];
4787        my @finalents;
4788        foreach my $de (keys %$dirents) {
4789                next if $dirents->{$de}->{kind} != $SVN::Node::dir;
4790                if ($depth > 1) {
4791                        my @args = ("$left/$de", $depth - 1, $r);
4792                        foreach my $dir ($self->get_dir_globbed(@args)) {
4793                                push @finalents, "$de/$dir";
4794                        }
4795                } else {
4796                        push @finalents, $de;
4797                }
4798        }
4799        @finalents;
4800}
4801
4802sub match_globs {
4803        my ($self, $exists, $paths, $globs, $r) = @_;
4804
4805        sub get_dir_check {
4806                my ($self, $exists, $g, $r) = @_;
4807
4808                my @dirs = $self->get_dir_globbed($g->{path}->{left},
4809                                                  $g->{path}->{depth},
4810                                                  $r);
4811
4812                foreach my $de (@dirs) {
4813                        my $p = $g->{path}->full_path($de);
4814                        next if $exists->{$p};
4815                        next if (length $g->{path}->{right} &&
4816                                 ($self->check_path($p, $r) !=
4817                                  $SVN::Node::dir));
4818                        $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
4819                                         $g->{ref}->full_path($de), 1);
4820                }
4821        }
4822        foreach my $g (@$globs) {
4823                if (my $path = $paths->{"/$g->{path}->{left}"}) {
4824                        if ($path->{action} =~ /^[AR]$/) {
4825                                get_dir_check($self, $exists, $g, $r);
4826                        }
4827                }
4828                foreach (keys %$paths) {
4829                        if (/$g->{path}->{left_regex}/ &&
4830                            !/$g->{path}->{regex}/) {
4831                                next if $paths->{$_}->{action} !~ /^[AR]$/;
4832                                get_dir_check($self, $exists, $g, $r);
4833                        }
4834                        next unless /$g->{path}->{regex}/;
4835                        my $p = $1;
4836                        my $pathname = $g->{path}->full_path($p);
4837                        next if $exists->{$pathname};
4838                        next if ($self->check_path($pathname, $r) !=
4839                                 $SVN::Node::dir);
4840                        $exists->{$pathname} = Git::SVN->init(
4841                                              $self->{url}, $pathname, undef,
4842                                              $g->{ref}->full_path($p), 1);
4843                }
4844                my $c = '';
4845                foreach (split m#/#, $g->{path}->{left}) {
4846                        $c .= "/$_";
4847                        next unless ($paths->{$c} &&
4848                                     ($paths->{$c}->{action} =~ /^[AR]$/));
4849                        get_dir_check($self, $exists, $g, $r);
4850                }
4851        }
4852        values %$exists;
4853}
4854
4855sub minimize_url {
4856        my ($self) = @_;
4857        return $self->{url} if ($self->{url} eq $self->{repos_root});
4858        my $url = $self->{repos_root};
4859        my @components = split(m!/!, $self->{svn_path});
4860        my $c = '';
4861        do {
4862                $url .= "/$c" if length $c;
4863                eval {
4864                        my $ra = (ref $self)->new($url);
4865                        my $latest = $ra->get_latest_revnum;
4866                        $ra->get_log("", $latest, 0, 1, 0, 1, sub {});
4867                };
4868        } while ($@ && ($c = shift @components));
4869        $url;
4870}
4871
4872sub can_do_switch {
4873        my $self = shift;
4874        unless (defined $can_do_switch) {
4875                my $pool = SVN::Pool->new;
4876                my $rep = eval {
4877                        $self->do_switch(1, '', 0, $self->{url},
4878                                         SVN::Delta::Editor->new, $pool);
4879                };
4880                if ($@) {
4881                        $can_do_switch = 0;
4882                } else {
4883                        $rep->abort_report($pool);
4884                        $can_do_switch = 1;
4885                }
4886                $pool->clear;
4887        }
4888        $can_do_switch;
4889}
4890
4891sub skip_unknown_revs {
4892        my ($err) = @_;
4893        my $errno = $err->apr_err();
4894        # Maybe the branch we're tracking didn't
4895        # exist when the repo started, so it's
4896        # not an error if it doesn't, just continue
4897        #
4898        # Wonderfully consistent library, eh?
4899        # 160013 - svn:// and file://
4900        # 175002 - http(s)://
4901        # 175007 - http(s):// (this repo required authorization, too...)
4902        #   More codes may be discovered later...
4903        if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
4904                my $err_key = $err->expanded_message;
4905                # revision numbers change every time, filter them out
4906                $err_key =~ s/\d+/\0/g;
4907                $err_key = "$errno\0$err_key";
4908                unless ($ignored_err{$err_key}) {
4909                        warn "W: Ignoring error from SVN, path probably ",
4910                             "does not exist: ($errno): ",
4911                             $err->expanded_message,"\n";
4912                        warn "W: Do not be alarmed at the above message ",
4913                             "git-svn is just searching aggressively for ",
4914                             "old history.\n",
4915                             "This may take a while on large repositories\n";
4916                        $ignored_err{$err_key} = 1;
4917                }
4918                return;
4919        }
4920        die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
4921}
4922
4923package Git::SVN::Log;
4924use strict;
4925use warnings;
4926use POSIX qw/strftime/;
4927use Time::Local;
4928use constant commit_log_separator => ('-' x 72) . "\n";
4929use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
4930            %rusers $show_commit $incremental/;
4931my $l_fmt;
4932
4933sub cmt_showable {
4934        my ($c) = @_;
4935        return 1 if defined $c->{r};
4936
4937        # big commit message got truncated by the 16k pretty buffer in rev-list
4938        if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
4939                                $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
4940                @{$c->{l}} = ();
4941                my @log = command(qw/cat-file commit/, $c->{c});
4942
4943                # shift off the headers
4944                shift @log while ($log[0] ne '');
4945                shift @log;
4946
4947                # TODO: make $c->{l} not have a trailing newline in the future
4948                @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
4949
4950                (undef, $c->{r}, undef) = ::extract_metadata(
4951                                (grep(/^git-svn-id: /, @log))[-1]);
4952        }
4953        return defined $c->{r};
4954}
4955
4956sub log_use_color {
4957        return $color || Git->repository->get_colorbool('color.diff');
4958}
4959
4960sub git_svn_log_cmd {
4961        my ($r_min, $r_max, @args) = @_;
4962        my $head = 'HEAD';
4963        my (@files, @log_opts);
4964        foreach my $x (@args) {
4965                if ($x eq '--' || @files) {
4966                        push @files, $x;
4967                } else {
4968                        if (::verify_ref("$x^0")) {
4969                                $head = $x;
4970                        } else {
4971                                push @log_opts, $x;
4972                        }
4973                }
4974        }
4975
4976        my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
4977        $gs ||= Git::SVN->_new;
4978        my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
4979                   $gs->refname);
4980        push @cmd, '-r' unless $non_recursive;
4981        push @cmd, qw/--raw --name-status/ if $verbose;
4982        push @cmd, '--color' if log_use_color();
4983        push @cmd, @log_opts;
4984        if (defined $r_max && $r_max == $r_min) {
4985                push @cmd, '--max-count=1';
4986                if (my $c = $gs->rev_map_get($r_max)) {
4987                        push @cmd, $c;
4988                }
4989        } elsif (defined $r_max) {
4990                if ($r_max < $r_min) {
4991                        ($r_min, $r_max) = ($r_max, $r_min);
4992                }
4993                my (undef, $c_max) = $gs->find_rev_before($r_max, 1, $r_min);
4994                my (undef, $c_min) = $gs->find_rev_after($r_min, 1, $r_max);
4995                # If there are no commits in the range, both $c_max and $c_min
4996                # will be undefined.  If there is at least 1 commit in the
4997                # range, both will be defined.
4998                return () if !defined $c_min || !defined $c_max;
4999                if ($c_min eq $c_max) {
5000                        push @cmd, '--max-count=1', $c_min;
5001                } else {
5002                        push @cmd, '--boundary', "$c_min..$c_max";
5003                }
5004        }
5005        return (@cmd, @files);
5006}
5007
5008# adapted from pager.c
5009sub config_pager {
5010        $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
5011        if (!defined $pager) {
5012                $pager = 'less';
5013        } elsif (length $pager == 0 || $pager eq 'cat') {
5014                $pager = undef;
5015        }
5016        $ENV{GIT_PAGER_IN_USE} = defined($pager);
5017}
5018
5019sub run_pager {
5020        return unless -t *STDOUT && defined $pager;
5021        pipe my ($rfd, $wfd) or return;
5022        defined(my $pid = fork) or ::fatal "Can't fork: $!";
5023        if (!$pid) {
5024                open STDOUT, '>&', $wfd or
5025                                     ::fatal "Can't redirect to stdout: $!";
5026                return;
5027        }
5028        open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!";
5029        $ENV{LESS} ||= 'FRSX';
5030        exec $pager or ::fatal "Can't run pager: $! ($pager)";
5031}
5032
5033sub format_svn_date {
5034        # some systmes don't handle or mishandle %z, so be creative.
5035        my $t = shift || time;
5036        my $gm = timelocal(gmtime($t));
5037        my $sign = qw( + + - )[ $t <=> $gm ];
5038        my $gmoff = sprintf("%s%02d%02d", $sign, (gmtime(abs($t - $gm)))[2,1]);
5039        return strftime("%Y-%m-%d %H:%M:%S $gmoff (%a, %d %b %Y)", localtime($t));
5040}
5041
5042sub parse_git_date {
5043        my ($t, $tz) = @_;
5044        # Date::Parse isn't in the standard Perl distro :(
5045        if ($tz =~ s/^\+//) {
5046                $t += tz_to_s_offset($tz);
5047        } elsif ($tz =~ s/^\-//) {
5048                $t -= tz_to_s_offset($tz);
5049        }
5050        return $t;
5051}
5052
5053sub set_local_timezone {
5054        if (defined $TZ) {
5055                $ENV{TZ} = $TZ;
5056        } else {
5057                delete $ENV{TZ};
5058        }
5059}
5060
5061sub tz_to_s_offset {
5062        my ($tz) = @_;
5063        $tz =~ s/(\d\d)$//;
5064        return ($1 * 60) + ($tz * 3600);
5065}
5066
5067sub get_author_info {
5068        my ($dest, $author, $t, $tz) = @_;
5069        $author =~ s/(?:^\s*|\s*$)//g;
5070        $dest->{a_raw} = $author;
5071        my $au;
5072        if ($::_authors) {
5073                $au = $rusers{$author} || undef;
5074        }
5075        if (!$au) {
5076                ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
5077        }
5078        $dest->{t} = $t;
5079        $dest->{tz} = $tz;
5080        $dest->{a} = $au;
5081        $dest->{t_utc} = parse_git_date($t, $tz);
5082}
5083
5084sub process_commit {
5085        my ($c, $r_min, $r_max, $defer) = @_;
5086        if (defined $r_min && defined $r_max) {
5087                if ($r_min == $c->{r} && $r_min == $r_max) {
5088                        show_commit($c);
5089                        return 0;
5090                }
5091                return 1 if $r_min == $r_max;
5092                if ($r_min < $r_max) {
5093                        # we need to reverse the print order
5094                        return 0 if (defined $limit && --$limit < 0);
5095                        push @$defer, $c;
5096                        return 1;
5097                }
5098                if ($r_min != $r_max) {
5099                        return 1 if ($r_min < $c->{r});
5100                        return 1 if ($r_max > $c->{r});
5101                }
5102        }
5103        return 0 if (defined $limit && --$limit < 0);
5104        show_commit($c);
5105        return 1;
5106}
5107
5108sub show_commit {
5109        my $c = shift;
5110        if ($oneline) {
5111                my $x = "\n";
5112                if (my $l = $c->{l}) {
5113                        while ($l->[0] =~ /^\s*$/) { shift @$l }
5114                        $x = $l->[0];
5115                }
5116                $l_fmt ||= 'A' . length($c->{r});
5117                print 'r',pack($l_fmt, $c->{r}),' | ';
5118                print "$c->{c} | " if $show_commit;
5119                print $x;
5120        } else {
5121                show_commit_normal($c);
5122        }
5123}
5124
5125sub show_commit_changed_paths {
5126        my ($c) = @_;
5127        return unless $c->{changed};
5128        print "Changed paths:\n", @{$c->{changed}};
5129}
5130
5131sub show_commit_normal {
5132        my ($c) = @_;
5133        print commit_log_separator, "r$c->{r} | ";
5134        print "$c->{c} | " if $show_commit;
5135        print "$c->{a} | ", format_svn_date($c->{t_utc}), ' | ';
5136        my $nr_line = 0;
5137
5138        if (my $l = $c->{l}) {
5139                while ($l->[$#$l] eq "\n" && $#$l > 0
5140                                          && $l->[($#$l - 1)] eq "\n") {
5141                        pop @$l;
5142                }
5143                $nr_line = scalar @$l;
5144                if (!$nr_line) {
5145                        print "1 line\n\n\n";
5146                } else {
5147                        if ($nr_line == 1) {
5148                                $nr_line = '1 line';
5149                        } else {
5150                                $nr_line .= ' lines';
5151                        }
5152                        print $nr_line, "\n";
5153                        show_commit_changed_paths($c);
5154                        print "\n";
5155                        print $_ foreach @$l;
5156                }
5157        } else {
5158                print "1 line\n";
5159                show_commit_changed_paths($c);
5160                print "\n";
5161
5162        }
5163        foreach my $x (qw/raw stat diff/) {
5164                if ($c->{$x}) {
5165                        print "\n";
5166                        print $_ foreach @{$c->{$x}}
5167                }
5168        }
5169}
5170
5171sub cmd_show_log {
5172        my (@args) = @_;
5173        my ($r_min, $r_max);
5174        my $r_last = -1; # prevent dupes
5175        set_local_timezone();
5176        if (defined $::_revision) {
5177                if ($::_revision =~ /^(\d+):(\d+)$/) {
5178                        ($r_min, $r_max) = ($1, $2);
5179                } elsif ($::_revision =~ /^\d+$/) {
5180                        $r_min = $r_max = $::_revision;
5181                } else {
5182                        ::fatal "-r$::_revision is not supported, use ",
5183                                "standard 'git log' arguments instead";
5184                }
5185        }
5186
5187        config_pager();
5188        @args = git_svn_log_cmd($r_min, $r_max, @args);
5189        if (!@args) {
5190                print commit_log_separator unless $incremental || $oneline;
5191                return;
5192        }
5193        my $log = command_output_pipe(@args);
5194        run_pager();
5195        my (@k, $c, $d, $stat);
5196        my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
5197        while (<$log>) {
5198                if (/^${esc_color}commit -?($::sha1_short)/o) {
5199                        my $cmt = $1;
5200                        if ($c && cmt_showable($c) && $c->{r} != $r_last) {
5201                                $r_last = $c->{r};
5202                                process_commit($c, $r_min, $r_max, \@k) or
5203                                                                goto out;
5204                        }
5205                        $d = undef;
5206                        $c = { c => $cmt };
5207                } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
5208                        get_author_info($c, $1, $2, $3);
5209                } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
5210                        # ignore
5211                } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
5212                        push @{$c->{raw}}, $_;
5213                } elsif (/^${esc_color}[ACRMDT]\t/) {
5214                        # we could add $SVN->{svn_path} here, but that requires
5215                        # remote access at the moment (repo_path_split)...
5216                        s#^(${esc_color})([ACRMDT])\t#$1   $2 #o;
5217                        push @{$c->{changed}}, $_;
5218                } elsif (/^${esc_color}diff /o) {
5219                        $d = 1;
5220                        push @{$c->{diff}}, $_;
5221                } elsif ($d) {
5222                        push @{$c->{diff}}, $_;
5223                } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
5224                          $esc_color*[\+\-]*$esc_color$/x) {
5225                        $stat = 1;
5226                        push @{$c->{stat}}, $_;
5227                } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
5228                        push @{$c->{stat}}, $_;
5229                        $stat = undef;
5230                } elsif (/^${esc_color}    (git-svn-id:.+)$/o) {
5231                        ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
5232                } elsif (s/^${esc_color}    //o) {
5233                        push @{$c->{l}}, $_;
5234                }
5235        }
5236        if ($c && defined $c->{r} && $c->{r} != $r_last) {
5237                $r_last = $c->{r};
5238                process_commit($c, $r_min, $r_max, \@k);
5239        }
5240        if (@k) {
5241                ($r_min, $r_max) = ($r_max, $r_min);
5242                process_commit($_, $r_min, $r_max) foreach reverse @k;
5243        }
5244out:
5245        close $log;
5246        print commit_log_separator unless $incremental || $oneline;
5247}
5248
5249sub cmd_blame {
5250        my $path = pop;
5251
5252        config_pager();
5253        run_pager();
5254
5255        my ($fh, $ctx, $rev);
5256
5257        if ($_git_format) {
5258                ($fh, $ctx) = command_output_pipe('blame', @_, $path);
5259                while (my $line = <$fh>) {
5260                        if ($line =~ /^\^?([[:xdigit:]]+)\s/) {
5261                                # Uncommitted edits show up as a rev ID of
5262                                # all zeros, which we can't look up with
5263                                # cmt_metadata
5264                                if ($1 !~ /^0+$/) {
5265                                        (undef, $rev, undef) =
5266                                                ::cmt_metadata($1);
5267                                        $rev = '0' if (!$rev);
5268                                } else {
5269                                        $rev = '0';
5270                                }
5271                                $rev = sprintf('%-10s', $rev);
5272                                $line =~ s/^\^?[[:xdigit:]]+(\s)/$rev$1/;
5273                        }
5274                        print $line;
5275                }
5276        } else {
5277                ($fh, $ctx) = command_output_pipe('blame', '-p', @_, 'HEAD',
5278                                                  '--', $path);
5279                my ($sha1);
5280                my %authors;
5281                my @buffer;
5282                my %dsha; #distinct sha keys
5283
5284                while (my $line = <$fh>) {
5285                        push @buffer, $line;
5286                        if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5287                                $dsha{$1} = 1;
5288                        }
5289                }
5290
5291                my $s2r = ::cmt_sha2rev_batch([keys %dsha]);
5292
5293                foreach my $line (@buffer) {
5294                        if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5295                                $rev = $s2r->{$1};
5296                                $rev = '0' if (!$rev)
5297                        }
5298                        elsif ($line =~ /^author (.*)/) {
5299                                $authors{$rev} = $1;
5300                                $authors{$rev} =~ s/\s/_/g;
5301                        }
5302                        elsif ($line =~ /^\t(.*)$/) {
5303                                printf("%6s %10s %s\n", $rev, $authors{$rev}, $1);
5304                        }
5305                }
5306        }
5307        command_close_pipe($fh, $ctx);
5308}
5309
5310package Git::SVN::Migration;
5311# these version numbers do NOT correspond to actual version numbers
5312# of git nor git-svn.  They are just relative.
5313#
5314# v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
5315#
5316# v1 layout: .git/$id/info/url, refs/remotes/$id
5317#
5318# v2 layout: .git/svn/$id/info/url, refs/remotes/$id
5319#
5320# v3 layout: .git/svn/$id, refs/remotes/$id
5321#            - info/url may remain for backwards compatibility
5322#            - this is what we migrate up to this layout automatically,
5323#            - this will be used by git svn init on single branches
5324# v3.1 layout (auto migrated):
5325#            - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
5326#              for backwards compatibility
5327#
5328# v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
5329#            - this is only created for newly multi-init-ed
5330#              repositories.  Similar in spirit to the
5331#              --use-separate-remotes option in git-clone (now default)
5332#            - we do not automatically migrate to this (following
5333#              the example set by core git)
5334#
5335# v5 layout: .rev_db.$UUID => .rev_map.$UUID
5336#            - newer, more-efficient format that uses 24-bytes per record
5337#              with no filler space.
5338#            - use xxd -c24 < .rev_map.$UUID to view and debug
5339#            - This is a one-way migration, repositories updated to the
5340#              new format will not be able to use old git-svn without
5341#              rebuilding the .rev_db.  Rebuilding the rev_db is not
5342#              possible if noMetadata or useSvmProps are set; but should
5343#              be no problem for users that use the (sensible) defaults.
5344use strict;
5345use warnings;
5346use Carp qw/croak/;
5347use File::Path qw/mkpath/;
5348use File::Basename qw/dirname basename/;
5349use vars qw/$_minimize/;
5350
5351sub migrate_from_v0 {
5352        my $git_dir = $ENV{GIT_DIR};
5353        return undef unless -d $git_dir;
5354        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5355        my $migrated = 0;
5356        while (<$fh>) {
5357                chomp;
5358                my ($id, $orig_ref) = ($_, $_);
5359                next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
5360                next unless -f "$git_dir/$id/info/url";
5361                my $new_ref = "refs/remotes/$id";
5362                if (::verify_ref("$new_ref^0")) {
5363                        print STDERR "W: $orig_ref is probably an old ",
5364                                     "branch used by an ancient version of ",
5365                                     "git-svn.\n",
5366                                     "However, $new_ref also exists.\n",
5367                                     "We will not be able ",
5368                                     "to use this branch until this ",
5369                                     "ambiguity is resolved.\n";
5370                        next;
5371                }
5372                print STDERR "Migrating from v0 layout...\n" if !$migrated;
5373                print STDERR "Renaming ref: $orig_ref => $new_ref\n";
5374                command_noisy('update-ref', $new_ref, $orig_ref);
5375                command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
5376                $migrated++;
5377        }
5378        command_close_pipe($fh, $ctx);
5379        print STDERR "Done migrating from v0 layout...\n" if $migrated;
5380        $migrated;
5381}
5382
5383sub migrate_from_v1 {
5384        my $git_dir = $ENV{GIT_DIR};
5385        my $migrated = 0;
5386        return $migrated unless -d $git_dir;
5387        my $svn_dir = "$git_dir/svn";
5388
5389        # just in case somebody used 'svn' as their $id at some point...
5390        return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
5391
5392        print STDERR "Migrating from a git-svn v1 layout...\n";
5393        mkpath([$svn_dir]);
5394        print STDERR "Data from a previous version of git-svn exists, but\n\t",
5395                     "$svn_dir\n\t(required for this version ",
5396                     "($::VERSION) of git-svn) does not exist.\n";
5397        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5398        while (<$fh>) {
5399                my $x = $_;
5400                next unless $x =~ s#^refs/remotes/##;
5401                chomp $x;
5402                next unless -f "$git_dir/$x/info/url";
5403                my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
5404                next unless $u;
5405                my $dn = dirname("$git_dir/svn/$x");
5406                mkpath([$dn]) unless -d $dn;
5407                if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
5408                        mkpath(["$git_dir/svn/svn"]);
5409                        print STDERR " - $git_dir/$x/info => ",
5410                                        "$git_dir/svn/$x/info\n";
5411                        rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
5412                               croak "$!: $x";
5413                        # don't worry too much about these, they probably
5414                        # don't exist with repos this old (save for index,
5415                        # and we can easily regenerate that)
5416                        foreach my $f (qw/unhandled.log index .rev_db/) {
5417                                rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
5418                        }
5419                } else {
5420                        print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
5421                        rename "$git_dir/$x", "$git_dir/svn/$x" or
5422                               croak "$!: $x";
5423                }
5424                $migrated++;
5425        }
5426        command_close_pipe($fh, $ctx);
5427        print STDERR "Done migrating from a git-svn v1 layout\n";
5428        $migrated;
5429}
5430
5431sub read_old_urls {
5432        my ($l_map, $pfx, $path) = @_;
5433        my @dir;
5434        foreach (<$path/*>) {
5435                if (-r "$_/info/url") {
5436                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
5437                        my $ref_id = $pfx . basename $_;
5438                        my $url = ::file_to_s("$_/info/url");
5439                        $l_map->{$ref_id} = $url;
5440                } elsif (-d $_) {
5441                        push @dir, $_;
5442                }
5443        }
5444        foreach (@dir) {
5445                my $x = $_;
5446                $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
5447                read_old_urls($l_map, $x, $_);
5448        }
5449}
5450
5451sub migrate_from_v2 {
5452        my @cfg = command(qw/config -l/);
5453        return if grep /^svn-remote\..+\.url=/, @cfg;
5454        my %l_map;
5455        read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
5456        my $migrated = 0;
5457
5458        foreach my $ref_id (sort keys %l_map) {
5459                eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
5460                if ($@) {
5461                        Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
5462                }
5463                $migrated++;
5464        }
5465        $migrated;
5466}
5467
5468sub minimize_connections {
5469        my $r = Git::SVN::read_all_remotes();
5470        my $new_urls = {};
5471        my $root_repos = {};
5472        foreach my $repo_id (keys %$r) {
5473                my $url = $r->{$repo_id}->{url} or next;
5474                my $fetch = $r->{$repo_id}->{fetch} or next;
5475                my $ra = Git::SVN::Ra->new($url);
5476
5477                # skip existing cases where we already connect to the root
5478                if (($ra->{url} eq $ra->{repos_root}) ||
5479                    ($ra->{repos_root} eq $repo_id)) {
5480                        $root_repos->{$ra->{url}} = $repo_id;
5481                        next;
5482                }
5483
5484                my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
5485                my $root_path = $ra->{url};
5486                $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
5487                foreach my $path (keys %$fetch) {
5488                        my $ref_id = $fetch->{$path};
5489                        my $gs = Git::SVN->new($ref_id, $repo_id, $path);
5490
5491                        # make sure we can read when connecting to
5492                        # a higher level of a repository
5493                        my ($last_rev, undef) = $gs->last_rev_commit;
5494                        if (!defined $last_rev) {
5495                                $last_rev = eval {
5496                                        $root_ra->get_latest_revnum;
5497                                };
5498                                next if $@;
5499                        }
5500                        my $new = $root_path;
5501                        $new .= length $path ? "/$path" : '';
5502                        eval {
5503                                $root_ra->get_log([$new], $last_rev, $last_rev,
5504                                                  0, 0, 1, sub { });
5505                        };
5506                        next if $@;
5507                        $new_urls->{$ra->{repos_root}}->{$new} =
5508                                { ref_id => $ref_id,
5509                                  old_repo_id => $repo_id,
5510                                  old_path => $path };
5511                }
5512        }
5513
5514        my @emptied;
5515        foreach my $url (keys %$new_urls) {
5516                # see if we can re-use an existing [svn-remote "repo_id"]
5517                # instead of creating a(n ugly) new section:
5518                my $repo_id = $root_repos->{$url} || $url;
5519
5520                my $fetch = $new_urls->{$url};
5521                foreach my $path (keys %$fetch) {
5522                        my $x = $fetch->{$path};
5523                        Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
5524                        my $pfx = "svn-remote.$x->{old_repo_id}";
5525
5526                        my $old_fetch = quotemeta("$x->{old_path}:".
5527                                                  "$x->{ref_id}");
5528                        command_noisy(qw/config --unset/,
5529                                      "$pfx.fetch", '^'. $old_fetch . '$');
5530                        delete $r->{$x->{old_repo_id}}->
5531                               {fetch}->{$x->{old_path}};
5532                        if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
5533                                command_noisy(qw/config --unset/,
5534                                              "$pfx.url");
5535                                push @emptied, $x->{old_repo_id}
5536                        }
5537                }
5538        }
5539        if (@emptied) {
5540                my $file = $ENV{GIT_CONFIG} || "$ENV{GIT_DIR}/config";
5541                print STDERR <<EOF;
5542The following [svn-remote] sections in your config file ($file) are empty
5543and can be safely removed:
5544EOF
5545                print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
5546        }
5547}
5548
5549sub migration_check {
5550        migrate_from_v0();
5551        migrate_from_v1();
5552        migrate_from_v2();
5553        minimize_connections() if $_minimize;
5554}
5555
5556package Git::IndexInfo;
5557use strict;
5558use warnings;
5559use Git qw/command_input_pipe command_close_pipe/;
5560
5561sub new {
5562        my ($class) = @_;
5563        my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
5564        bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
5565}
5566
5567sub remove {
5568        my ($self, $path) = @_;
5569        if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
5570                return ++$self->{nr};
5571        }
5572        undef;
5573}
5574
5575sub update {
5576        my ($self, $mode, $hash, $path) = @_;
5577        if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
5578                return ++$self->{nr};
5579        }
5580        undef;
5581}
5582
5583sub DESTROY {
5584        my ($self) = @_;
5585        command_close_pipe($self->{gui}, $self->{ctx});
5586}
5587
5588package Git::SVN::GlobSpec;
5589use strict;
5590use warnings;
5591
5592sub new {
5593        my ($class, $glob) = @_;
5594        my $re = $glob;
5595        $re =~ s!/+$!!g; # no need for trailing slashes
5596        $re =~ m!^([^*]*)(\*(?:/\*)*)(.*)$!;
5597        my $temp = $re;
5598        my ($left, $right) = ($1, $3);
5599        $re = $2;
5600        my $depth = $re =~ tr/*/*/;
5601        if ($depth != $temp =~ tr/*/*/) {
5602                die "Only one set of wildcard directories " .
5603                        "(e.g. '*' or '*/*/*') is supported: '$glob'\n";
5604        }
5605        if ($depth == 0) {
5606                die "One '*' is needed for glob: '$glob'\n";
5607        }
5608        $re =~ s!\*!\[^/\]*!g;
5609        $re = quotemeta($left) . "($re)" . quotemeta($right);
5610        if (length $left && !($left =~ s!/+$!!g)) {
5611                die "Missing trailing '/' on left side of: '$glob' ($left)\n";
5612        }
5613        if (length $right && !($right =~ s!^/+!!g)) {
5614                die "Missing leading '/' on right side of: '$glob' ($right)\n";
5615        }
5616        my $left_re = qr/^\/\Q$left\E(\/|$)/;
5617        bless { left => $left, right => $right, left_regex => $left_re,
5618                regex => qr/$re/, glob => $glob, depth => $depth }, $class;
5619}
5620
5621sub full_path {
5622        my ($self, $path) = @_;
5623        return (length $self->{left} ? "$self->{left}/" : '') .
5624               $path . (length $self->{right} ? "/$self->{right}" : '');
5625}
5626
5627__END__
5628
5629Data structures:
5630
5631
5632$remotes = { # returned by read_all_remotes()
5633        'svn' => {
5634                # svn-remote.svn.url=https://svn.musicpd.org
5635                url => 'https://svn.musicpd.org',
5636                # svn-remote.svn.fetch=mpd/trunk:trunk
5637                fetch => {
5638                        'mpd/trunk' => 'trunk',
5639                },
5640                # svn-remote.svn.tags=mpd/tags/*:tags/*
5641                tags => {
5642                        path => {
5643                                left => 'mpd/tags',
5644                                right => '',
5645                                regex => qr!mpd/tags/([^/]+)$!,
5646                                glob => 'tags/*',
5647                        },
5648                        ref => {
5649                                left => 'tags',
5650                                right => '',
5651                                regex => qr!tags/([^/]+)$!,
5652                                glob => 'tags/*',
5653                        },
5654                }
5655        }
5656};
5657
5658$log_entry hashref as returned by libsvn_log_entry()
5659{
5660        log => 'whitespace-formatted log entry
5661',                                              # trailing newline is preserved
5662        revision => '8',                        # integer
5663        date => '2004-02-24T17:01:44.108345Z',  # commit date
5664        author => 'committer name'
5665};
5666
5667
5668# this is generated by generate_diff();
5669@mods = array of diff-index line hashes, each element represents one line
5670        of diff-index output
5671
5672diff-index line ($m hash)
5673{
5674        mode_a => first column of diff-index output, no leading ':',
5675        mode_b => second column of diff-index output,
5676        sha1_b => sha1sum of the final blob,
5677        chg => change type [MCRADT],
5678        file_a => original file name of a file (iff chg is 'C' or 'R')
5679        file_b => new/current file name of a file (any chg)
5680}
5681;
5682
5683# retval of read_url_paths{,_all}();
5684$l_map = {
5685        # repository root url
5686        'https://svn.musicpd.org' => {
5687                # repository path               # GIT_SVN_ID
5688                'mpd/trunk'             =>      'trunk',
5689                'mpd/tags/0.11.5'       =>      'tags/0.11.5',
5690        },
5691}
5692
5693Notes:
5694        I don't trust the each() function on unless I created %hash myself
5695        because the internal iterator may not have started at base.