git-svn.perlon commit Avoid writing to buffer in add_excludes_from_file_1() (b5041c5)
   1#!/usr/bin/env perl
   2# Copyright (C) 2006, Eric Wong <normalperson@yhbt.net>
   3# License: GPL v2 or later
   4use warnings;
   5use strict;
   6use vars qw/    $AUTHOR $VERSION
   7                $sha1 $sha1_short $_revision $_repository
   8                $_q $_authors $_authors_prog %users/;
   9$AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
  10$VERSION = '@@GIT_VERSION@@';
  11
  12# From which subdir have we been invoked?
  13my $cmd_dir_prefix = eval {
  14        command_oneline([qw/rev-parse --show-prefix/], STDERR => 0)
  15} || '';
  16
  17my $git_dir_user_set = 1 if defined $ENV{GIT_DIR};
  18$ENV{GIT_DIR} ||= '.git';
  19$Git::SVN::default_repo_id = 'svn';
  20$Git::SVN::default_ref_id = $ENV{GIT_SVN_ID} || 'git-svn';
  21$Git::SVN::Ra::_log_window_size = 100;
  22$Git::SVN::_minimize_url = 'unset';
  23
  24$Git::SVN::Log::TZ = $ENV{TZ};
  25$ENV{TZ} = 'UTC';
  26$| = 1; # unbuffer STDOUT
  27
  28sub fatal (@) { print STDERR "@_\n"; exit 1 }
  29require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
  30require SVN::Ra;
  31require SVN::Delta;
  32if ($SVN::Core::VERSION lt '1.1.0') {
  33        fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)";
  34}
  35my $can_compress = eval { require Compress::Zlib; 1};
  36push @Git::SVN::Ra::ISA, 'SVN::Ra';
  37push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
  38push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
  39use Carp qw/croak/;
  40use Digest::MD5;
  41use IO::File qw//;
  42use File::Basename qw/dirname basename/;
  43use File::Path qw/mkpath/;
  44use File::Spec;
  45use File::Find;
  46use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev/;
  47use IPC::Open3;
  48use Git;
  49
  50BEGIN {
  51        # import functions from Git into our packages, en masse
  52        no strict 'refs';
  53        foreach (qw/command command_oneline command_noisy command_output_pipe
  54                    command_input_pipe command_close_pipe
  55                    command_bidi_pipe command_close_bidi_pipe/) {
  56                for my $package ( qw(SVN::Git::Editor SVN::Git::Fetcher
  57                        Git::SVN::Migration Git::SVN::Log Git::SVN),
  58                        __PACKAGE__) {
  59                        *{"${package}::$_"} = \&{"Git::$_"};
  60                }
  61        }
  62}
  63
  64my ($SVN);
  65
  66$sha1 = qr/[a-f\d]{40}/;
  67$sha1_short = qr/[a-f\d]{4,40}/;
  68my ($_stdin, $_help, $_edit,
  69        $_message, $_file, $_branch_dest,
  70        $_template, $_shared,
  71        $_version, $_fetch_all, $_no_rebase, $_fetch_parent,
  72        $_merge, $_strategy, $_dry_run, $_local,
  73        $_prefix, $_no_checkout, $_url, $_verbose,
  74        $_git_format, $_commit_url, $_tag);
  75$Git::SVN::_follow_parent = 1;
  76$_q ||= 0;
  77my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
  78                    'config-dir=s' => \$Git::SVN::Ra::config_dir,
  79                    'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache,
  80                    'ignore-paths=s' => \$SVN::Git::Fetcher::_ignore_regex );
  81my %fc_opts = ( 'follow-parent|follow!' => \$Git::SVN::_follow_parent,
  82                'authors-file|A=s' => \$_authors,
  83                'authors-prog=s' => \$_authors_prog,
  84                'repack:i' => \$Git::SVN::_repack,
  85                'noMetadata' => \$Git::SVN::_no_metadata,
  86                'useSvmProps' => \$Git::SVN::_use_svm_props,
  87                'useSvnsyncProps' => \$Git::SVN::_use_svnsync_props,
  88                'log-window-size=i' => \$Git::SVN::Ra::_log_window_size,
  89                'no-checkout' => \$_no_checkout,
  90                'quiet|q+' => \$_q,
  91                'repack-flags|repack-args|repack-opts=s' =>
  92                   \$Git::SVN::_repack_flags,
  93                'use-log-author' => \$Git::SVN::_use_log_author,
  94                'add-author-from' => \$Git::SVN::_add_author_from,
  95                'localtime' => \$Git::SVN::_localtime,
  96                %remote_opts );
  97
  98my ($_trunk, @_tags, @_branches, $_stdlayout);
  99my %icv;
 100my %init_opts = ( 'template=s' => \$_template, 'shared:s' => \$_shared,
 101                  'trunk|T=s' => \$_trunk, 'tags|t=s@' => \@_tags,
 102                  'branches|b=s@' => \@_branches, 'prefix=s' => \$_prefix,
 103                  'stdlayout|s' => \$_stdlayout,
 104                  'minimize-url|m!' => \$Git::SVN::_minimize_url,
 105                  'no-metadata' => sub { $icv{noMetadata} = 1 },
 106                  'use-svm-props' => sub { $icv{useSvmProps} = 1 },
 107                  'use-svnsync-props' => sub { $icv{useSvnsyncProps} = 1 },
 108                  'rewrite-root=s' => sub { $icv{rewriteRoot} = $_[1] },
 109                  %remote_opts );
 110my %cmt_opts = ( 'edit|e' => \$_edit,
 111                'rmdir' => \$SVN::Git::Editor::_rmdir,
 112                'find-copies-harder' => \$SVN::Git::Editor::_find_copies_harder,
 113                'l=i' => \$SVN::Git::Editor::_rename_limit,
 114                'copy-similarity|C=i'=> \$SVN::Git::Editor::_cp_similarity
 115);
 116
 117my %cmd = (
 118        fetch => [ \&cmd_fetch, "Download new revisions from SVN",
 119                        { 'revision|r=s' => \$_revision,
 120                          'fetch-all|all' => \$_fetch_all,
 121                          'parent|p' => \$_fetch_parent,
 122                           %fc_opts } ],
 123        clone => [ \&cmd_clone, "Initialize and fetch revisions",
 124                        { 'revision|r=s' => \$_revision,
 125                           %fc_opts, %init_opts } ],
 126        init => [ \&cmd_init, "Initialize a repo for tracking" .
 127                          " (requires URL argument)",
 128                          \%init_opts ],
 129        'multi-init' => [ \&cmd_multi_init,
 130                          "Deprecated alias for ".
 131                          "'$0 init -T<trunk> -b<branches> -t<tags>'",
 132                          \%init_opts ],
 133        dcommit => [ \&cmd_dcommit,
 134                     'Commit several diffs to merge with upstream',
 135                        { 'merge|m|M' => \$_merge,
 136                          'strategy|s=s' => \$_strategy,
 137                          'verbose|v' => \$_verbose,
 138                          'dry-run|n' => \$_dry_run,
 139                          'fetch-all|all' => \$_fetch_all,
 140                          'commit-url=s' => \$_commit_url,
 141                          'revision|r=i' => \$_revision,
 142                          'no-rebase' => \$_no_rebase,
 143                        %cmt_opts, %fc_opts } ],
 144        branch => [ \&cmd_branch,
 145                    'Create a branch in the SVN repository',
 146                    { 'message|m=s' => \$_message,
 147                      'destination|d=s' => \$_branch_dest,
 148                      'dry-run|n' => \$_dry_run,
 149                      'tag|t' => \$_tag } ],
 150        tag => [ sub { $_tag = 1; cmd_branch(@_) },
 151                 'Create a tag in the SVN repository',
 152                 { 'message|m=s' => \$_message,
 153                   'destination|d=s' => \$_branch_dest,
 154                   'dry-run|n' => \$_dry_run } ],
 155        'set-tree' => [ \&cmd_set_tree,
 156                        "Set an SVN repository to a git tree-ish",
 157                        { 'stdin' => \$_stdin, %cmt_opts, %fc_opts, } ],
 158        'create-ignore' => [ \&cmd_create_ignore,
 159                             'Create a .gitignore per svn:ignore',
 160                             { 'revision|r=i' => \$_revision
 161                             } ],
 162        'propget' => [ \&cmd_propget,
 163                       'Print the value of a property on a file or directory',
 164                       { 'revision|r=i' => \$_revision } ],
 165        'proplist' => [ \&cmd_proplist,
 166                       'List all properties of a file or directory',
 167                       { 'revision|r=i' => \$_revision } ],
 168        'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
 169                        { 'revision|r=i' => \$_revision
 170                        } ],
 171        'show-externals' => [ \&cmd_show_externals, "Show svn:externals listings",
 172                        { 'revision|r=i' => \$_revision
 173                        } ],
 174        'multi-fetch' => [ \&cmd_multi_fetch,
 175                           "Deprecated alias for $0 fetch --all",
 176                           { 'revision|r=s' => \$_revision, %fc_opts } ],
 177        'migrate' => [ sub { },
 178                       # no-op, we automatically run this anyways,
 179                       'Migrate configuration/metadata/layout from
 180                        previous versions of git-svn',
 181                       { 'minimize' => \$Git::SVN::Migration::_minimize,
 182                         %remote_opts } ],
 183        'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
 184                        { 'limit=i' => \$Git::SVN::Log::limit,
 185                          'revision|r=s' => \$_revision,
 186                          'verbose|v' => \$Git::SVN::Log::verbose,
 187                          'incremental' => \$Git::SVN::Log::incremental,
 188                          'oneline' => \$Git::SVN::Log::oneline,
 189                          'show-commit' => \$Git::SVN::Log::show_commit,
 190                          'non-recursive' => \$Git::SVN::Log::non_recursive,
 191                          'authors-file|A=s' => \$_authors,
 192                          'color' => \$Git::SVN::Log::color,
 193                          'pager=s' => \$Git::SVN::Log::pager
 194                        } ],
 195        'find-rev' => [ \&cmd_find_rev,
 196                        "Translate between SVN revision numbers and tree-ish",
 197                        {} ],
 198        'rebase' => [ \&cmd_rebase, "Fetch and rebase your working directory",
 199                        { 'merge|m|M' => \$_merge,
 200                          'verbose|v' => \$_verbose,
 201                          'strategy|s=s' => \$_strategy,
 202                          'local|l' => \$_local,
 203                          'fetch-all|all' => \$_fetch_all,
 204                          'dry-run|n' => \$_dry_run,
 205                          %fc_opts } ],
 206        'commit-diff' => [ \&cmd_commit_diff,
 207                           'Commit a diff between two trees',
 208                        { 'message|m=s' => \$_message,
 209                          'file|F=s' => \$_file,
 210                          'revision|r=s' => \$_revision,
 211                        %cmt_opts } ],
 212        'info' => [ \&cmd_info,
 213                    "Show info about the latest SVN revision
 214                     on the current branch",
 215                    { 'url' => \$_url, } ],
 216        'blame' => [ \&Git::SVN::Log::cmd_blame,
 217                    "Show what revision and author last modified each line of a file",
 218                    { 'git-format' => \$_git_format } ],
 219        'reset' => [ \&cmd_reset,
 220                     "Undo fetches back to the specified SVN revision",
 221                     { 'revision|r=s' => \$_revision,
 222                       'parent|p' => \$_fetch_parent } ],
 223        'gc' => [ \&cmd_gc,
 224                  "Compress unhandled.log files in .git/svn and remove " .
 225                  "index files in .git/svn",
 226                {} ],
 227);
 228
 229my $cmd;
 230for (my $i = 0; $i < @ARGV; $i++) {
 231        if (defined $cmd{$ARGV[$i]}) {
 232                $cmd = $ARGV[$i];
 233                splice @ARGV, $i, 1;
 234                last;
 235        } elsif ($ARGV[$i] eq 'help') {
 236                $cmd = $ARGV[$i+1];
 237                usage(0);
 238        }
 239};
 240
 241# make sure we're always running at the top-level working directory
 242unless ($cmd && $cmd =~ /(?:clone|init|multi-init)$/) {
 243        unless (-d $ENV{GIT_DIR}) {
 244                if ($git_dir_user_set) {
 245                        die "GIT_DIR=$ENV{GIT_DIR} explicitly set, ",
 246                            "but it is not a directory\n";
 247                }
 248                my $git_dir = delete $ENV{GIT_DIR};
 249                my $cdup = undef;
 250                git_cmd_try {
 251                        $cdup = command_oneline(qw/rev-parse --show-cdup/);
 252                        $git_dir = '.' unless ($cdup);
 253                        chomp $cdup if ($cdup);
 254                        $cdup = "." unless ($cdup && length $cdup);
 255                } "Already at toplevel, but $git_dir not found\n";
 256                chdir $cdup or die "Unable to chdir up to '$cdup'\n";
 257                unless (-d $git_dir) {
 258                        die "$git_dir still not found after going to ",
 259                            "'$cdup'\n";
 260                }
 261                $ENV{GIT_DIR} = $git_dir;
 262        }
 263        $_repository = Git->repository(Repository => $ENV{GIT_DIR});
 264}
 265
 266my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
 267
 268read_repo_config(\%opts);
 269if ($cmd && ($cmd eq 'log' || $cmd eq 'blame')) {
 270        Getopt::Long::Configure('pass_through');
 271}
 272my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
 273                    'minimize-connections' => \$Git::SVN::Migration::_minimize,
 274                    'id|i=s' => \$Git::SVN::default_ref_id,
 275                    'svn-remote|remote|R=s' => sub {
 276                       $Git::SVN::no_reuse_existing = 1;
 277                       $Git::SVN::default_repo_id = $_[1] });
 278exit 1 if (!$rv && $cmd && $cmd ne 'log');
 279
 280usage(0) if $_help;
 281version() if $_version;
 282usage(1) unless defined $cmd;
 283load_authors() if $_authors;
 284if (defined $_authors_prog) {
 285        $_authors_prog = "'" . File::Spec->rel2abs($_authors_prog) . "'";
 286}
 287
 288unless ($cmd =~ /^(?:clone|init|multi-init|commit-diff)$/) {
 289        Git::SVN::Migration::migration_check();
 290}
 291Git::SVN::init_vars();
 292eval {
 293        Git::SVN::verify_remotes_sanity();
 294        $cmd{$cmd}->[0]->(@ARGV);
 295};
 296fatal $@ if $@;
 297post_fetch_checkout();
 298exit 0;
 299
 300####################### primary functions ######################
 301sub usage {
 302        my $exit = shift || 0;
 303        my $fd = $exit ? \*STDERR : \*STDOUT;
 304        print $fd <<"";
 305git-svn - bidirectional operations between a single Subversion tree and git
 306Usage: git svn <command> [options] [arguments]\n
 307
 308        print $fd "Available commands:\n" unless $cmd;
 309
 310        foreach (sort keys %cmd) {
 311                next if $cmd && $cmd ne $_;
 312                next if /^multi-/; # don't show deprecated commands
 313                print $fd '  ',pack('A17',$_),$cmd{$_}->[1],"\n";
 314                foreach (sort keys %{$cmd{$_}->[2]}) {
 315                        # mixed-case options are for .git/config only
 316                        next if /[A-Z]/ && /^[a-z]+$/i;
 317                        # prints out arguments as they should be passed:
 318                        my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
 319                        print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
 320                                                        "--$_" : "-$_" }
 321                                                split /\|/,$_)," $x\n";
 322                }
 323        }
 324        print $fd <<"";
 325\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
 326arbitrary identifier if you're tracking multiple SVN branches/repositories in
 327one git repository and want to keep them separate.  See git-svn(1) for more
 328information.
 329
 330        exit $exit;
 331}
 332
 333sub version {
 334        print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
 335        exit 0;
 336}
 337
 338sub do_git_init_db {
 339        unless (-d $ENV{GIT_DIR}) {
 340                my @init_db = ('init');
 341                push @init_db, "--template=$_template" if defined $_template;
 342                if (defined $_shared) {
 343                        if ($_shared =~ /[a-z]/) {
 344                                push @init_db, "--shared=$_shared";
 345                        } else {
 346                                push @init_db, "--shared";
 347                        }
 348                }
 349                command_noisy(@init_db);
 350                $_repository = Git->repository(Repository => ".git");
 351        }
 352        command_noisy('config', 'core.autocrlf', 'false');
 353        my $set;
 354        my $pfx = "svn-remote.$Git::SVN::default_repo_id";
 355        foreach my $i (keys %icv) {
 356                die "'$set' and '$i' cannot both be set\n" if $set;
 357                next unless defined $icv{$i};
 358                command_noisy('config', "$pfx.$i", $icv{$i});
 359                $set = $i;
 360        }
 361        my $ignore_regex = \$SVN::Git::Fetcher::_ignore_regex;
 362        command_noisy('config', "$pfx.ignore-paths", $$ignore_regex)
 363                if defined $$ignore_regex;
 364}
 365
 366sub init_subdir {
 367        my $repo_path = shift or return;
 368        mkpath([$repo_path]) unless -d $repo_path;
 369        chdir $repo_path or die "Couldn't chdir to $repo_path: $!\n";
 370        $ENV{GIT_DIR} = '.git';
 371        $_repository = Git->repository(Repository => $ENV{GIT_DIR});
 372}
 373
 374sub cmd_clone {
 375        my ($url, $path) = @_;
 376        if (!defined $path &&
 377            (defined $_trunk || @_branches || @_tags ||
 378             defined $_stdlayout) &&
 379            $url !~ m#^[a-z\+]+://#) {
 380                $path = $url;
 381        }
 382        $path = basename($url) if !defined $path || !length $path;
 383        cmd_init($url, $path);
 384        Git::SVN::fetch_all($Git::SVN::default_repo_id);
 385        command_oneline('config', 'svn.authorsfile', $_authors) if $_authors;
 386}
 387
 388sub cmd_init {
 389        if (defined $_stdlayout) {
 390                $_trunk = 'trunk' if (!defined $_trunk);
 391                @_tags = 'tags' if (! @_tags);
 392                @_branches = 'branches' if (! @_branches);
 393        }
 394        if (defined $_trunk || @_branches || @_tags) {
 395                return cmd_multi_init(@_);
 396        }
 397        my $url = shift or die "SVN repository location required ",
 398                               "as a command-line argument\n";
 399        $url = canonicalize_url($url);
 400        init_subdir(@_);
 401        do_git_init_db();
 402
 403        if ($Git::SVN::_minimize_url eq 'unset') {
 404                $Git::SVN::_minimize_url = 0;
 405        }
 406
 407        Git::SVN->init($url);
 408}
 409
 410sub cmd_fetch {
 411        if (grep /^\d+=./, @_) {
 412                die "'<rev>=<commit>' fetch arguments are ",
 413                    "no longer supported.\n";
 414        }
 415        my ($remote) = @_;
 416        if (@_ > 1) {
 417                die "Usage: $0 fetch [--all] [--parent] [svn-remote]\n";
 418        }
 419        if ($_fetch_parent) {
 420                my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 421                unless ($gs) {
 422                        die "Unable to determine upstream SVN information from ",
 423                            "working tree history\n";
 424                }
 425                # just fetch, don't checkout.
 426                $_no_checkout = 'true';
 427                $_fetch_all ? $gs->fetch_all : $gs->fetch;
 428        } elsif ($_fetch_all) {
 429                cmd_multi_fetch();
 430        } else {
 431                $remote ||= $Git::SVN::default_repo_id;
 432                Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
 433        }
 434}
 435
 436sub cmd_set_tree {
 437        my (@commits) = @_;
 438        if ($_stdin || !@commits) {
 439                print "Reading from stdin...\n";
 440                @commits = ();
 441                while (<STDIN>) {
 442                        if (/\b($sha1_short)\b/o) {
 443                                unshift @commits, $1;
 444                        }
 445                }
 446        }
 447        my @revs;
 448        foreach my $c (@commits) {
 449                my @tmp = command('rev-parse',$c);
 450                if (scalar @tmp == 1) {
 451                        push @revs, $tmp[0];
 452                } elsif (scalar @tmp > 1) {
 453                        push @revs, reverse(command('rev-list',@tmp));
 454                } else {
 455                        fatal "Failed to rev-parse $c";
 456                }
 457        }
 458        my $gs = Git::SVN->new;
 459        my ($r_last, $cmt_last) = $gs->last_rev_commit;
 460        $gs->fetch;
 461        if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
 462                fatal "There are new revisions that were fetched ",
 463                      "and need to be merged (or acknowledged) ",
 464                      "before committing.\nlast rev: $r_last\n",
 465                      " current: $gs->{last_rev}";
 466        }
 467        $gs->set_tree($_) foreach @revs;
 468        print "Done committing ",scalar @revs," revisions to SVN\n";
 469        unlink $gs->{index};
 470}
 471
 472sub cmd_dcommit {
 473        my $head = shift;
 474        git_cmd_try { command_oneline(qw/diff-index --quiet HEAD/) }
 475                'Cannot dcommit with a dirty index.  Commit your changes first, '
 476                . "or stash them with `git stash'.\n";
 477        $head ||= 'HEAD';
 478
 479        my $old_head;
 480        if ($head ne 'HEAD') {
 481                $old_head = eval {
 482                        command_oneline([qw/symbolic-ref -q HEAD/])
 483                };
 484                if ($old_head) {
 485                        $old_head =~ s{^refs/heads/}{};
 486                } else {
 487                        $old_head = eval { command_oneline(qw/rev-parse HEAD/) };
 488                }
 489                command(['checkout', $head], STDERR => 0);
 490        }
 491
 492        my @refs;
 493        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD', \@refs);
 494        unless ($gs) {
 495                die "Unable to determine upstream SVN information from ",
 496                    "$head history.\nPerhaps the repository is empty.";
 497        }
 498
 499        if (defined $_commit_url) {
 500                $url = $_commit_url;
 501        } else {
 502                $url = eval { command_oneline('config', '--get',
 503                              "svn-remote.$gs->{repo_id}.commiturl") };
 504                if (!$url) {
 505                        $url = $gs->full_url
 506                }
 507        }
 508
 509        my $last_rev = $_revision if defined $_revision;
 510        if ($url) {
 511                print "Committing to $url ...\n";
 512        }
 513        my ($linear_refs, $parents) = linearize_history($gs, \@refs);
 514        if ($_no_rebase && scalar(@$linear_refs) > 1) {
 515                warn "Attempting to commit more than one change while ",
 516                     "--no-rebase is enabled.\n",
 517                     "If these changes depend on each other, re-running ",
 518                     "without --no-rebase may be required."
 519        }
 520        my $expect_url = $url;
 521        Git::SVN::remove_username($expect_url);
 522        while (1) {
 523                my $d = shift @$linear_refs or last;
 524                unless (defined $last_rev) {
 525                        (undef, $last_rev, undef) = cmt_metadata("$d~1");
 526                        unless (defined $last_rev) {
 527                                fatal "Unable to extract revision information ",
 528                                      "from commit $d~1";
 529                        }
 530                }
 531                if ($_dry_run) {
 532                        print "diff-tree $d~1 $d\n";
 533                } else {
 534                        my $cmt_rev;
 535                        my %ed_opts = ( r => $last_rev,
 536                                        log => get_commit_entry($d)->{log},
 537                                        ra => Git::SVN::Ra->new($url),
 538                                        config => SVN::Core::config_get_config(
 539                                                $Git::SVN::Ra::config_dir
 540                                        ),
 541                                        tree_a => "$d~1",
 542                                        tree_b => $d,
 543                                        editor_cb => sub {
 544                                               print "Committed r$_[0]\n";
 545                                               $cmt_rev = $_[0];
 546                                        },
 547                                        svn_path => '');
 548                        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
 549                                print "No changes\n$d~1 == $d\n";
 550                        } elsif ($parents->{$d} && @{$parents->{$d}}) {
 551                                $gs->{inject_parents_dcommit}->{$cmt_rev} =
 552                                                               $parents->{$d};
 553                        }
 554                        $_fetch_all ? $gs->fetch_all : $gs->fetch;
 555                        $last_rev = $cmt_rev;
 556                        next if $_no_rebase;
 557
 558                        # we always want to rebase against the current HEAD,
 559                        # not any head that was passed to us
 560                        my @diff = command('diff-tree', $d,
 561                                           $gs->refname, '--');
 562                        my @finish;
 563                        if (@diff) {
 564                                @finish = rebase_cmd();
 565                                print STDERR "W: $d and ", $gs->refname,
 566                                             " differ, using @finish:\n",
 567                                             join("\n", @diff), "\n";
 568                        } else {
 569                                print "No changes between current HEAD and ",
 570                                      $gs->refname,
 571                                      "\nResetting to the latest ",
 572                                      $gs->refname, "\n";
 573                                @finish = qw/reset --mixed/;
 574                        }
 575                        command_noisy(@finish, $gs->refname);
 576                        if (@diff) {
 577                                @refs = ();
 578                                my ($url_, $rev_, $uuid_, $gs_) =
 579                                              working_head_info('HEAD', \@refs);
 580                                my ($linear_refs_, $parents_) =
 581                                              linearize_history($gs_, \@refs);
 582                                if (scalar(@$linear_refs) !=
 583                                    scalar(@$linear_refs_)) {
 584                                        fatal "# of revisions changed ",
 585                                          "\nbefore:\n",
 586                                          join("\n", @$linear_refs),
 587                                          "\n\nafter:\n",
 588                                          join("\n", @$linear_refs_), "\n",
 589                                          'If you are attempting to commit ',
 590                                          "merges, try running:\n\t",
 591                                          'git rebase --interactive',
 592                                          '--preserve-merges ',
 593                                          $gs->refname,
 594                                          "\nBefore dcommitting";
 595                                }
 596                                if ($url_ ne $expect_url) {
 597                                        fatal "URL mismatch after rebase: ",
 598                                              "$url_ != $expect_url";
 599                                }
 600                                if ($uuid_ ne $uuid) {
 601                                        fatal "uuid mismatch after rebase: ",
 602                                              "$uuid_ != $uuid";
 603                                }
 604                                # remap parents
 605                                my (%p, @l, $i);
 606                                for ($i = 0; $i < scalar @$linear_refs; $i++) {
 607                                        my $new = $linear_refs_->[$i] or next;
 608                                        $p{$new} =
 609                                                $parents->{$linear_refs->[$i]};
 610                                        push @l, $new;
 611                                }
 612                                $parents = \%p;
 613                                $linear_refs = \@l;
 614                        }
 615                }
 616        }
 617
 618        if ($old_head) {
 619                my $new_head = command_oneline(qw/rev-parse HEAD/);
 620                my $new_is_symbolic = eval {
 621                        command_oneline(qw/symbolic-ref -q HEAD/);
 622                };
 623                if ($new_is_symbolic) {
 624                        print "dcommitted the branch ", $head, "\n";
 625                } else {
 626                        print "dcommitted on a detached HEAD because you gave ",
 627                              "a revision argument.\n",
 628                              "The rewritten commit is: ", $new_head, "\n";
 629                }
 630                command(['checkout', $old_head], STDERR => 0);
 631        }
 632
 633        unlink $gs->{index};
 634}
 635
 636sub cmd_branch {
 637        my ($branch_name, $head) = @_;
 638
 639        unless (defined $branch_name && length $branch_name) {
 640                die(($_tag ? "tag" : "branch") . " name required\n");
 641        }
 642        $head ||= 'HEAD';
 643
 644        my ($src, $rev, undef, $gs) = working_head_info($head);
 645
 646        my $remote = Git::SVN::read_all_remotes()->{$gs->{repo_id}};
 647        my $allglobs = $remote->{ $_tag ? 'tags' : 'branches' };
 648        my $glob;
 649        if ($#{$allglobs} == 0) {
 650                $glob = $allglobs->[0];
 651        } else {
 652                unless(defined $_branch_dest) {
 653                        die "Multiple ",
 654                            $_tag ? "tag" : "branch",
 655                            " paths defined for Subversion repository.\n",
 656                            "You must specify where you want to create the ",
 657                            $_tag ? "tag" : "branch",
 658                            " with the --destination argument.\n";
 659                }
 660                foreach my $g (@{$allglobs}) {
 661                        # SVN::Git::Editor could probably be moved to Git.pm..
 662                        my $re = SVN::Git::Editor::glob2pat($g->{path}->{left});
 663                        if ($_branch_dest =~ /$re/) {
 664                                $glob = $g;
 665                                last;
 666                        }
 667                }
 668                unless (defined $glob) {
 669                        my $dest_re = qr/\b\Q$_branch_dest\E\b/;
 670                        foreach my $g (@{$allglobs}) {
 671                                $g->{path}->{left} =~ /$dest_re/ or next;
 672                                if (defined $glob) {
 673                                        die "Ambiguous destination: ",
 674                                            $_branch_dest, "\nmatches both '",
 675                                            $glob->{path}->{left}, "' and '",
 676                                            $g->{path}->{left}, "'\n";
 677                                }
 678                                $glob = $g;
 679                        }
 680                        unless (defined $glob) {
 681                                die "Unknown ",
 682                                    $_tag ? "tag" : "branch",
 683                                    " destination $_branch_dest\n";
 684                        }
 685                }
 686        }
 687        my ($lft, $rgt) = @{ $glob->{path} }{qw/left right/};
 688        my $dst = join '/', $remote->{url}, $lft, $branch_name, ($rgt || ());
 689
 690        my $ctx = SVN::Client->new(
 691                auth    => Git::SVN::Ra::_auth_providers(),
 692                log_msg => sub {
 693                        ${ $_[0] } = defined $_message
 694                                ? $_message
 695                                : 'Create ' . ($_tag ? 'tag ' : 'branch ' )
 696                                . $branch_name;
 697                },
 698        );
 699
 700        eval {
 701                $ctx->ls($dst, 'HEAD', 0);
 702        } and die "branch ${branch_name} already exists\n";
 703
 704        print "Copying ${src} at r${rev} to ${dst}...\n";
 705        $ctx->copy($src, $rev, $dst)
 706                unless $_dry_run;
 707
 708        $gs->fetch_all;
 709}
 710
 711sub cmd_find_rev {
 712        my $revision_or_hash = shift or die "SVN or git revision required ",
 713                                            "as a command-line argument\n";
 714        my $result;
 715        if ($revision_or_hash =~ /^r\d+$/) {
 716                my $head = shift;
 717                $head ||= 'HEAD';
 718                my @refs;
 719                my (undef, undef, $uuid, $gs) = working_head_info($head, \@refs);
 720                unless ($gs) {
 721                        die "Unable to determine upstream SVN information from ",
 722                            "$head history\n";
 723                }
 724                my $desired_revision = substr($revision_or_hash, 1);
 725                $result = $gs->rev_map_get($desired_revision, $uuid);
 726        } else {
 727                my (undef, $rev, undef) = cmt_metadata($revision_or_hash);
 728                $result = $rev;
 729        }
 730        print "$result\n" if $result;
 731}
 732
 733sub cmd_rebase {
 734        command_noisy(qw/update-index --refresh/);
 735        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 736        unless ($gs) {
 737                die "Unable to determine upstream SVN information from ",
 738                    "working tree history\n";
 739        }
 740        if ($_dry_run) {
 741                print "Remote Branch: " . $gs->refname . "\n";
 742                print "SVN URL: " . $url . "\n";
 743                return;
 744        }
 745        if (command(qw/diff-index HEAD --/)) {
 746                print STDERR "Cannot rebase with uncommited changes:\n";
 747                command_noisy('status');
 748                exit 1;
 749        }
 750        unless ($_local) {
 751                # rebase will checkout for us, so no need to do it explicitly
 752                $_no_checkout = 'true';
 753                $_fetch_all ? $gs->fetch_all : $gs->fetch;
 754        }
 755        command_noisy(rebase_cmd(), $gs->refname);
 756}
 757
 758sub cmd_show_ignore {
 759        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 760        $gs ||= Git::SVN->new;
 761        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 762        $gs->prop_walk($gs->{path}, $r, sub {
 763                my ($gs, $path, $props) = @_;
 764                print STDOUT "\n# $path\n";
 765                my $s = $props->{'svn:ignore'} or return;
 766                $s =~ s/[\r\n]+/\n/g;
 767                $s =~ s/^\n+//;
 768                chomp $s;
 769                $s =~ s#^#$path#gm;
 770                print STDOUT "$s\n";
 771        });
 772}
 773
 774sub cmd_show_externals {
 775        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 776        $gs ||= Git::SVN->new;
 777        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 778        $gs->prop_walk($gs->{path}, $r, sub {
 779                my ($gs, $path, $props) = @_;
 780                print STDOUT "\n# $path\n";
 781                my $s = $props->{'svn:externals'} or return;
 782                $s =~ s/[\r\n]+/\n/g;
 783                chomp $s;
 784                $s =~ s#^#$path#gm;
 785                print STDOUT "$s\n";
 786        });
 787}
 788
 789sub cmd_create_ignore {
 790        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 791        $gs ||= Git::SVN->new;
 792        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 793        $gs->prop_walk($gs->{path}, $r, sub {
 794                my ($gs, $path, $props) = @_;
 795                # $path is of the form /path/to/dir/
 796                $path = '.' . $path;
 797                # SVN can have attributes on empty directories,
 798                # which git won't track
 799                mkpath([$path]) unless -d $path;
 800                my $ignore = $path . '.gitignore';
 801                my $s = $props->{'svn:ignore'} or return;
 802                open(GITIGNORE, '>', $ignore)
 803                  or fatal("Failed to open `$ignore' for writing: $!");
 804                $s =~ s/[\r\n]+/\n/g;
 805                $s =~ s/^\n+//;
 806                chomp $s;
 807                # Prefix all patterns so that the ignore doesn't apply
 808                # to sub-directories.
 809                $s =~ s#^#/#gm;
 810                print GITIGNORE "$s\n";
 811                close(GITIGNORE)
 812                  or fatal("Failed to close `$ignore': $!");
 813                command_noisy('add', '-f', $ignore);
 814        });
 815}
 816
 817sub canonicalize_path {
 818        my ($path) = @_;
 819        my $dot_slash_added = 0;
 820        if (substr($path, 0, 1) ne "/") {
 821                $path = "./" . $path;
 822                $dot_slash_added = 1;
 823        }
 824        # File::Spec->canonpath doesn't collapse x/../y into y (for a
 825        # good reason), so let's do this manually.
 826        $path =~ s#/+#/#g;
 827        $path =~ s#/\.(?:/|$)#/#g;
 828        $path =~ s#/[^/]+/\.\.##g;
 829        $path =~ s#/$##g;
 830        $path =~ s#^\./## if $dot_slash_added;
 831        $path =~ s#^/##;
 832        $path =~ s#^\.$##;
 833        return $path;
 834}
 835
 836sub canonicalize_url {
 837        my ($url) = @_;
 838        $url =~ s#^([^:]+://[^/]*/)(.*)$#$1 . canonicalize_path($2)#e;
 839        return $url;
 840}
 841
 842# get_svnprops(PATH)
 843# ------------------
 844# Helper for cmd_propget and cmd_proplist below.
 845sub get_svnprops {
 846        my $path = shift;
 847        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 848        $gs ||= Git::SVN->new;
 849
 850        # prefix THE PATH by the sub-directory from which the user
 851        # invoked us.
 852        $path = $cmd_dir_prefix . $path;
 853        fatal("No such file or directory: $path") unless -e $path;
 854        my $is_dir = -d $path ? 1 : 0;
 855        $path = $gs->{path} . '/' . $path;
 856
 857        # canonicalize the path (otherwise libsvn will abort or fail to
 858        # find the file)
 859        $path = canonicalize_path($path);
 860
 861        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 862        my $props;
 863        if ($is_dir) {
 864                (undef, undef, $props) = $gs->ra->get_dir($path, $r);
 865        }
 866        else {
 867                (undef, $props) = $gs->ra->get_file($path, $r, undef);
 868        }
 869        return $props;
 870}
 871
 872# cmd_propget (PROP, PATH)
 873# ------------------------
 874# Print the SVN property PROP for PATH.
 875sub cmd_propget {
 876        my ($prop, $path) = @_;
 877        $path = '.' if not defined $path;
 878        usage(1) if not defined $prop;
 879        my $props = get_svnprops($path);
 880        if (not defined $props->{$prop}) {
 881                fatal("`$path' does not have a `$prop' SVN property.");
 882        }
 883        print $props->{$prop} . "\n";
 884}
 885
 886# cmd_proplist (PATH)
 887# -------------------
 888# Print the list of SVN properties for PATH.
 889sub cmd_proplist {
 890        my $path = shift;
 891        $path = '.' if not defined $path;
 892        my $props = get_svnprops($path);
 893        print "Properties on '$path':\n";
 894        foreach (sort keys %{$props}) {
 895                print "  $_\n";
 896        }
 897}
 898
 899sub cmd_multi_init {
 900        my $url = shift;
 901        unless (defined $_trunk || @_branches || @_tags) {
 902                usage(1);
 903        }
 904
 905        $_prefix = '' unless defined $_prefix;
 906        if (defined $url) {
 907                $url = canonicalize_url($url);
 908                init_subdir(@_);
 909        }
 910        do_git_init_db();
 911        if (defined $_trunk) {
 912                my $trunk_ref = $_prefix . 'trunk';
 913                # try both old-style and new-style lookups:
 914                my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
 915                unless ($gs_trunk) {
 916                        my ($trunk_url, $trunk_path) =
 917                                              complete_svn_url($url, $_trunk);
 918                        $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
 919                                                   undef, $trunk_ref);
 920                }
 921        }
 922        return unless @_branches || @_tags;
 923        my $ra = $url ? Git::SVN::Ra->new($url) : undef;
 924        foreach my $path (@_branches) {
 925                complete_url_ls_init($ra, $path, '--branches/-b', $_prefix);
 926        }
 927        foreach my $path (@_tags) {
 928                complete_url_ls_init($ra, $path, '--tags/-t', $_prefix.'tags/');
 929        }
 930}
 931
 932sub cmd_multi_fetch {
 933        my $remotes = Git::SVN::read_all_remotes();
 934        foreach my $repo_id (sort keys %$remotes) {
 935                if ($remotes->{$repo_id}->{url}) {
 936                        Git::SVN::fetch_all($repo_id, $remotes);
 937                }
 938        }
 939}
 940
 941# this command is special because it requires no metadata
 942sub cmd_commit_diff {
 943        my ($ta, $tb, $url) = @_;
 944        my $usage = "Usage: $0 commit-diff -r<revision> ".
 945                    "<tree-ish> <tree-ish> [<URL>]";
 946        fatal($usage) if (!defined $ta || !defined $tb);
 947        my $svn_path = '';
 948        if (!defined $url) {
 949                my $gs = eval { Git::SVN->new };
 950                if (!$gs) {
 951                        fatal("Needed URL or usable git-svn --id in ",
 952                              "the command-line\n", $usage);
 953                }
 954                $url = $gs->{url};
 955                $svn_path = $gs->{path};
 956        }
 957        unless (defined $_revision) {
 958                fatal("-r|--revision is a required argument\n", $usage);
 959        }
 960        if (defined $_message && defined $_file) {
 961                fatal("Both --message/-m and --file/-F specified ",
 962                      "for the commit message.\n",
 963                      "I have no idea what you mean");
 964        }
 965        if (defined $_file) {
 966                $_message = file_to_s($_file);
 967        } else {
 968                $_message ||= get_commit_entry($tb)->{log};
 969        }
 970        my $ra ||= Git::SVN::Ra->new($url);
 971        my $r = $_revision;
 972        if ($r eq 'HEAD') {
 973                $r = $ra->get_latest_revnum;
 974        } elsif ($r !~ /^\d+$/) {
 975                die "revision argument: $r not understood by git-svn\n";
 976        }
 977        my %ed_opts = ( r => $r,
 978                        log => $_message,
 979                        ra => $ra,
 980                        tree_a => $ta,
 981                        tree_b => $tb,
 982                        editor_cb => sub { print "Committed r$_[0]\n" },
 983                        svn_path => $svn_path );
 984        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
 985                print "No changes\n$ta == $tb\n";
 986        }
 987}
 988
 989sub escape_uri_only {
 990        my ($uri) = @_;
 991        my @tmp;
 992        foreach (split m{/}, $uri) {
 993                s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
 994                push @tmp, $_;
 995        }
 996        join('/', @tmp);
 997}
 998
 999sub escape_url {
1000        my ($url) = @_;
1001        if ($url =~ m#^([^:]+)://([^/]*)(.*)$#) {
1002                my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
1003                $url = "$scheme://$domain$uri";
1004        }
1005        $url;
1006}
1007
1008sub cmd_info {
1009        my $path = canonicalize_path(defined($_[0]) ? $_[0] : ".");
1010        my $fullpath = canonicalize_path($cmd_dir_prefix . $path);
1011        if (exists $_[1]) {
1012                die "Too many arguments specified\n";
1013        }
1014
1015        my ($file_type, $diff_status) = find_file_type_and_diff_status($path);
1016
1017        if (!$file_type && !$diff_status) {
1018                print STDERR "svn: '$path' is not under version control\n";
1019                exit 1;
1020        }
1021
1022        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1023        unless ($gs) {
1024                die "Unable to determine upstream SVN information from ",
1025                    "working tree history\n";
1026        }
1027
1028        # canonicalize_path() will return "" to make libsvn 1.5.x happy,
1029        $path = "." if $path eq "";
1030
1031        my $full_url = $url . ($fullpath eq "" ? "" : "/$fullpath");
1032
1033        if ($_url) {
1034                print escape_url($full_url), "\n";
1035                return;
1036        }
1037
1038        my $result = "Path: $path\n";
1039        $result .= "Name: " . basename($path) . "\n" if $file_type ne "dir";
1040        $result .= "URL: " . escape_url($full_url) . "\n";
1041
1042        eval {
1043                my $repos_root = $gs->repos_root;
1044                Git::SVN::remove_username($repos_root);
1045                $result .= "Repository Root: " . escape_url($repos_root) . "\n";
1046        };
1047        if ($@) {
1048                $result .= "Repository Root: (offline)\n";
1049        }
1050        $result .= "Repository UUID: $uuid\n" unless $diff_status eq "A" &&
1051                ($SVN::Core::VERSION le '1.5.4' || $file_type ne "dir");
1052        $result .= "Revision: " . ($diff_status eq "A" ? 0 : $rev) . "\n";
1053
1054        $result .= "Node Kind: " .
1055                   ($file_type eq "dir" ? "directory" : "file") . "\n";
1056
1057        my $schedule = $diff_status eq "A"
1058                       ? "add"
1059                       : ($diff_status eq "D" ? "delete" : "normal");
1060        $result .= "Schedule: $schedule\n";
1061
1062        if ($diff_status eq "A") {
1063                print $result, "\n";
1064                return;
1065        }
1066
1067        my ($lc_author, $lc_rev, $lc_date_utc);
1068        my @args = Git::SVN::Log::git_svn_log_cmd($rev, $rev, "--", $fullpath);
1069        my $log = command_output_pipe(@args);
1070        my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
1071        while (<$log>) {
1072                if (/^${esc_color}author (.+) <[^>]+> (\d+) ([\-\+]?\d+)$/o) {
1073                        $lc_author = $1;
1074                        $lc_date_utc = Git::SVN::Log::parse_git_date($2, $3);
1075                } elsif (/^${esc_color}    (git-svn-id:.+)$/o) {
1076                        (undef, $lc_rev, undef) = ::extract_metadata($1);
1077                }
1078        }
1079        close $log;
1080
1081        Git::SVN::Log::set_local_timezone();
1082
1083        $result .= "Last Changed Author: $lc_author\n";
1084        $result .= "Last Changed Rev: $lc_rev\n";
1085        $result .= "Last Changed Date: " .
1086                   Git::SVN::Log::format_svn_date($lc_date_utc) . "\n";
1087
1088        if ($file_type ne "dir") {
1089                my $text_last_updated_date =
1090                    ($diff_status eq "D" ? $lc_date_utc : (stat $path)[9]);
1091                $result .=
1092                    "Text Last Updated: " .
1093                    Git::SVN::Log::format_svn_date($text_last_updated_date) .
1094                    "\n";
1095                my $checksum;
1096                if ($diff_status eq "D") {
1097                        my ($fh, $ctx) =
1098                            command_output_pipe(qw(cat-file blob), "HEAD:$path");
1099                        if ($file_type eq "link") {
1100                                my $file_name = <$fh>;
1101                                $checksum = md5sum("link $file_name");
1102                        } else {
1103                                $checksum = md5sum($fh);
1104                        }
1105                        command_close_pipe($fh, $ctx);
1106                } elsif ($file_type eq "link") {
1107                        my $file_name =
1108                            command(qw(cat-file blob), "HEAD:$path");
1109                        $checksum =
1110                            md5sum("link " . $file_name);
1111                } else {
1112                        open FILE, "<", $path or die $!;
1113                        $checksum = md5sum(\*FILE);
1114                        close FILE or die $!;
1115                }
1116                $result .= "Checksum: " . $checksum . "\n";
1117        }
1118
1119        print $result, "\n";
1120}
1121
1122sub cmd_reset {
1123        my $target = shift || $_revision or die "SVN revision required\n";
1124        $target = $1 if $target =~ /^r(\d+)$/;
1125        $target =~ /^\d+$/ or die "Numeric SVN revision expected\n";
1126        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1127        unless ($gs) {
1128                die "Unable to determine upstream SVN information from ".
1129                    "history\n";
1130        }
1131        my ($r, $c) = $gs->find_rev_before($target, not $_fetch_parent);
1132        $gs->rev_map_set($r, $c, 'reset', $uuid);
1133        print "r$r = $c ($gs->{ref_id})\n";
1134}
1135
1136sub cmd_gc {
1137        if (!$can_compress) {
1138                warn "Compress::Zlib could not be found; unhandled.log " .
1139                     "files will not be compressed.\n";
1140        }
1141        find({ wanted => \&gc_directory, no_chdir => 1}, "$ENV{GIT_DIR}/svn");
1142}
1143
1144########################### utility functions #########################
1145
1146sub rebase_cmd {
1147        my @cmd = qw/rebase/;
1148        push @cmd, '-v' if $_verbose;
1149        push @cmd, qw/--merge/ if $_merge;
1150        push @cmd, "--strategy=$_strategy" if $_strategy;
1151        @cmd;
1152}
1153
1154sub post_fetch_checkout {
1155        return if $_no_checkout;
1156        my $gs = $Git::SVN::_head or return;
1157        return if verify_ref('refs/heads/master^0');
1158
1159        my $valid_head = verify_ref('HEAD^0');
1160        command_noisy(qw(update-ref refs/heads/master), $gs->refname);
1161        return if ($valid_head || !verify_ref('HEAD^0'));
1162
1163        return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
1164        my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
1165        return if -f $index;
1166
1167        return if command_oneline(qw/rev-parse --is-inside-work-tree/) eq 'false';
1168        return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
1169        command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
1170        print STDERR "Checked out HEAD:\n  ",
1171                     $gs->full_url, " r", $gs->last_rev, "\n";
1172}
1173
1174sub complete_svn_url {
1175        my ($url, $path) = @_;
1176        $path =~ s#/+$##;
1177        if ($path !~ m#^[a-z\+]+://#) {
1178                if (!defined $url || $url !~ m#^[a-z\+]+://#) {
1179                        fatal("E: '$path' is not a complete URL ",
1180                              "and a separate URL is not specified");
1181                }
1182                return ($url, $path);
1183        }
1184        return ($path, '');
1185}
1186
1187sub complete_url_ls_init {
1188        my ($ra, $repo_path, $switch, $pfx) = @_;
1189        unless ($repo_path) {
1190                print STDERR "W: $switch not specified\n";
1191                return;
1192        }
1193        $repo_path =~ s#/+$##;
1194        if ($repo_path =~ m#^[a-z\+]+://#) {
1195                $ra = Git::SVN::Ra->new($repo_path);
1196                $repo_path = '';
1197        } else {
1198                $repo_path =~ s#^/+##;
1199                unless ($ra) {
1200                        fatal("E: '$repo_path' is not a complete URL ",
1201                              "and a separate URL is not specified");
1202                }
1203        }
1204        my $url = $ra->{url};
1205        my $gs = Git::SVN->init($url, undef, undef, undef, 1);
1206        my $k = "svn-remote.$gs->{repo_id}.url";
1207        my $orig_url = eval { command_oneline(qw/config --get/, $k) };
1208        if ($orig_url && ($orig_url ne $gs->{url})) {
1209                die "$k already set: $orig_url\n",
1210                    "wanted to set to: $gs->{url}\n";
1211        }
1212        command_oneline('config', $k, $gs->{url}) unless $orig_url;
1213        my $remote_path = "$gs->{path}/$repo_path";
1214        $remote_path =~ s#/+#/#g;
1215        $remote_path =~ s#^/##g;
1216        $remote_path .= "/*" if $remote_path !~ /\*/;
1217        my ($n) = ($switch =~ /^--(\w+)/);
1218        if (length $pfx && $pfx !~ m#/$#) {
1219                die "--prefix='$pfx' must have a trailing slash '/'\n";
1220        }
1221        command_noisy('config',
1222                      '--add',
1223                      "svn-remote.$gs->{repo_id}.$n",
1224                      "$remote_path:refs/remotes/$pfx*" .
1225                        ('/*' x (($remote_path =~ tr/*/*/) - 1)) );
1226}
1227
1228sub verify_ref {
1229        my ($ref) = @_;
1230        eval { command_oneline([ 'rev-parse', '--verify', $ref ],
1231                               { STDERR => 0 }); };
1232}
1233
1234sub get_tree_from_treeish {
1235        my ($treeish) = @_;
1236        # $treeish can be a symbolic ref, too:
1237        my $type = command_oneline(qw/cat-file -t/, $treeish);
1238        my $expected;
1239        while ($type eq 'tag') {
1240                ($treeish, $type) = command(qw/cat-file tag/, $treeish);
1241        }
1242        if ($type eq 'commit') {
1243                $expected = (grep /^tree /, command(qw/cat-file commit/,
1244                                                    $treeish))[0];
1245                ($expected) = ($expected =~ /^tree ($sha1)$/o);
1246                die "Unable to get tree from $treeish\n" unless $expected;
1247        } elsif ($type eq 'tree') {
1248                $expected = $treeish;
1249        } else {
1250                die "$treeish is a $type, expected tree, tag or commit\n";
1251        }
1252        return $expected;
1253}
1254
1255sub get_commit_entry {
1256        my ($treeish) = shift;
1257        my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
1258        my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
1259        my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
1260        open my $log_fh, '>', $commit_editmsg or croak $!;
1261
1262        my $type = command_oneline(qw/cat-file -t/, $treeish);
1263        if ($type eq 'commit' || $type eq 'tag') {
1264                my ($msg_fh, $ctx) = command_output_pipe('cat-file',
1265                                                         $type, $treeish);
1266                my $in_msg = 0;
1267                my $author;
1268                my $saw_from = 0;
1269                my $msgbuf = "";
1270                while (<$msg_fh>) {
1271                        if (!$in_msg) {
1272                                $in_msg = 1 if (/^\s*$/);
1273                                $author = $1 if (/^author (.*>)/);
1274                        } elsif (/^git-svn-id: /) {
1275                                # skip this for now, we regenerate the
1276                                # correct one on re-fetch anyways
1277                                # TODO: set *:merge properties or like...
1278                        } else {
1279                                if (/^From:/ || /^Signed-off-by:/) {
1280                                        $saw_from = 1;
1281                                }
1282                                $msgbuf .= $_;
1283                        }
1284                }
1285                $msgbuf =~ s/\s+$//s;
1286                if ($Git::SVN::_add_author_from && defined($author)
1287                    && !$saw_from) {
1288                        $msgbuf .= "\n\nFrom: $author";
1289                }
1290                print $log_fh $msgbuf or croak $!;
1291                command_close_pipe($msg_fh, $ctx);
1292        }
1293        close $log_fh or croak $!;
1294
1295        if ($_edit || ($type eq 'tree')) {
1296                my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
1297                # TODO: strip out spaces, comments, like git-commit.sh
1298                system($editor, $commit_editmsg);
1299        }
1300        rename $commit_editmsg, $commit_msg or croak $!;
1301        {
1302                require Encode;
1303                # SVN requires messages to be UTF-8 when entering the repo
1304                local $/;
1305                open $log_fh, '<', $commit_msg or croak $!;
1306                binmode $log_fh;
1307                chomp($log_entry{log} = <$log_fh>);
1308
1309                my $enc = Git::config('i18n.commitencoding') || 'UTF-8';
1310                my $msg = $log_entry{log};
1311
1312                eval { $msg = Encode::decode($enc, $msg, 1) };
1313                if ($@) {
1314                        die "Could not decode as $enc:\n", $msg,
1315                            "\nPerhaps you need to set i18n.commitencoding\n";
1316                }
1317
1318                eval { $msg = Encode::encode('UTF-8', $msg, 1) };
1319                die "Could not encode as UTF-8:\n$msg\n" if $@;
1320
1321                $log_entry{log} = $msg;
1322
1323                close $log_fh or croak $!;
1324        }
1325        unlink $commit_msg;
1326        \%log_entry;
1327}
1328
1329sub s_to_file {
1330        my ($str, $file, $mode) = @_;
1331        open my $fd,'>',$file or croak $!;
1332        print $fd $str,"\n" or croak $!;
1333        close $fd or croak $!;
1334        chmod ($mode &~ umask, $file) if (defined $mode);
1335}
1336
1337sub file_to_s {
1338        my $file = shift;
1339        open my $fd,'<',$file or croak "$!: file: $file\n";
1340        local $/;
1341        my $ret = <$fd>;
1342        close $fd or croak $!;
1343        $ret =~ s/\s*$//s;
1344        return $ret;
1345}
1346
1347# '<svn username> = real-name <email address>' mapping based on git-svnimport:
1348sub load_authors {
1349        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1350        my $log = $cmd eq 'log';
1351        while (<$authors>) {
1352                chomp;
1353                next unless /^(.+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
1354                my ($user, $name, $email) = ($1, $2, $3);
1355                if ($log) {
1356                        $Git::SVN::Log::rusers{"$name <$email>"} = $user;
1357                } else {
1358                        $users{$user} = [$name, $email];
1359                }
1360        }
1361        close $authors or croak $!;
1362}
1363
1364# convert GetOpt::Long specs for use by git-config
1365sub read_repo_config {
1366        return unless -d $ENV{GIT_DIR};
1367        my $opts = shift;
1368        my @config_only;
1369        foreach my $o (keys %$opts) {
1370                # if we have mixedCase and a long option-only, then
1371                # it's a config-only variable that we don't need for
1372                # the command-line.
1373                push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
1374                my $v = $opts->{$o};
1375                my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
1376                $key =~ s/-//g;
1377                my $arg = 'git config';
1378                $arg .= ' --int' if ($o =~ /[:=]i$/);
1379                $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
1380                if (ref $v eq 'ARRAY') {
1381                        chomp(my @tmp = `$arg --get-all svn.$key`);
1382                        @$v = @tmp if @tmp;
1383                } else {
1384                        chomp(my $tmp = `$arg --get svn.$key`);
1385                        if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
1386                                $$v = $tmp;
1387                        }
1388                }
1389        }
1390        delete @$opts{@config_only} if @config_only;
1391}
1392
1393sub extract_metadata {
1394        my $id = shift or return (undef, undef, undef);
1395        my ($url, $rev, $uuid) = ($id =~ /^\s*git-svn-id:\s+(.*)\@(\d+)
1396                                                        \s([a-f\d\-]+)$/ix);
1397        if (!defined $rev || !$uuid || !$url) {
1398                # some of the original repositories I made had
1399                # identifiers like this:
1400                ($rev, $uuid) = ($id =~/^\s*git-svn-id:\s(\d+)\@([a-f\d\-]+)/i);
1401        }
1402        return ($url, $rev, $uuid);
1403}
1404
1405sub cmt_metadata {
1406        return extract_metadata((grep(/^git-svn-id: /,
1407                command(qw/cat-file commit/, shift)))[-1]);
1408}
1409
1410sub cmt_sha2rev_batch {
1411        my %s2r;
1412        my ($pid, $in, $out, $ctx) = command_bidi_pipe(qw/cat-file --batch/);
1413        my $list = shift;
1414
1415        foreach my $sha (@{$list}) {
1416                my $first = 1;
1417                my $size = 0;
1418                print $out $sha, "\n";
1419
1420                while (my $line = <$in>) {
1421                        if ($first && $line =~ /^[[:xdigit:]]{40}\smissing$/) {
1422                                last;
1423                        } elsif ($first &&
1424                               $line =~ /^[[:xdigit:]]{40}\scommit\s(\d+)$/) {
1425                                $first = 0;
1426                                $size = $1;
1427                                next;
1428                        } elsif ($line =~ /^(git-svn-id: )/) {
1429                                my (undef, $rev, undef) =
1430                                                      extract_metadata($line);
1431                                $s2r{$sha} = $rev;
1432                        }
1433
1434                        $size -= length($line);
1435                        last if ($size == 0);
1436                }
1437        }
1438
1439        command_close_bidi_pipe($pid, $in, $out, $ctx);
1440
1441        return \%s2r;
1442}
1443
1444sub working_head_info {
1445        my ($head, $refs) = @_;
1446        my @args = ('log', '--no-color', '--first-parent', '--pretty=medium');
1447        my ($fh, $ctx) = command_output_pipe(@args, $head);
1448        my $hash;
1449        my %max;
1450        while (<$fh>) {
1451                if ( m{^commit ($::sha1)$} ) {
1452                        unshift @$refs, $hash if $hash and $refs;
1453                        $hash = $1;
1454                        next;
1455                }
1456                next unless s{^\s*(git-svn-id:)}{$1};
1457                my ($url, $rev, $uuid) = extract_metadata($_);
1458                if (defined $url && defined $rev) {
1459                        next if $max{$url} and $max{$url} < $rev;
1460                        if (my $gs = Git::SVN->find_by_url($url)) {
1461                                my $c = $gs->rev_map_get($rev, $uuid);
1462                                if ($c && $c eq $hash) {
1463                                        close $fh; # break the pipe
1464                                        return ($url, $rev, $uuid, $gs);
1465                                } else {
1466                                        $max{$url} ||= $gs->rev_map_max;
1467                                }
1468                        }
1469                }
1470        }
1471        command_close_pipe($fh, $ctx);
1472        (undef, undef, undef, undef);
1473}
1474
1475sub read_commit_parents {
1476        my ($parents, $c) = @_;
1477        chomp(my $p = command_oneline(qw/rev-list --parents -1/, $c));
1478        $p =~ s/^($c)\s*// or die "rev-list --parents -1 $c failed!\n";
1479        @{$parents->{$c}} = split(/ /, $p);
1480}
1481
1482sub linearize_history {
1483        my ($gs, $refs) = @_;
1484        my %parents;
1485        foreach my $c (@$refs) {
1486                read_commit_parents(\%parents, $c);
1487        }
1488
1489        my @linear_refs;
1490        my %skip = ();
1491        my $last_svn_commit = $gs->last_commit;
1492        foreach my $c (reverse @$refs) {
1493                next if $c eq $last_svn_commit;
1494                last if $skip{$c};
1495
1496                unshift @linear_refs, $c;
1497                $skip{$c} = 1;
1498
1499                # we only want the first parent to diff against for linear
1500                # history, we save the rest to inject when we finalize the
1501                # svn commit
1502                my $fp_a = verify_ref("$c~1");
1503                my $fp_b = shift @{$parents{$c}} if $parents{$c};
1504                if (!$fp_a || !$fp_b) {
1505                        die "Commit $c\n",
1506                            "has no parent commit, and therefore ",
1507                            "nothing to diff against.\n",
1508                            "You should be working from a repository ",
1509                            "originally created by git-svn\n";
1510                }
1511                if ($fp_a ne $fp_b) {
1512                        die "$c~1 = $fp_a, however parsing commit $c ",
1513                            "revealed that:\n$c~1 = $fp_b\nBUG!\n";
1514                }
1515
1516                foreach my $p (@{$parents{$c}}) {
1517                        $skip{$p} = 1;
1518                }
1519        }
1520        (\@linear_refs, \%parents);
1521}
1522
1523sub find_file_type_and_diff_status {
1524        my ($path) = @_;
1525        return ('dir', '') if $path eq '';
1526
1527        my $diff_output =
1528            command_oneline(qw(diff --cached --name-status --), $path) || "";
1529        my $diff_status = (split(' ', $diff_output))[0] || "";
1530
1531        my $ls_tree = command_oneline(qw(ls-tree HEAD), $path) || "";
1532
1533        return (undef, undef) if !$diff_status && !$ls_tree;
1534
1535        if ($diff_status eq "A") {
1536                return ("link", $diff_status) if -l $path;
1537                return ("dir", $diff_status) if -d $path;
1538                return ("file", $diff_status);
1539        }
1540
1541        my $mode = (split(' ', $ls_tree))[0] || "";
1542
1543        return ("link", $diff_status) if $mode eq "120000";
1544        return ("dir", $diff_status) if $mode eq "040000";
1545        return ("file", $diff_status);
1546}
1547
1548sub md5sum {
1549        my $arg = shift;
1550        my $ref = ref $arg;
1551        my $md5 = Digest::MD5->new();
1552        if ($ref eq 'GLOB' || $ref eq 'IO::File' || $ref eq 'File::Temp') {
1553                $md5->addfile($arg) or croak $!;
1554        } elsif ($ref eq 'SCALAR') {
1555                $md5->add($$arg) or croak $!;
1556        } elsif (!$ref) {
1557                $md5->add($arg) or croak $!;
1558        } else {
1559                ::fatal "Can't provide MD5 hash for unknown ref type: '", $ref, "'";
1560        }
1561        return $md5->hexdigest();
1562}
1563
1564sub gc_directory {
1565        if ($can_compress && -f $_ && basename($_) eq "unhandled.log") {
1566                my $out_filename = $_ . ".gz";
1567                open my $in_fh, "<", $_ or die "Unable to open $_: $!\n";
1568                binmode $in_fh;
1569                my $gz = Compress::Zlib::gzopen($out_filename, "ab") or
1570                                die "Unable to open $out_filename: $!\n";
1571
1572                my $res;
1573                while ($res = sysread($in_fh, my $str, 1024)) {
1574                        $gz->gzwrite($str) or
1575                                die "Unable to write: ".$gz->gzerror()."!\n";
1576                }
1577                unlink $_ or die "unlink $File::Find::name: $!\n";
1578        } elsif (-f $_ && basename($_) eq "index") {
1579                unlink $_ or die "unlink $_: $!\n";
1580        }
1581}
1582
1583package Git::SVN;
1584use strict;
1585use warnings;
1586use Fcntl qw/:DEFAULT :seek/;
1587use constant rev_map_fmt => 'NH40';
1588use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
1589            $_repack $_repack_flags $_use_svm_props $_head
1590            $_use_svnsync_props $no_reuse_existing $_minimize_url
1591            $_use_log_author $_add_author_from $_localtime/;
1592use Carp qw/croak/;
1593use File::Path qw/mkpath/;
1594use File::Copy qw/copy/;
1595use IPC::Open3;
1596
1597my ($_gc_nr, $_gc_period);
1598
1599# properties that we do not log:
1600my %SKIP_PROP;
1601BEGIN {
1602        %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
1603                                        svn:special svn:executable
1604                                        svn:entry:committed-rev
1605                                        svn:entry:last-author
1606                                        svn:entry:uuid
1607                                        svn:entry:committed-date/;
1608
1609        # some options are read globally, but can be overridden locally
1610        # per [svn-remote "..."] section.  Command-line options will *NOT*
1611        # override options set in an [svn-remote "..."] section
1612        no strict 'refs';
1613        for my $option (qw/follow_parent no_metadata use_svm_props
1614                           use_svnsync_props/) {
1615                my $key = $option;
1616                $key =~ tr/_//d;
1617                my $prop = "-$option";
1618                *$option = sub {
1619                        my ($self) = @_;
1620                        return $self->{$prop} if exists $self->{$prop};
1621                        my $k = "svn-remote.$self->{repo_id}.$key";
1622                        eval { command_oneline(qw/config --get/, $k) };
1623                        if ($@) {
1624                                $self->{$prop} = ${"Git::SVN::_$option"};
1625                        } else {
1626                                my $v = command_oneline(qw/config --bool/,$k);
1627                                $self->{$prop} = $v eq 'false' ? 0 : 1;
1628                        }
1629                        return $self->{$prop};
1630                }
1631        }
1632}
1633
1634
1635my (%LOCKFILES, %INDEX_FILES);
1636END {
1637        unlink keys %LOCKFILES if %LOCKFILES;
1638        unlink keys %INDEX_FILES if %INDEX_FILES;
1639}
1640
1641sub resolve_local_globs {
1642        my ($url, $fetch, $glob_spec) = @_;
1643        return unless defined $glob_spec;
1644        my $ref = $glob_spec->{ref};
1645        my $path = $glob_spec->{path};
1646        foreach (command(qw#for-each-ref --format=%(refname) refs/remotes#)) {
1647                next unless m#^refs/remotes/$ref->{regex}$#;
1648                my $p = $1;
1649                my $pathname = desanitize_refname($path->full_path($p));
1650                my $refname = desanitize_refname($ref->full_path($p));
1651                if (my $existing = $fetch->{$pathname}) {
1652                        if ($existing ne $refname) {
1653                                die "Refspec conflict:\n",
1654                                    "existing: refs/remotes/$existing\n",
1655                                    " globbed: refs/remotes/$refname\n";
1656                        }
1657                        my $u = (::cmt_metadata("refs/remotes/$refname"))[0];
1658                        $u =~ s!^\Q$url\E(/|$)!! or die
1659                          "refs/remotes/$refname: '$url' not found in '$u'\n";
1660                        if ($pathname ne $u) {
1661                                warn "W: Refspec glob conflict ",
1662                                     "(ref: refs/remotes/$refname):\n",
1663                                     "expected path: $pathname\n",
1664                                     "    real path: $u\n",
1665                                     "Continuing ahead with $u\n";
1666                                next;
1667                        }
1668                } else {
1669                        $fetch->{$pathname} = $refname;
1670                }
1671        }
1672}
1673
1674sub parse_revision_argument {
1675        my ($base, $head) = @_;
1676        if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
1677                return ($base, $head);
1678        }
1679        return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
1680        return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
1681        return ($head, $head) if ($::_revision eq 'HEAD');
1682        return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
1683        return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
1684        die "revision argument: $::_revision not understood by git-svn\n";
1685}
1686
1687sub fetch_all {
1688        my ($repo_id, $remotes) = @_;
1689        if (ref $repo_id) {
1690                my $gs = $repo_id;
1691                $repo_id = undef;
1692                $repo_id = $gs->{repo_id};
1693        }
1694        $remotes ||= read_all_remotes();
1695        my $remote = $remotes->{$repo_id} or
1696                     die "[svn-remote \"$repo_id\"] unknown\n";
1697        my $fetch = $remote->{fetch};
1698        my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
1699        my (@gs, @globs);
1700        my $ra = Git::SVN::Ra->new($url);
1701        my $uuid = $ra->get_uuid;
1702        my $head = $ra->get_latest_revnum;
1703        $ra->get_log("", $head, 0, 1, 0, 1, sub { $head = $_[1] });
1704        my $base = defined $fetch ? $head : 0;
1705
1706        # read the max revs for wildcard expansion (branches/*, tags/*)
1707        foreach my $t (qw/branches tags/) {
1708                defined $remote->{$t} or next;
1709                push @globs, @{$remote->{$t}};
1710
1711                my $max_rev = eval { tmp_config(qw/--int --get/,
1712                                         "svn-remote.$repo_id.${t}-maxRev") };
1713                if (defined $max_rev && ($max_rev < $base)) {
1714                        $base = $max_rev;
1715                } elsif (!defined $max_rev) {
1716                        $base = 0;
1717                }
1718        }
1719
1720        if ($fetch) {
1721                foreach my $p (sort keys %$fetch) {
1722                        my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
1723                        my $lr = $gs->rev_map_max;
1724                        if (defined $lr) {
1725                                $base = $lr if ($lr < $base);
1726                        }
1727                        push @gs, $gs;
1728                }
1729        }
1730
1731        ($base, $head) = parse_revision_argument($base, $head);
1732        $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
1733}
1734
1735sub read_all_remotes {
1736        my $r = {};
1737        my $use_svm_props = eval { command_oneline(qw/config --bool
1738            svn.useSvmProps/) };
1739        $use_svm_props = $use_svm_props eq 'true' if $use_svm_props;
1740        foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
1741                if (m!^(.+)\.fetch=\s*(.*)\s*:\s*(.+)\s*$!) {
1742                        my ($remote, $local_ref, $_remote_ref) = ($1, $2, $3);
1743                        die("svn-remote.$remote: remote ref '$_remote_ref' "
1744                            . "must start with 'refs/remotes/'\n")
1745                                unless $_remote_ref =~ m{^refs/remotes/(.+)};
1746                        my $remote_ref = $1;
1747                        $local_ref =~ s{^/}{};
1748                        $r->{$remote}->{fetch}->{$local_ref} = $remote_ref;
1749                        $r->{$remote}->{svm} = {} if $use_svm_props;
1750                } elsif (m!^(.+)\.usesvmprops=\s*(.*)\s*$!) {
1751                        $r->{$1}->{svm} = {};
1752                } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
1753                        $r->{$1}->{url} = $2;
1754                } elsif (m!^(.+)\.(branches|tags)=
1755                           (.*):refs/remotes/(.+)\s*$/!x) {
1756                        my ($p, $g) = ($3, $4);
1757                        my $rs = {
1758                            t => $2,
1759                            remote => $1,
1760                            path => Git::SVN::GlobSpec->new($p),
1761                            ref => Git::SVN::GlobSpec->new($g) };
1762                        if (length($rs->{ref}->{right}) != 0) {
1763                                die "The '*' glob character must be the last ",
1764                                    "character of '$g'\n";
1765                        }
1766                        push @{ $r->{$1}->{$2} }, $rs;
1767                }
1768        }
1769
1770        map {
1771                if (defined $r->{$_}->{svm}) {
1772                        my $svm;
1773                        eval {
1774                                my $section = "svn-remote.$_";
1775                                $svm = {
1776                                        source => tmp_config('--get',
1777                                            "$section.svm-source"),
1778                                        replace => tmp_config('--get',
1779                                            "$section.svm-replace"),
1780                                }
1781                        };
1782                        $r->{$_}->{svm} = $svm;
1783                }
1784        } keys %$r;
1785
1786        $r;
1787}
1788
1789sub init_vars {
1790        $_gc_nr = $_gc_period = 1000;
1791        if (defined $_repack || defined $_repack_flags) {
1792               warn "Repack options are obsolete; they have no effect.\n";
1793        }
1794}
1795
1796sub verify_remotes_sanity {
1797        return unless -d $ENV{GIT_DIR};
1798        my %seen;
1799        foreach (command(qw/config -l/)) {
1800                if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
1801                        if ($seen{$1}) {
1802                                die "Remote ref refs/remote/$1 is tracked by",
1803                                    "\n  \"$_\"\nand\n  \"$seen{$1}\"\n",
1804                                    "Please resolve this ambiguity in ",
1805                                    "your git configuration file before ",
1806                                    "continuing\n";
1807                        }
1808                        $seen{$1} = $_;
1809                }
1810        }
1811}
1812
1813sub find_existing_remote {
1814        my ($url, $remotes) = @_;
1815        return undef if $no_reuse_existing;
1816        my $existing;
1817        foreach my $repo_id (keys %$remotes) {
1818                my $u = $remotes->{$repo_id}->{url} or next;
1819                next if $u ne $url;
1820                $existing = $repo_id;
1821                last;
1822        }
1823        $existing;
1824}
1825
1826sub init_remote_config {
1827        my ($self, $url, $no_write) = @_;
1828        $url =~ s!/+$!!; # strip trailing slash
1829        my $r = read_all_remotes();
1830        my $existing = find_existing_remote($url, $r);
1831        if ($existing) {
1832                unless ($no_write) {
1833                        print STDERR "Using existing ",
1834                                     "[svn-remote \"$existing\"]\n";
1835                }
1836                $self->{repo_id} = $existing;
1837        } elsif ($_minimize_url) {
1838                my $min_url = Git::SVN::Ra->new($url)->minimize_url;
1839                $existing = find_existing_remote($min_url, $r);
1840                if ($existing) {
1841                        unless ($no_write) {
1842                                print STDERR "Using existing ",
1843                                             "[svn-remote \"$existing\"]\n";
1844                        }
1845                        $self->{repo_id} = $existing;
1846                }
1847                if ($min_url ne $url) {
1848                        unless ($no_write) {
1849                                print STDERR "Using higher level of URL: ",
1850                                             "$url => $min_url\n";
1851                        }
1852                        my $old_path = $self->{path};
1853                        $self->{path} = $url;
1854                        $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
1855                        if (length $old_path) {
1856                                $self->{path} .= "/$old_path";
1857                        }
1858                        $url = $min_url;
1859                }
1860        }
1861        my $orig_url;
1862        if (!$existing) {
1863                # verify that we aren't overwriting anything:
1864                $orig_url = eval {
1865                        command_oneline('config', '--get',
1866                                        "svn-remote.$self->{repo_id}.url")
1867                };
1868                if ($orig_url && ($orig_url ne $url)) {
1869                        die "svn-remote.$self->{repo_id}.url already set: ",
1870                            "$orig_url\nwanted to set to: $url\n";
1871                }
1872        }
1873        my ($xrepo_id, $xpath) = find_ref($self->refname);
1874        if (defined $xpath) {
1875                die "svn-remote.$xrepo_id.fetch already set to track ",
1876                    "$xpath:refs/remotes/", $self->refname, "\n";
1877        }
1878        unless ($no_write) {
1879                command_noisy('config',
1880                              "svn-remote.$self->{repo_id}.url", $url);
1881                $self->{path} =~ s{^/}{};
1882                command_noisy('config', '--add',
1883                              "svn-remote.$self->{repo_id}.fetch",
1884                              "$self->{path}:".$self->refname);
1885        }
1886        $self->{url} = $url;
1887}
1888
1889sub find_by_url { # repos_root and, path are optional
1890        my ($class, $full_url, $repos_root, $path) = @_;
1891
1892        return undef unless defined $full_url;
1893        remove_username($full_url);
1894        remove_username($repos_root) if defined $repos_root;
1895        my $remotes = read_all_remotes();
1896        if (defined $full_url && defined $repos_root && !defined $path) {
1897                $path = $full_url;
1898                $path =~ s#^\Q$repos_root\E(?:/|$)##;
1899        }
1900        foreach my $repo_id (keys %$remotes) {
1901                my $u = $remotes->{$repo_id}->{url} or next;
1902                remove_username($u);
1903                next if defined $repos_root && $repos_root ne $u;
1904
1905                my $fetch = $remotes->{$repo_id}->{fetch} || {};
1906                foreach my $t (qw/branches tags/) {
1907                        foreach my $globspec (@{$remotes->{$repo_id}->{$t}}) {
1908                                resolve_local_globs($u, $fetch, $globspec);
1909                        }
1910                }
1911                my $p = $path;
1912                my $rwr = rewrite_root({repo_id => $repo_id});
1913                my $svm = $remotes->{$repo_id}->{svm}
1914                        if defined $remotes->{$repo_id}->{svm};
1915                unless (defined $p) {
1916                        $p = $full_url;
1917                        my $z = $u;
1918                        my $prefix = '';
1919                        if ($rwr) {
1920                                $z = $rwr;
1921                                remove_username($z);
1922                        } elsif (defined $svm) {
1923                                $z = $svm->{source};
1924                                $prefix = $svm->{replace};
1925                                $prefix =~ s#^\Q$u\E(?:/|$)##;
1926                                $prefix =~ s#/$##;
1927                        }
1928                        $p =~ s#^\Q$z\E(?:/|$)#$prefix# or next;
1929                }
1930                foreach my $f (keys %$fetch) {
1931                        next if $f ne $p;
1932                        return Git::SVN->new($fetch->{$f}, $repo_id, $f);
1933                }
1934        }
1935        undef;
1936}
1937
1938sub init {
1939        my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
1940        my $self = _new($class, $repo_id, $ref_id, $path);
1941        if (defined $url) {
1942                $self->init_remote_config($url, $no_write);
1943        }
1944        $self;
1945}
1946
1947sub find_ref {
1948        my ($ref_id) = @_;
1949        foreach (command(qw/config -l/)) {
1950                next unless m!^svn-remote\.(.+)\.fetch=
1951                              \s*(.*)\s*:\s*refs/remotes/(.+)\s*$!x;
1952                my ($repo_id, $path, $ref) = ($1, $2, $3);
1953                if ($ref eq $ref_id) {
1954                        $path = '' if ($path =~ m#^\./?#);
1955                        return ($repo_id, $path);
1956                }
1957        }
1958        (undef, undef, undef);
1959}
1960
1961sub new {
1962        my ($class, $ref_id, $repo_id, $path) = @_;
1963        if (defined $ref_id && !defined $repo_id && !defined $path) {
1964                ($repo_id, $path) = find_ref($ref_id);
1965                if (!defined $repo_id) {
1966                        die "Could not find a \"svn-remote.*.fetch\" key ",
1967                            "in the repository configuration matching: ",
1968                            "refs/remotes/$ref_id\n";
1969                }
1970        }
1971        my $self = _new($class, $repo_id, $ref_id, $path);
1972        if (!defined $self->{path} || !length $self->{path}) {
1973                my $fetch = command_oneline('config', '--get',
1974                                            "svn-remote.$repo_id.fetch",
1975                                            ":refs/remotes/$ref_id\$") or
1976                     die "Failed to read \"svn-remote.$repo_id.fetch\" ",
1977                         "\":refs/remotes/$ref_id\$\" in config\n";
1978                ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
1979        }
1980        $self->{url} = command_oneline('config', '--get',
1981                                       "svn-remote.$repo_id.url") or
1982                  die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
1983        $self->rebuild;
1984        $self;
1985}
1986
1987sub refname {
1988        my ($refname) = "refs/remotes/$_[0]->{ref_id}" ;
1989
1990        # It cannot end with a slash /, we'll throw up on this because
1991        # SVN can't have directories with a slash in their name, either:
1992        if ($refname =~ m{/$}) {
1993                die "ref: '$refname' ends with a trailing slash, this is ",
1994                    "not permitted by git nor Subversion\n";
1995        }
1996
1997        # It cannot have ASCII control character space, tilde ~, caret ^,
1998        # colon :, question-mark ?, asterisk *, space, or open bracket [
1999        # anywhere.
2000        #
2001        # Additionally, % must be escaped because it is used for escaping
2002        # and we want our escaped refname to be reversible
2003        $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
2004
2005        # no slash-separated component can begin with a dot .
2006        # /.* becomes /%2E*
2007        $refname =~ s{/\.}{/%2E}g;
2008
2009        # It cannot have two consecutive dots .. anywhere
2010        # .. becomes %2E%2E
2011        $refname =~ s{\.\.}{%2E%2E}g;
2012
2013        return $refname;
2014}
2015
2016sub desanitize_refname {
2017        my ($refname) = @_;
2018        $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
2019        return $refname;
2020}
2021
2022sub svm_uuid {
2023        my ($self) = @_;
2024        return $self->{svm}->{uuid} if $self->svm;
2025        $self->ra;
2026        unless ($self->{svm}) {
2027                die "SVM UUID not cached, and reading remotely failed\n";
2028        }
2029        $self->{svm}->{uuid};
2030}
2031
2032sub svm {
2033        my ($self) = @_;
2034        return $self->{svm} if $self->{svm};
2035        my $svm;
2036        # see if we have it in our config, first:
2037        eval {
2038                my $section = "svn-remote.$self->{repo_id}";
2039                $svm = {
2040                  source => tmp_config('--get', "$section.svm-source"),
2041                  uuid => tmp_config('--get', "$section.svm-uuid"),
2042                  replace => tmp_config('--get', "$section.svm-replace"),
2043                }
2044        };
2045        if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
2046                $self->{svm} = $svm;
2047        }
2048        $self->{svm};
2049}
2050
2051sub _set_svm_vars {
2052        my ($self, $ra) = @_;
2053        return $ra if $self->svm;
2054
2055        my @err = ( "useSvmProps set, but failed to read SVM properties\n",
2056                    "(svm:source, svm:uuid) ",
2057                    "from the following URLs:\n" );
2058        sub read_svm_props {
2059                my ($self, $ra, $path, $r) = @_;
2060                my $props = ($ra->get_dir($path, $r))[2];
2061                my $src = $props->{'svm:source'};
2062                my $uuid = $props->{'svm:uuid'};
2063                return undef if (!$src || !$uuid);
2064
2065                chomp($src, $uuid);
2066
2067                $uuid =~ m{^[0-9a-f\-]{30,}$}i
2068                    or die "doesn't look right - svm:uuid is '$uuid'\n";
2069
2070                # the '!' is used to mark the repos_root!/relative/path
2071                $src =~ s{/?!/?}{/};
2072                $src =~ s{/+$}{}; # no trailing slashes please
2073                # username is of no interest
2074                $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
2075
2076                my $replace = $ra->{url};
2077                $replace .= "/$path" if length $path;
2078
2079                my $section = "svn-remote.$self->{repo_id}";
2080                tmp_config("$section.svm-source", $src);
2081                tmp_config("$section.svm-replace", $replace);
2082                tmp_config("$section.svm-uuid", $uuid);
2083                $self->{svm} = {
2084                        source => $src,
2085                        uuid => $uuid,
2086                        replace => $replace
2087                };
2088        }
2089
2090        my $r = $ra->get_latest_revnum;
2091        my $path = $self->{path};
2092        my %tried;
2093        while (length $path) {
2094                unless ($tried{"$self->{url}/$path"}) {
2095                        return $ra if $self->read_svm_props($ra, $path, $r);
2096                        $tried{"$self->{url}/$path"} = 1;
2097                }
2098                $path =~ s#/?[^/]+$##;
2099        }
2100        die "Path: '$path' should be ''\n" if $path ne '';
2101        return $ra if $self->read_svm_props($ra, $path, $r);
2102        $tried{"$self->{url}/$path"} = 1;
2103
2104        if ($ra->{repos_root} eq $self->{url}) {
2105                die @err, (map { "  $_\n" } keys %tried), "\n";
2106        }
2107
2108        # nope, make sure we're connected to the repository root:
2109        my $ok;
2110        my @tried_b;
2111        $path = $ra->{svn_path};
2112        $ra = Git::SVN::Ra->new($ra->{repos_root});
2113        while (length $path) {
2114                unless ($tried{"$ra->{url}/$path"}) {
2115                        $ok = $self->read_svm_props($ra, $path, $r);
2116                        last if $ok;
2117                        $tried{"$ra->{url}/$path"} = 1;
2118                }
2119                $path =~ s#/?[^/]+$##;
2120        }
2121        die "Path: '$path' should be ''\n" if $path ne '';
2122        $ok ||= $self->read_svm_props($ra, $path, $r);
2123        $tried{"$ra->{url}/$path"} = 1;
2124        if (!$ok) {
2125                die @err, (map { "  $_\n" } keys %tried), "\n";
2126        }
2127        Git::SVN::Ra->new($self->{url});
2128}
2129
2130sub svnsync {
2131        my ($self) = @_;
2132        return $self->{svnsync} if $self->{svnsync};
2133
2134        if ($self->no_metadata) {
2135                die "Can't have both 'noMetadata' and ",
2136                    "'useSvnsyncProps' options set!\n";
2137        }
2138        if ($self->rewrite_root) {
2139                die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
2140                    "options set!\n";
2141        }
2142
2143        my $svnsync;
2144        # see if we have it in our config, first:
2145        eval {
2146                my $section = "svn-remote.$self->{repo_id}";
2147
2148                my $url = tmp_config('--get', "$section.svnsync-url");
2149                ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2150                   die "doesn't look right - svn:sync-from-url is '$url'\n";
2151
2152                my $uuid = tmp_config('--get', "$section.svnsync-uuid");
2153                ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2154                   die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2155
2156                $svnsync = { url => $url, uuid => $uuid }
2157        };
2158        if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
2159                return $self->{svnsync} = $svnsync;
2160        }
2161
2162        my $err = "useSvnsyncProps set, but failed to read " .
2163                  "svnsync property: svn:sync-from-";
2164        my $rp = $self->ra->rev_proplist(0);
2165
2166        my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
2167        ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2168                   die "doesn't look right - svn:sync-from-url is '$url'\n";
2169
2170        my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
2171        ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2172                   die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2173
2174        my $section = "svn-remote.$self->{repo_id}";
2175        tmp_config('--add', "$section.svnsync-uuid", $uuid);
2176        tmp_config('--add', "$section.svnsync-url", $url);
2177        return $self->{svnsync} = { url => $url, uuid => $uuid };
2178}
2179
2180# this allows us to memoize our SVN::Ra UUID locally and avoid a
2181# remote lookup (useful for 'git svn log').
2182sub ra_uuid {
2183        my ($self) = @_;
2184        unless ($self->{ra_uuid}) {
2185                my $key = "svn-remote.$self->{repo_id}.uuid";
2186                my $uuid = eval { tmp_config('--get', $key) };
2187                if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/i) {
2188                        $self->{ra_uuid} = $uuid;
2189                } else {
2190                        die "ra_uuid called without URL\n" unless $self->{url};
2191                        $self->{ra_uuid} = $self->ra->get_uuid;
2192                        tmp_config('--add', $key, $self->{ra_uuid});
2193                }
2194        }
2195        $self->{ra_uuid};
2196}
2197
2198sub _set_repos_root {
2199        my ($self, $repos_root) = @_;
2200        my $k = "svn-remote.$self->{repo_id}.reposRoot";
2201        $repos_root ||= $self->ra->{repos_root};
2202        tmp_config($k, $repos_root);
2203        $repos_root;
2204}
2205
2206sub repos_root {
2207        my ($self) = @_;
2208        my $k = "svn-remote.$self->{repo_id}.reposRoot";
2209        eval { tmp_config('--get', $k) } || $self->_set_repos_root;
2210}
2211
2212sub ra {
2213        my ($self) = shift;
2214        my $ra = Git::SVN::Ra->new($self->{url});
2215        $self->_set_repos_root($ra->{repos_root});
2216        if ($self->use_svm_props && !$self->{svm}) {
2217                if ($self->no_metadata) {
2218                        die "Can't have both 'noMetadata' and ",
2219                            "'useSvmProps' options set!\n";
2220                } elsif ($self->use_svnsync_props) {
2221                        die "Can't have both 'useSvnsyncProps' and ",
2222                            "'useSvmProps' options set!\n";
2223                }
2224                $ra = $self->_set_svm_vars($ra);
2225                $self->{-want_revprops} = 1;
2226        }
2227        $ra;
2228}
2229
2230# prop_walk(PATH, REV, SUB)
2231# -------------------------
2232# Recursively traverse PATH at revision REV and invoke SUB for each
2233# directory that contains a SVN property.  SUB will be invoked as
2234# follows:  &SUB(gs, path, props);  where `gs' is this instance of
2235# Git::SVN, `path' the path to the directory where the properties
2236# `props' were found.  The `path' will be relative to point of checkout,
2237# that is, if url://repo/trunk is the current Git branch, and that
2238# directory contains a sub-directory `d', SUB will be invoked with `/d/'
2239# as `path' (note the trailing `/').
2240sub prop_walk {
2241        my ($self, $path, $rev, $sub) = @_;
2242
2243        $path =~ s#^/##;
2244        my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev);
2245        $path =~ s#^/*#/#g;
2246        my $p = $path;
2247        # Strip the irrelevant part of the path.
2248        $p =~ s#^/+\Q$self->{path}\E(/|$)#/#;
2249        # Ensure the path is terminated by a `/'.
2250        $p =~ s#/*$#/#;
2251
2252        # The properties contain all the internal SVN stuff nobody
2253        # (usually) cares about.
2254        my $interesting_props = 0;
2255        foreach (keys %{$props}) {
2256                # If it doesn't start with `svn:', it must be a
2257                # user-defined property.
2258                ++$interesting_props and next if $_ !~ /^svn:/;
2259                # FIXME: Fragile, if SVN adds new public properties,
2260                # this needs to be updated.
2261                ++$interesting_props if /^svn:(?:ignore|keywords|executable
2262                                                 |eol-style|mime-type
2263                                                 |externals|needs-lock)$/x;
2264        }
2265        &$sub($self, $p, $props) if $interesting_props;
2266
2267        foreach (sort keys %$dirent) {
2268                next if $dirent->{$_}->{kind} != $SVN::Node::dir;
2269                $self->prop_walk($self->{path} . $p . $_, $rev, $sub);
2270        }
2271}
2272
2273sub last_rev { ($_[0]->last_rev_commit)[0] }
2274sub last_commit { ($_[0]->last_rev_commit)[1] }
2275
2276# returns the newest SVN revision number and newest commit SHA1
2277sub last_rev_commit {
2278        my ($self) = @_;
2279        if (defined $self->{last_rev} && defined $self->{last_commit}) {
2280                return ($self->{last_rev}, $self->{last_commit});
2281        }
2282        my $c = ::verify_ref($self->refname.'^0');
2283        if ($c && !$self->use_svm_props && !$self->no_metadata) {
2284                my $rev = (::cmt_metadata($c))[1];
2285                if (defined $rev) {
2286                        ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
2287                        return ($rev, $c);
2288                }
2289        }
2290        my $map_path = $self->map_path;
2291        unless (-e $map_path) {
2292                ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
2293                return (undef, undef);
2294        }
2295        my ($rev, $commit) = $self->rev_map_max(1);
2296        ($self->{last_rev}, $self->{last_commit}) = ($rev, $commit);
2297        return ($rev, $commit);
2298}
2299
2300sub get_fetch_range {
2301        my ($self, $min, $max) = @_;
2302        $max ||= $self->ra->get_latest_revnum;
2303        $min ||= $self->rev_map_max;
2304        (++$min, $max);
2305}
2306
2307sub tmp_config {
2308        my (@args) = @_;
2309        my $old_def_config = "$ENV{GIT_DIR}/svn/config";
2310        my $config = "$ENV{GIT_DIR}/svn/.metadata";
2311        if (! -f $config && -f $old_def_config) {
2312                rename $old_def_config, $config or
2313                       die "Failed rename $old_def_config => $config: $!\n";
2314        }
2315        my $old_config = $ENV{GIT_CONFIG};
2316        $ENV{GIT_CONFIG} = $config;
2317        $@ = undef;
2318        my @ret = eval {
2319                unless (-f $config) {
2320                        mkfile($config);
2321                        open my $fh, '>', $config or
2322                            die "Can't open $config: $!\n";
2323                        print $fh "; This file is used internally by ",
2324                                  "git-svn\n" or die
2325                                  "Couldn't write to $config: $!\n";
2326                        print $fh "; You should not have to edit it\n" or
2327                              die "Couldn't write to $config: $!\n";
2328                        close $fh or die "Couldn't close $config: $!\n";
2329                }
2330                command('config', @args);
2331        };
2332        my $err = $@;
2333        if (defined $old_config) {
2334                $ENV{GIT_CONFIG} = $old_config;
2335        } else {
2336                delete $ENV{GIT_CONFIG};
2337        }
2338        die $err if $err;
2339        wantarray ? @ret : $ret[0];
2340}
2341
2342sub tmp_index_do {
2343        my ($self, $sub) = @_;
2344        my $old_index = $ENV{GIT_INDEX_FILE};
2345        $ENV{GIT_INDEX_FILE} = $self->{index};
2346        $@ = undef;
2347        my @ret = eval {
2348                my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
2349                mkpath([$dir]) unless -d $dir;
2350                &$sub;
2351        };
2352        my $err = $@;
2353        if (defined $old_index) {
2354                $ENV{GIT_INDEX_FILE} = $old_index;
2355        } else {
2356                delete $ENV{GIT_INDEX_FILE};
2357        }
2358        die $err if $err;
2359        wantarray ? @ret : $ret[0];
2360}
2361
2362sub assert_index_clean {
2363        my ($self, $treeish) = @_;
2364
2365        $self->tmp_index_do(sub {
2366                command_noisy('read-tree', $treeish) unless -e $self->{index};
2367                my $x = command_oneline('write-tree');
2368                my ($y) = (command(qw/cat-file commit/, $treeish) =~
2369                           /^tree ($::sha1)/mo);
2370                return if $y eq $x;
2371
2372                warn "Index mismatch: $y != $x\nrereading $treeish\n";
2373                unlink $self->{index} or die "unlink $self->{index}: $!\n";
2374                command_noisy('read-tree', $treeish);
2375                $x = command_oneline('write-tree');
2376                if ($y ne $x) {
2377                        ::fatal "trees ($treeish) $y != $x\n",
2378                                "Something is seriously wrong...";
2379                }
2380        });
2381}
2382
2383sub get_commit_parents {
2384        my ($self, $log_entry) = @_;
2385        my (%seen, @ret, @tmp);
2386        # legacy support for 'set-tree'; this is only used by set_tree_cb:
2387        if (my $ip = $self->{inject_parents}) {
2388                if (my $commit = delete $ip->{$log_entry->{revision}}) {
2389                        push @tmp, $commit;
2390                }
2391        }
2392        if (my $cur = ::verify_ref($self->refname.'^0')) {
2393                push @tmp, $cur;
2394        }
2395        if (my $ipd = $self->{inject_parents_dcommit}) {
2396                if (my $commit = delete $ipd->{$log_entry->{revision}}) {
2397                        push @tmp, @$commit;
2398                }
2399        }
2400        push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
2401        while (my $p = shift @tmp) {
2402                next if $seen{$p};
2403                $seen{$p} = 1;
2404                push @ret, $p;
2405                # MAXPARENT is defined to 16 in commit-tree.c:
2406                last if @ret >= 16;
2407        }
2408        if (@tmp) {
2409                die "r$log_entry->{revision}: No room for parents:\n\t",
2410                    join("\n\t", @tmp), "\n";
2411        }
2412        @ret;
2413}
2414
2415sub rewrite_root {
2416        my ($self) = @_;
2417        return $self->{-rewrite_root} if exists $self->{-rewrite_root};
2418        my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
2419        my $rwr = eval { command_oneline(qw/config --get/, $k) };
2420        if ($rwr) {
2421                $rwr =~ s#/+$##;
2422                if ($rwr !~ m#^[a-z\+]+://#) {
2423                        die "$rwr is not a valid URL (key: $k)\n";
2424                }
2425        }
2426        $self->{-rewrite_root} = $rwr;
2427}
2428
2429sub metadata_url {
2430        my ($self) = @_;
2431        ($self->rewrite_root || $self->{url}) .
2432           (length $self->{path} ? '/' . $self->{path} : '');
2433}
2434
2435sub full_url {
2436        my ($self) = @_;
2437        $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
2438}
2439
2440
2441sub set_commit_header_env {
2442        my ($log_entry) = @_;
2443        my %env;
2444        foreach my $ned (qw/NAME EMAIL DATE/) {
2445                foreach my $ac (qw/AUTHOR COMMITTER/) {
2446                        $env{"GIT_${ac}_${ned}"} = $ENV{"GIT_${ac}_${ned}"};
2447                }
2448        }
2449
2450        $ENV{GIT_AUTHOR_NAME} = $log_entry->{name};
2451        $ENV{GIT_AUTHOR_EMAIL} = $log_entry->{email};
2452        $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
2453
2454        $ENV{GIT_COMMITTER_NAME} = (defined $log_entry->{commit_name})
2455                                                ? $log_entry->{commit_name}
2456                                                : $log_entry->{name};
2457        $ENV{GIT_COMMITTER_EMAIL} = (defined $log_entry->{commit_email})
2458                                                ? $log_entry->{commit_email}
2459                                                : $log_entry->{email};
2460        \%env;
2461}
2462
2463sub restore_commit_header_env {
2464        my ($env) = @_;
2465        foreach my $ned (qw/NAME EMAIL DATE/) {
2466                foreach my $ac (qw/AUTHOR COMMITTER/) {
2467                        my $k = "GIT_${ac}_${ned}";
2468                        if (defined $env->{$k}) {
2469                                $ENV{$k} = $env->{$k};
2470                        } else {
2471                                delete $ENV{$k};
2472                        }
2473                }
2474        }
2475}
2476
2477sub gc {
2478        command_noisy('gc', '--auto');
2479};
2480
2481sub do_git_commit {
2482        my ($self, $log_entry) = @_;
2483        my $lr = $self->last_rev;
2484        if (defined $lr && $lr >= $log_entry->{revision}) {
2485                die "Last fetched revision of ", $self->refname,
2486                    " was r$lr, but we are about to fetch: ",
2487                    "r$log_entry->{revision}!\n";
2488        }
2489        if (my $c = $self->rev_map_get($log_entry->{revision})) {
2490                croak "$log_entry->{revision} = $c already exists! ",
2491                      "Why are we refetching it?\n";
2492        }
2493        my $old_env = set_commit_header_env($log_entry);
2494        my $tree = $log_entry->{tree};
2495        if (!defined $tree) {
2496                $tree = $self->tmp_index_do(sub {
2497                                            command_oneline('write-tree') });
2498        }
2499        die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
2500
2501        my @exec = ('git', 'commit-tree', $tree);
2502        foreach ($self->get_commit_parents($log_entry)) {
2503                push @exec, '-p', $_;
2504        }
2505        defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
2506                                                                   or croak $!;
2507        binmode $msg_fh;
2508
2509        # we always get UTF-8 from SVN, but we may want our commits in
2510        # a different encoding.
2511        if (my $enc = Git::config('i18n.commitencoding')) {
2512                require Encode;
2513                Encode::from_to($log_entry->{log}, 'UTF-8', $enc);
2514        }
2515        print $msg_fh $log_entry->{log} or croak $!;
2516        restore_commit_header_env($old_env);
2517        unless ($self->no_metadata) {
2518                print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
2519                              or croak $!;
2520        }
2521        $msg_fh->flush == 0 or croak $!;
2522        close $msg_fh or croak $!;
2523        chomp(my $commit = do { local $/; <$out_fh> });
2524        close $out_fh or croak $!;
2525        waitpid $pid, 0;
2526        croak $? if $?;
2527        if ($commit !~ /^$::sha1$/o) {
2528                die "Failed to commit, invalid sha1: $commit\n";
2529        }
2530
2531        $self->rev_map_set($log_entry->{revision}, $commit, 1);
2532
2533        $self->{last_rev} = $log_entry->{revision};
2534        $self->{last_commit} = $commit;
2535        print "r$log_entry->{revision}" unless $::_q > 1;
2536        if (defined $log_entry->{svm_revision}) {
2537                 print " (\@$log_entry->{svm_revision})" unless $::_q > 1;
2538                 $self->rev_map_set($log_entry->{svm_revision}, $commit,
2539                                   0, $self->svm_uuid);
2540        }
2541        print " = $commit ($self->{ref_id})\n" unless $::_q > 1;
2542        if (--$_gc_nr == 0) {
2543                $_gc_nr = $_gc_period;
2544                gc();
2545        }
2546        return $commit;
2547}
2548
2549sub match_paths {
2550        my ($self, $paths, $r) = @_;
2551        return 1 if $self->{path} eq '';
2552        if (my $path = $paths->{"/$self->{path}"}) {
2553                return ($path->{action} eq 'D') ? 0 : 1;
2554        }
2555        $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
2556        if (grep /$self->{path_regex}/, keys %$paths) {
2557                return 1;
2558        }
2559        my $c = '';
2560        foreach (split m#/#, $self->{path}) {
2561                $c .= "/$_";
2562                next unless ($paths->{$c} &&
2563                             ($paths->{$c}->{action} =~ /^[AR]$/));
2564                if ($self->ra->check_path($self->{path}, $r) ==
2565                    $SVN::Node::dir) {
2566                        return 1;
2567                }
2568        }
2569        return 0;
2570}
2571
2572sub find_parent_branch {
2573        my ($self, $paths, $rev) = @_;
2574        return undef unless $self->follow_parent;
2575        unless (defined $paths) {
2576                my $err_handler = $SVN::Error::handler;
2577                $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
2578                $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1,
2579                                   sub { $paths = $_[0] });
2580                $SVN::Error::handler = $err_handler;
2581        }
2582        return undef unless defined $paths;
2583
2584        # look for a parent from another branch:
2585        my @b_path_components = split m#/#, $self->{path};
2586        my @a_path_components;
2587        my $i;
2588        while (@b_path_components) {
2589                $i = $paths->{'/'.join('/', @b_path_components)};
2590                last if $i && defined $i->{copyfrom_path};
2591                unshift(@a_path_components, pop(@b_path_components));
2592        }
2593        return undef unless defined $i && defined $i->{copyfrom_path};
2594        my $branch_from = $i->{copyfrom_path};
2595        if (@a_path_components) {
2596                print STDERR "branch_from: $branch_from => ";
2597                $branch_from .= '/'.join('/', @a_path_components);
2598                print STDERR $branch_from, "\n";
2599        }
2600        my $r = $i->{copyfrom_rev};
2601        my $repos_root = $self->ra->{repos_root};
2602        my $url = $self->ra->{url};
2603        my $new_url = $url . $branch_from;
2604        print STDERR  "Found possible branch point: ",
2605                      "$new_url => ", $self->full_url, ", $r\n";
2606        $branch_from =~ s#^/##;
2607        my $gs = $self->other_gs($new_url, $url,
2608                                 $branch_from, $r, $self->{ref_id});
2609        my ($r0, $parent) = $gs->find_rev_before($r, 1);
2610        {
2611                my ($base, $head);
2612                if (!defined $r0 || !defined $parent) {
2613                        ($base, $head) = parse_revision_argument(0, $r);
2614                } else {
2615                        if ($r0 < $r) {
2616                                $gs->ra->get_log([$gs->{path}], $r0 + 1, $r, 1,
2617                                        0, 1, sub { $base = $_[1] - 1 });
2618                        }
2619                }
2620                if (defined $base && $base <= $r) {
2621                        $gs->fetch($base, $r);
2622                }
2623                ($r0, $parent) = $gs->find_rev_before($r, 1);
2624        }
2625        if (defined $r0 && defined $parent) {
2626                print STDERR "Found branch parent: ($self->{ref_id}) $parent\n";
2627                my $ed;
2628                if ($self->ra->can_do_switch) {
2629                        $self->assert_index_clean($parent);
2630                        print STDERR "Following parent with do_switch\n";
2631                        # do_switch works with svn/trunk >= r22312, but that
2632                        # is not included with SVN 1.4.3 (the latest version
2633                        # at the moment), so we can't rely on it
2634                        $self->{last_rev} = $r0;
2635                        $self->{last_commit} = $parent;
2636                        $ed = SVN::Git::Fetcher->new($self, $gs->{path});
2637                        $gs->ra->gs_do_switch($r0, $rev, $gs,
2638                                              $self->full_url, $ed)
2639                          or die "SVN connection failed somewhere...\n";
2640                } elsif ($self->ra->trees_match($new_url, $r0,
2641                                                $self->full_url, $rev)) {
2642                        print STDERR "Trees match:\n",
2643                                     "  $new_url\@$r0\n",
2644                                     "  ${\$self->full_url}\@$rev\n",
2645                                     "Following parent with no changes\n";
2646                        $self->tmp_index_do(sub {
2647                            command_noisy('read-tree', $parent);
2648                        });
2649                        $self->{last_commit} = $parent;
2650                } else {
2651                        print STDERR "Following parent with do_update\n";
2652                        $ed = SVN::Git::Fetcher->new($self);
2653                        $self->ra->gs_do_update($rev, $rev, $self, $ed)
2654                          or die "SVN connection failed somewhere...\n";
2655                }
2656                print STDERR "Successfully followed parent\n";
2657                return $self->make_log_entry($rev, [$parent], $ed);
2658        }
2659        return undef;
2660}
2661
2662sub do_fetch {
2663        my ($self, $paths, $rev) = @_;
2664        my $ed;
2665        my ($last_rev, @parents);
2666        if (my $lc = $self->last_commit) {
2667                # we can have a branch that was deleted, then re-added
2668                # under the same name but copied from another path, in
2669                # which case we'll have multiple parents (we don't
2670                # want to break the original ref, nor lose copypath info):
2671                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2672                        push @{$log_entry->{parents}}, $lc;
2673                        return $log_entry;
2674                }
2675                $ed = SVN::Git::Fetcher->new($self);
2676                $last_rev = $self->{last_rev};
2677                $ed->{c} = $lc;
2678                @parents = ($lc);
2679        } else {
2680                $last_rev = $rev;
2681                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2682                        return $log_entry;
2683                }
2684                $ed = SVN::Git::Fetcher->new($self);
2685        }
2686        unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
2687                die "SVN connection failed somewhere...\n";
2688        }
2689        $self->make_log_entry($rev, \@parents, $ed);
2690}
2691
2692sub get_untracked {
2693        my ($self, $ed) = @_;
2694        my @out;
2695        my $h = $ed->{empty};
2696        foreach (sort keys %$h) {
2697                my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
2698                push @out, "  $act: " . uri_encode($_);
2699                warn "W: $act: $_\n";
2700        }
2701        foreach my $t (qw/dir_prop file_prop/) {
2702                $h = $ed->{$t} or next;
2703                foreach my $path (sort keys %$h) {
2704                        my $ppath = $path eq '' ? '.' : $path;
2705                        foreach my $prop (sort keys %{$h->{$path}}) {
2706                                next if $SKIP_PROP{$prop};
2707                                my $v = $h->{$path}->{$prop};
2708                                my $t_ppath_prop = "$t: " .
2709                                                    uri_encode($ppath) . ' ' .
2710                                                    uri_encode($prop);
2711                                if (defined $v) {
2712                                        push @out, "  +$t_ppath_prop " .
2713                                                   uri_encode($v);
2714                                } else {
2715                                        push @out, "  -$t_ppath_prop";
2716                                }
2717                        }
2718                }
2719        }
2720        foreach my $t (qw/absent_file absent_directory/) {
2721                $h = $ed->{$t} or next;
2722                foreach my $parent (sort keys %$h) {
2723                        foreach my $path (sort @{$h->{$parent}}) {
2724                                push @out, "  $t: " .
2725                                           uri_encode("$parent/$path");
2726                                warn "W: $t: $parent/$path ",
2727                                     "Insufficient permissions?\n";
2728                        }
2729                }
2730        }
2731        \@out;
2732}
2733
2734# parse_svn_date(DATE)
2735# --------------------
2736# Given a date (in UTC) from Subversion, return a string in the format
2737# "<TZ Offset> <local date/time>" that Git will use.
2738#
2739# By default the parsed date will be in UTC; if $Git::SVN::_localtime
2740# is true we'll convert it to the local timezone instead.
2741sub parse_svn_date {
2742        my $date = shift || return '+0000 1970-01-01 00:00:00';
2743        my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2744                                            (\d\d)\:(\d\d)\:(\d\d)\.\d*Z$/x) or
2745                                         croak "Unable to parse date: $date\n";
2746        my $parsed_date;    # Set next.
2747
2748        if ($Git::SVN::_localtime) {
2749                # Translate the Subversion datetime to an epoch time.
2750                # Begin by switching ourselves to $date's timezone, UTC.
2751                my $old_env_TZ = $ENV{TZ};
2752                $ENV{TZ} = 'UTC';
2753
2754                my $epoch_in_UTC =
2755                    POSIX::strftime('%s', $S, $M, $H, $d, $m - 1, $Y - 1900);
2756
2757                # Determine our local timezone (including DST) at the
2758                # time of $epoch_in_UTC.  $Git::SVN::Log::TZ stored the
2759                # value of TZ, if any, at the time we were run.
2760                if (defined $Git::SVN::Log::TZ) {
2761                        $ENV{TZ} = $Git::SVN::Log::TZ;
2762                } else {
2763                        delete $ENV{TZ};
2764                }
2765
2766                my $our_TZ =
2767                    POSIX::strftime('%Z', $S, $M, $H, $d, $m - 1, $Y - 1900);
2768
2769                # This converts $epoch_in_UTC into our local timezone.
2770                my ($sec, $min, $hour, $mday, $mon, $year,
2771                    $wday, $yday, $isdst) = localtime($epoch_in_UTC);
2772
2773                $parsed_date = sprintf('%s %04d-%02d-%02d %02d:%02d:%02d',
2774                                       $our_TZ, $year + 1900, $mon + 1,
2775                                       $mday, $hour, $min, $sec);
2776
2777                # Reset us to the timezone in effect when we entered
2778                # this routine.
2779                if (defined $old_env_TZ) {
2780                        $ENV{TZ} = $old_env_TZ;
2781                } else {
2782                        delete $ENV{TZ};
2783                }
2784        } else {
2785                $parsed_date = "+0000 $Y-$m-$d $H:$M:$S";
2786        }
2787
2788        return $parsed_date;
2789}
2790
2791sub other_gs {
2792        my ($self, $new_url, $url,
2793            $branch_from, $r, $old_ref_id) = @_;
2794        my $gs = Git::SVN->find_by_url($new_url, $url, $branch_from);
2795        unless ($gs) {
2796                my $ref_id = $old_ref_id;
2797                $ref_id =~ s/\@\d+$//;
2798                $ref_id .= "\@$r";
2799                # just grow a tail if we're not unique enough :x
2800                $ref_id .= '-' while find_ref($ref_id);
2801                print STDERR "Initializing parent: $ref_id\n";
2802                my ($u, $p, $repo_id) = ($new_url, '', $ref_id);
2803                if ($u =~ s#^\Q$url\E(/|$)##) {
2804                        $p = $u;
2805                        $u = $url;
2806                        $repo_id = $self->{repo_id};
2807                }
2808                $gs = Git::SVN->init($u, $p, $repo_id, $ref_id, 1);
2809        }
2810        $gs
2811}
2812
2813sub call_authors_prog {
2814        my ($orig_author) = @_;
2815        my $author = `$::_authors_prog $orig_author`;
2816        if ($? != 0) {
2817                die "$::_authors_prog failed with exit code $?\n"
2818        }
2819        if ($author =~ /^\s*(.+?)\s*<(.*)>\s*$/) {
2820                my ($name, $email) = ($1, $2);
2821                $email = undef if length $2 == 0;
2822                return [$name, $email];
2823        } else {
2824                die "Author: $orig_author: $::_authors_prog returned "
2825                        . "invalid author format: $author\n";
2826        }
2827}
2828
2829sub check_author {
2830        my ($author) = @_;
2831        if (!defined $author || length $author == 0) {
2832                $author = '(no author)';
2833        }
2834        if (!defined $::users{$author}) {
2835                if (defined $::_authors_prog) {
2836                        $::users{$author} = call_authors_prog($author);
2837                } elsif (defined $::_authors) {
2838                        die "Author: $author not defined in $::_authors file\n";
2839                }
2840        }
2841        $author;
2842}
2843
2844sub make_log_entry {
2845        my ($self, $rev, $parents, $ed) = @_;
2846        my $untracked = $self->get_untracked($ed);
2847
2848        open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
2849        print $un "r$rev\n" or croak $!;
2850        print $un $_, "\n" foreach @$untracked;
2851        my %log_entry = ( parents => $parents || [], revision => $rev,
2852                          log => '');
2853
2854        my $headrev;
2855        my $logged = delete $self->{logged_rev_props};
2856        if (!$logged || $self->{-want_revprops}) {
2857                my $rp = $self->ra->rev_proplist($rev);
2858                foreach (sort keys %$rp) {
2859                        my $v = $rp->{$_};
2860                        if (/^svn:(author|date|log)$/) {
2861                                $log_entry{$1} = $v;
2862                        } elsif ($_ eq 'svm:headrev') {
2863                                $headrev = $v;
2864                        } else {
2865                                print $un "  rev_prop: ", uri_encode($_), ' ',
2866                                          uri_encode($v), "\n";
2867                        }
2868                }
2869        } else {
2870                map { $log_entry{$_} = $logged->{$_} } keys %$logged;
2871        }
2872        close $un or croak $!;
2873
2874        $log_entry{date} = parse_svn_date($log_entry{date});
2875        $log_entry{log} .= "\n";
2876        my $author = $log_entry{author} = check_author($log_entry{author});
2877        my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
2878                                                       : ($author, undef);
2879
2880        my ($commit_name, $commit_email) = ($name, $email);
2881        if ($_use_log_author) {
2882                my $name_field;
2883                if ($log_entry{log} =~ /From:\s+(.*\S)\s*\n/i) {
2884                        $name_field = $1;
2885                } elsif ($log_entry{log} =~ /Signed-off-by:\s+(.*\S)\s*\n/i) {
2886                        $name_field = $1;
2887                }
2888                if (!defined $name_field) {
2889                        if (!defined $email) {
2890                                $email = $name;
2891                        }
2892                } elsif ($name_field =~ /(.*?)\s+<(.*)>/) {
2893                        ($name, $email) = ($1, $2);
2894                } elsif ($name_field =~ /(.*)@/) {
2895                        ($name, $email) = ($1, $name_field);
2896                } else {
2897                        ($name, $email) = ($name_field, $name_field);
2898                }
2899        }
2900        if (defined $headrev && $self->use_svm_props) {
2901                if ($self->rewrite_root) {
2902                        die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
2903                            "options set!\n";
2904                }
2905                my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$}i;
2906                # we don't want "SVM: initializing mirror for junk" ...
2907                return undef if $r == 0;
2908                my $svm = $self->svm;
2909                if ($uuid ne $svm->{uuid}) {
2910                        die "UUID mismatch on SVM path:\n",
2911                            "expected: $svm->{uuid}\n",
2912                            "     got: $uuid\n";
2913                }
2914                my $full_url = $self->full_url;
2915                $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
2916                             die "Failed to replace '$svm->{replace}' with ",
2917                                 "'$svm->{source}' in $full_url\n";
2918                # throw away username for storing in records
2919                remove_username($full_url);
2920                $log_entry{metadata} = "$full_url\@$r $uuid";
2921                $log_entry{svm_revision} = $r;
2922                $email ||= "$author\@$uuid";
2923                $commit_email ||= "$author\@$uuid";
2924        } elsif ($self->use_svnsync_props) {
2925                my $full_url = $self->svnsync->{url};
2926                $full_url .= "/$self->{path}" if length $self->{path};
2927                remove_username($full_url);
2928                my $uuid = $self->svnsync->{uuid};
2929                $log_entry{metadata} = "$full_url\@$rev $uuid";
2930                $email ||= "$author\@$uuid";
2931                $commit_email ||= "$author\@$uuid";
2932        } else {
2933                my $url = $self->metadata_url;
2934                remove_username($url);
2935                $log_entry{metadata} = "$url\@$rev " .
2936                                       $self->ra->get_uuid;
2937                $email ||= "$author\@" . $self->ra->get_uuid;
2938                $commit_email ||= "$author\@" . $self->ra->get_uuid;
2939        }
2940        $log_entry{name} = $name;
2941        $log_entry{email} = $email;
2942        $log_entry{commit_name} = $commit_name;
2943        $log_entry{commit_email} = $commit_email;
2944        \%log_entry;
2945}
2946
2947sub fetch {
2948        my ($self, $min_rev, $max_rev, @parents) = @_;
2949        my ($last_rev, $last_commit) = $self->last_rev_commit;
2950        my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
2951        $self->ra->gs_fetch_loop_common($base, $head, [$self]);
2952}
2953
2954sub set_tree_cb {
2955        my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
2956        $self->{inject_parents} = { $rev => $tree };
2957        $self->fetch(undef, undef);
2958}
2959
2960sub set_tree {
2961        my ($self, $tree) = (shift, shift);
2962        my $log_entry = ::get_commit_entry($tree);
2963        unless ($self->{last_rev}) {
2964                ::fatal("Must have an existing revision to commit");
2965        }
2966        my %ed_opts = ( r => $self->{last_rev},
2967                        log => $log_entry->{log},
2968                        ra => $self->ra,
2969                        tree_a => $self->{last_commit},
2970                        tree_b => $tree,
2971                        editor_cb => sub {
2972                               $self->set_tree_cb($log_entry, $tree, @_) },
2973                        svn_path => $self->{path} );
2974        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
2975                print "No changes\nr$self->{last_rev} = $tree\n";
2976        }
2977}
2978
2979sub rebuild_from_rev_db {
2980        my ($self, $path) = @_;
2981        my $r = -1;
2982        open my $fh, '<', $path or croak "open: $!";
2983        binmode $fh or croak "binmode: $!";
2984        while (<$fh>) {
2985                length($_) == 41 or croak "inconsistent size in ($_) != 41";
2986                chomp($_);
2987                ++$r;
2988                next if $_ eq ('0' x 40);
2989                $self->rev_map_set($r, $_);
2990                print "r$r = $_\n";
2991        }
2992        close $fh or croak "close: $!";
2993        unlink $path or croak "unlink: $!";
2994}
2995
2996sub rebuild {
2997        my ($self) = @_;
2998        my $map_path = $self->map_path;
2999        my $partial = (-e $map_path && ! -z $map_path);
3000        return unless ::verify_ref($self->refname.'^0');
3001        if (!$partial && ($self->use_svm_props || $self->no_metadata)) {
3002                my $rev_db = $self->rev_db_path;
3003                $self->rebuild_from_rev_db($rev_db);
3004                if ($self->use_svm_props) {
3005                        my $svm_rev_db = $self->rev_db_path($self->svm_uuid);
3006                        $self->rebuild_from_rev_db($svm_rev_db);
3007                }
3008                $self->unlink_rev_db_symlink;
3009                return;
3010        }
3011        print "Rebuilding $map_path ...\n" if (!$partial);
3012        my ($base_rev, $head) = ($partial ? $self->rev_map_max_norebuild(1) :
3013                (undef, undef));
3014        my ($log, $ctx) =
3015            command_output_pipe(qw/rev-list --pretty=raw --no-color --reverse/,
3016                                ($head ? "$head.." : "") . $self->refname,
3017                                '--');
3018        my $metadata_url = $self->metadata_url;
3019        remove_username($metadata_url);
3020        my $svn_uuid = $self->ra_uuid;
3021        my $c;
3022        while (<$log>) {
3023                if ( m{^commit ($::sha1)$} ) {
3024                        $c = $1;
3025                        next;
3026                }
3027                next unless s{^\s*(git-svn-id:)}{$1};
3028                my ($url, $rev, $uuid) = ::extract_metadata($_);
3029                remove_username($url);
3030
3031                # ignore merges (from set-tree)
3032                next if (!defined $rev || !$uuid);
3033
3034                # if we merged or otherwise started elsewhere, this is
3035                # how we break out of it
3036                if (($uuid ne $svn_uuid) ||
3037                    ($metadata_url && $url && ($url ne $metadata_url))) {
3038                        next;
3039                }
3040                if ($partial && $head) {
3041                        print "Partial-rebuilding $map_path ...\n";
3042                        print "Currently at $base_rev = $head\n";
3043                        $head = undef;
3044                }
3045
3046                $self->rev_map_set($rev, $c);
3047                print "r$rev = $c\n";
3048        }
3049        command_close_pipe($log, $ctx);
3050        print "Done rebuilding $map_path\n" if (!$partial || !$head);
3051        my $rev_db_path = $self->rev_db_path;
3052        if (-f $self->rev_db_path) {
3053                unlink $self->rev_db_path or croak "unlink: $!";
3054        }
3055        $self->unlink_rev_db_symlink;
3056}
3057
3058# rev_map:
3059# Tie::File seems to be prone to offset errors if revisions get sparse,
3060# it's not that fast, either.  Tie::File is also not in Perl 5.6.  So
3061# one of my favorite modules is out :<  Next up would be one of the DBM
3062# modules, but I'm not sure which is most portable...
3063#
3064# This is the replacement for the rev_db format, which was too big
3065# and inefficient for large repositories with a lot of sparse history
3066# (mainly tags)
3067#
3068# The format is this:
3069#   - 24 bytes for every record,
3070#     * 4 bytes for the integer representing an SVN revision number
3071#     * 20 bytes representing the sha1 of a git commit
3072#   - No empty padding records like the old format
3073#     (except the last record, which can be overwritten)
3074#   - new records are written append-only since SVN revision numbers
3075#     increase monotonically
3076#   - lookups on SVN revision number are done via a binary search
3077#   - Piping the file to xxd -c24 is a good way of dumping it for
3078#     viewing or editing (piped back through xxd -r), should the need
3079#     ever arise.
3080#   - The last record can be padding revision with an all-zero sha1
3081#     This is used to optimize fetch performance when using multiple
3082#     "fetch" directives in .git/config
3083#
3084# These files are disposable unless noMetadata or useSvmProps is set
3085
3086sub _rev_map_set {
3087        my ($fh, $rev, $commit) = @_;
3088
3089        binmode $fh or croak "binmode: $!";
3090        my $size = (stat($fh))[7];
3091        ($size % 24) == 0 or croak "inconsistent size: $size";
3092
3093        my $wr_offset = 0;
3094        if ($size > 0) {
3095                sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3096                my $read = sysread($fh, my $buf, 24) or croak "read: $!";
3097                $read == 24 or croak "read only $read bytes (!= 24)";
3098                my ($last_rev, $last_commit) = unpack(rev_map_fmt, $buf);
3099                if ($last_commit eq ('0' x40)) {
3100                        if ($size >= 48) {
3101                                sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3102                                $read = sysread($fh, $buf, 24) or
3103                                    croak "read: $!";
3104                                $read == 24 or
3105                                    croak "read only $read bytes (!= 24)";
3106                                ($last_rev, $last_commit) =
3107                                    unpack(rev_map_fmt, $buf);
3108                                if ($last_commit eq ('0' x40)) {
3109                                        croak "inconsistent .rev_map\n";
3110                                }
3111                        }
3112                        if ($last_rev >= $rev) {
3113                                croak "last_rev is higher!: $last_rev >= $rev";
3114                        }
3115                        $wr_offset = -24;
3116                }
3117        }
3118        sysseek($fh, $wr_offset, SEEK_END) or croak "seek: $!";
3119        syswrite($fh, pack(rev_map_fmt, $rev, $commit), 24) == 24 or
3120          croak "write: $!";
3121}
3122
3123sub _rev_map_reset {
3124        my ($fh, $rev, $commit) = @_;
3125        my $c = _rev_map_get($fh, $rev);
3126        $c eq $commit or die "_rev_map_reset(@_) commit $c does not match!\n";
3127        my $offset = sysseek($fh, 0, SEEK_CUR) or croak "seek: $!";
3128        truncate $fh, $offset or croak "truncate: $!";
3129}
3130
3131sub mkfile {
3132        my ($path) = @_;
3133        unless (-e $path) {
3134                my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
3135                mkpath([$dir]) unless -d $dir;
3136                open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
3137                close $fh or die "Couldn't close (create) $path: $!\n";
3138        }
3139}
3140
3141sub rev_map_set {
3142        my ($self, $rev, $commit, $update_ref, $uuid) = @_;
3143        length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
3144        my $db = $self->map_path($uuid);
3145        my $db_lock = "$db.lock";
3146        my $sig;
3147        $update_ref ||= 0;
3148        if ($update_ref) {
3149                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3150                            $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
3151        }
3152        mkfile($db);
3153
3154        $LOCKFILES{$db_lock} = 1;
3155        my $sync;
3156        # both of these options make our .rev_db file very, very important
3157        # and we can't afford to lose it because rebuild() won't work
3158        if ($self->use_svm_props || $self->no_metadata) {
3159                $sync = 1;
3160                copy($db, $db_lock) or die "rev_map_set(@_): ",
3161                                           "Failed to copy: ",
3162                                           "$db => $db_lock ($!)\n";
3163        } else {
3164                rename $db, $db_lock or die "rev_map_set(@_): ",
3165                                            "Failed to rename: ",
3166                                            "$db => $db_lock ($!)\n";
3167        }
3168
3169        sysopen(my $fh, $db_lock, O_RDWR | O_CREAT)
3170             or croak "Couldn't open $db_lock: $!\n";
3171        $update_ref eq 'reset' ? _rev_map_reset($fh, $rev, $commit) :
3172                                 _rev_map_set($fh, $rev, $commit);
3173        if ($sync) {
3174                $fh->flush or die "Couldn't flush $db_lock: $!\n";
3175                $fh->sync or die "Couldn't sync $db_lock: $!\n";
3176        }
3177        close $fh or croak $!;
3178        if ($update_ref) {
3179                $_head = $self;
3180                my $note = "";
3181                $note = " ($update_ref)" if ($update_ref !~ /^\d*$/);
3182                command_noisy('update-ref', '-m', "r$rev$note",
3183                              $self->refname, $commit);
3184        }
3185        rename $db_lock, $db or die "rev_map_set(@_): ", "Failed to rename: ",
3186                                    "$db_lock => $db ($!)\n";
3187        delete $LOCKFILES{$db_lock};
3188        if ($update_ref) {
3189                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3190                            $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
3191                kill $sig, $$ if defined $sig;
3192        }
3193}
3194
3195# If want_commit, this will return an array of (rev, commit) where
3196# commit _must_ be a valid commit in the archive.
3197# Otherwise, it'll return the max revision (whether or not the
3198# commit is valid or just a 0x40 placeholder).
3199sub rev_map_max {
3200        my ($self, $want_commit) = @_;
3201        $self->rebuild;
3202        my ($r, $c) = $self->rev_map_max_norebuild($want_commit);
3203        $want_commit ? ($r, $c) : $r;
3204}
3205
3206sub rev_map_max_norebuild {
3207        my ($self, $want_commit) = @_;
3208        my $map_path = $self->map_path;
3209        stat $map_path or return $want_commit ? (0, undef) : 0;
3210        sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3211        binmode $fh or croak "binmode: $!";
3212        my $size = (stat($fh))[7];
3213        ($size % 24) == 0 or croak "inconsistent size: $size";
3214
3215        if ($size == 0) {
3216                close $fh or croak "close: $!";
3217                return $want_commit ? (0, undef) : 0;
3218        }
3219
3220        sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3221        sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3222        my ($r, $c) = unpack(rev_map_fmt, $buf);
3223        if ($want_commit && $c eq ('0' x40)) {
3224                if ($size < 48) {
3225                        return $want_commit ? (0, undef) : 0;
3226                }
3227                sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3228                sysread($fh, $buf, 24) == 24 or croak "read: $!";
3229                ($r, $c) = unpack(rev_map_fmt, $buf);
3230                if ($c eq ('0'x40)) {
3231                        croak "Penultimate record is all-zeroes in $map_path";
3232                }
3233        }
3234        close $fh or croak "close: $!";
3235        $want_commit ? ($r, $c) : $r;
3236}
3237
3238sub rev_map_get {
3239        my ($self, $rev, $uuid) = @_;
3240        my $map_path = $self->map_path($uuid);
3241        return undef unless -e $map_path;
3242
3243        sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3244        my $c = _rev_map_get($fh, $rev);
3245        close($fh) or croak "close: $!";
3246        $c
3247}
3248
3249sub _rev_map_get {
3250        my ($fh, $rev) = @_;
3251
3252        binmode $fh or croak "binmode: $!";
3253        my $size = (stat($fh))[7];
3254        ($size % 24) == 0 or croak "inconsistent size: $size";
3255
3256        if ($size == 0) {
3257                return undef;
3258        }
3259
3260        my ($l, $u) = (0, $size - 24);
3261        my ($r, $c, $buf);
3262
3263        while ($l <= $u) {
3264                my $i = int(($l/24 + $u/24) / 2) * 24;
3265                sysseek($fh, $i, SEEK_SET) or croak "seek: $!";
3266                sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3267                my ($r, $c) = unpack('NH40', $buf);
3268
3269                if ($r < $rev) {
3270                        $l = $i + 24;
3271                } elsif ($r > $rev) {
3272                        $u = $i - 24;
3273                } else { # $r == $rev
3274                        return $c eq ('0' x 40) ? undef : $c;
3275                }
3276        }
3277        undef;
3278}
3279
3280# Finds the first svn revision that exists on (if $eq_ok is true) or
3281# before $rev for the current branch.  It will not search any lower
3282# than $min_rev.  Returns the git commit hash and svn revision number
3283# if found, else (undef, undef).
3284sub find_rev_before {
3285        my ($self, $rev, $eq_ok, $min_rev) = @_;
3286        --$rev unless $eq_ok;
3287        $min_rev ||= 1;
3288        my $max_rev = $self->rev_map_max;
3289        $rev = $max_rev if ($rev > $max_rev);
3290        while ($rev >= $min_rev) {
3291                if (my $c = $self->rev_map_get($rev)) {
3292                        return ($rev, $c);
3293                }
3294                --$rev;
3295        }
3296        return (undef, undef);
3297}
3298
3299# Finds the first svn revision that exists on (if $eq_ok is true) or
3300# after $rev for the current branch.  It will not search any higher
3301# than $max_rev.  Returns the git commit hash and svn revision number
3302# if found, else (undef, undef).
3303sub find_rev_after {
3304        my ($self, $rev, $eq_ok, $max_rev) = @_;
3305        ++$rev unless $eq_ok;
3306        $max_rev ||= $self->rev_map_max;
3307        while ($rev <= $max_rev) {
3308                if (my $c = $self->rev_map_get($rev)) {
3309                        return ($rev, $c);
3310                }
3311                ++$rev;
3312        }
3313        return (undef, undef);
3314}
3315
3316sub _new {
3317        my ($class, $repo_id, $ref_id, $path) = @_;
3318        unless (defined $repo_id && length $repo_id) {
3319                $repo_id = $Git::SVN::default_repo_id;
3320        }
3321        unless (defined $ref_id && length $ref_id) {
3322                $_prefix = '' unless defined($_prefix);
3323                $_[2] = $ref_id = $_prefix . $Git::SVN::default_ref_id;
3324        }
3325        $_[1] = $repo_id;
3326        my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
3327        $_[3] = $path = '' unless (defined $path);
3328        mkpath(["$ENV{GIT_DIR}/svn"]);
3329        bless {
3330                ref_id => $ref_id, dir => $dir, index => "$dir/index",
3331                path => $path, config => "$ENV{GIT_DIR}/svn/config",
3332                map_root => "$dir/.rev_map", repo_id => $repo_id }, $class;
3333}
3334
3335# for read-only access of old .rev_db formats
3336sub unlink_rev_db_symlink {
3337        my ($self) = @_;
3338        my $link = $self->rev_db_path;
3339        $link =~ s/\.[\w-]+$// or croak "missing UUID at the end of $link";
3340        if (-l $link) {
3341                unlink $link or croak "unlink: $link failed!";
3342        }
3343}
3344
3345sub rev_db_path {
3346        my ($self, $uuid) = @_;
3347        my $db_path = $self->map_path($uuid);
3348        $db_path =~ s{/\.rev_map\.}{/\.rev_db\.}
3349            or croak "map_path: $db_path does not contain '/.rev_map.' !";
3350        $db_path;
3351}
3352
3353# the new replacement for .rev_db
3354sub map_path {
3355        my ($self, $uuid) = @_;
3356        $uuid ||= $self->ra_uuid;
3357        "$self->{map_root}.$uuid";
3358}
3359
3360sub uri_encode {
3361        my ($f) = @_;
3362        $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
3363        $f
3364}
3365
3366sub remove_username {
3367        $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
3368}
3369
3370package Git::SVN::Prompt;
3371use strict;
3372use warnings;
3373require SVN::Core;
3374use vars qw/$_no_auth_cache $_username/;
3375
3376sub simple {
3377        my ($cred, $realm, $default_username, $may_save, $pool) = @_;
3378        $may_save = undef if $_no_auth_cache;
3379        $default_username = $_username if defined $_username;
3380        if (defined $default_username && length $default_username) {
3381                if (defined $realm && length $realm) {
3382                        print STDERR "Authentication realm: $realm\n";
3383                        STDERR->flush;
3384                }
3385                $cred->username($default_username);
3386        } else {
3387                username($cred, $realm, $may_save, $pool);
3388        }
3389        $cred->password(_read_password("Password for '" .
3390                                       $cred->username . "': ", $realm));
3391        $cred->may_save($may_save);
3392        $SVN::_Core::SVN_NO_ERROR;
3393}
3394
3395sub ssl_server_trust {
3396        my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
3397        $may_save = undef if $_no_auth_cache;
3398        print STDERR "Error validating server certificate for '$realm':\n";
3399        {
3400                no warnings 'once';
3401                # All variables SVN::Auth::SSL::* are used only once,
3402                # so we're shutting up Perl warnings about this.
3403                if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
3404                        print STDERR " - The certificate is not issued ",
3405                            "by a trusted authority. Use the\n",
3406                            "   fingerprint to validate ",
3407                            "the certificate manually!\n";
3408                }
3409                if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
3410                        print STDERR " - The certificate hostname ",
3411                            "does not match.\n";
3412                }
3413                if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
3414                        print STDERR " - The certificate is not yet valid.\n";
3415                }
3416                if ($failures & $SVN::Auth::SSL::EXPIRED) {
3417                        print STDERR " - The certificate has expired.\n";
3418                }
3419                if ($failures & $SVN::Auth::SSL::OTHER) {
3420                        print STDERR " - The certificate has ",
3421                            "an unknown error.\n";
3422                }
3423        } # no warnings 'once'
3424        printf STDERR
3425                "Certificate information:\n".
3426                " - Hostname: %s\n".
3427                " - Valid: from %s until %s\n".
3428                " - Issuer: %s\n".
3429                " - Fingerprint: %s\n",
3430                map $cert_info->$_, qw(hostname valid_from valid_until
3431                                       issuer_dname fingerprint);
3432        my $choice;
3433prompt:
3434        print STDERR $may_save ?
3435              "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
3436              "(R)eject or accept (t)emporarily? ";
3437        STDERR->flush;
3438        $choice = lc(substr(<STDIN> || 'R', 0, 1));
3439        if ($choice =~ /^t$/i) {
3440                $cred->may_save(undef);
3441        } elsif ($choice =~ /^r$/i) {
3442                return -1;
3443        } elsif ($may_save && $choice =~ /^p$/i) {
3444                $cred->may_save($may_save);
3445        } else {
3446                goto prompt;
3447        }
3448        $cred->accepted_failures($failures);
3449        $SVN::_Core::SVN_NO_ERROR;
3450}
3451
3452sub ssl_client_cert {
3453        my ($cred, $realm, $may_save, $pool) = @_;
3454        $may_save = undef if $_no_auth_cache;
3455        print STDERR "Client certificate filename: ";
3456        STDERR->flush;
3457        chomp(my $filename = <STDIN>);
3458        $cred->cert_file($filename);
3459        $cred->may_save($may_save);
3460        $SVN::_Core::SVN_NO_ERROR;
3461}
3462
3463sub ssl_client_cert_pw {
3464        my ($cred, $realm, $may_save, $pool) = @_;
3465        $may_save = undef if $_no_auth_cache;
3466        $cred->password(_read_password("Password: ", $realm));
3467        $cred->may_save($may_save);
3468        $SVN::_Core::SVN_NO_ERROR;
3469}
3470
3471sub username {
3472        my ($cred, $realm, $may_save, $pool) = @_;
3473        $may_save = undef if $_no_auth_cache;
3474        if (defined $realm && length $realm) {
3475                print STDERR "Authentication realm: $realm\n";
3476        }
3477        my $username;
3478        if (defined $_username) {
3479                $username = $_username;
3480        } else {
3481                print STDERR "Username: ";
3482                STDERR->flush;
3483                chomp($username = <STDIN>);
3484        }
3485        $cred->username($username);
3486        $cred->may_save($may_save);
3487        $SVN::_Core::SVN_NO_ERROR;
3488}
3489
3490sub _read_password {
3491        my ($prompt, $realm) = @_;
3492        print STDERR $prompt;
3493        STDERR->flush;
3494        require Term::ReadKey;
3495        Term::ReadKey::ReadMode('noecho');
3496        my $password = '';
3497        while (defined(my $key = Term::ReadKey::ReadKey(0))) {
3498                last if $key =~ /[\012\015]/; # \n\r
3499                $password .= $key;
3500        }
3501        Term::ReadKey::ReadMode('restore');
3502        print STDERR "\n";
3503        STDERR->flush;
3504        $password;
3505}
3506
3507package SVN::Git::Fetcher;
3508use vars qw/@ISA/;
3509use strict;
3510use warnings;
3511use Carp qw/croak/;
3512use File::Temp qw/tempfile/;
3513use IO::File qw//;
3514use vars qw/$_ignore_regex/;
3515
3516# file baton members: path, mode_a, mode_b, pool, fh, blob, base
3517sub new {
3518        my ($class, $git_svn, $switch_path) = @_;
3519        my $self = SVN::Delta::Editor->new;
3520        bless $self, $class;
3521        if (exists $git_svn->{last_commit}) {
3522                $self->{c} = $git_svn->{last_commit};
3523                $self->{empty_symlinks} =
3524                                  _mark_empty_symlinks($git_svn, $switch_path);
3525        }
3526        $self->{ignore_regex} = eval { command_oneline('config', '--get',
3527                             "svn-remote.$git_svn->{repo_id}.ignore-paths") };
3528        $self->{empty} = {};
3529        $self->{dir_prop} = {};
3530        $self->{file_prop} = {};
3531        $self->{absent_dir} = {};
3532        $self->{absent_file} = {};
3533        $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
3534        $self;
3535}
3536
3537# this uses the Ra object, so it must be called before do_{switch,update},
3538# not inside them (when the Git::SVN::Fetcher object is passed) to
3539# do_{switch,update}
3540sub _mark_empty_symlinks {
3541        my ($git_svn, $switch_path) = @_;
3542        my $bool = Git::config_bool('svn.brokenSymlinkWorkaround');
3543        return {} if (!defined($bool)) || (defined($bool) && ! $bool);
3544
3545        my %ret;
3546        my ($rev, $cmt) = $git_svn->last_rev_commit;
3547        return {} unless ($rev && $cmt);
3548
3549        # allow the warning to be printed for each revision we fetch to
3550        # ensure the user sees it.  The user can also disable the workaround
3551        # on the repository even while git svn is running and the next
3552        # revision fetched will skip this expensive function.
3553        my $printed_warning;
3554        chomp(my $empty_blob = `git hash-object -t blob --stdin < /dev/null`);
3555        my ($ls, $ctx) = command_output_pipe(qw/ls-tree -r -z/, $cmt);
3556        local $/ = "\0";
3557        my $pfx = defined($switch_path) ? $switch_path : $git_svn->{path};
3558        $pfx .= '/' if length($pfx);
3559        while (<$ls>) {
3560                chomp;
3561                s/\A100644 blob $empty_blob\t//o or next;
3562                unless ($printed_warning) {
3563                        print STDERR "Scanning for empty symlinks, ",
3564                                     "this may take a while if you have ",
3565                                     "many empty files\n",
3566                                     "You may disable this with `",
3567                                     "git config svn.brokenSymlinkWorkaround ",
3568                                     "false'.\n",
3569                                     "This may be done in a different ",
3570                                     "terminal without restarting ",
3571                                     "git svn\n";
3572                        $printed_warning = 1;
3573                }
3574                my $path = $_;
3575                my (undef, $props) =
3576                               $git_svn->ra->get_file($pfx.$path, $rev, undef);
3577                if ($props->{'svn:special'}) {
3578                        $ret{$path} = 1;
3579                }
3580        }
3581        command_close_pipe($ls, $ctx);
3582        \%ret;
3583}
3584
3585# returns true if a given path is inside a ".git" directory
3586sub in_dot_git {
3587        $_[0] =~ m{(?:^|/)\.git(?:/|$)};
3588}
3589
3590# return value: 0 -- don't ignore, 1 -- ignore
3591sub is_path_ignored {
3592        my ($self, $path) = @_;
3593        return 1 if in_dot_git($path);
3594        return 1 if defined($self->{ignore_regex}) &&
3595                    $path =~ m!$self->{ignore_regex}!;
3596        return 0 unless defined($_ignore_regex);
3597        return 1 if $path =~ m!$_ignore_regex!o;
3598        return 0;
3599}
3600
3601sub set_path_strip {
3602        my ($self, $path) = @_;
3603        $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
3604}
3605
3606sub open_root {
3607        { path => '' };
3608}
3609
3610sub open_directory {
3611        my ($self, $path, $pb, $rev) = @_;
3612        { path => $path };
3613}
3614
3615sub git_path {
3616        my ($self, $path) = @_;
3617        if ($self->{path_strip}) {
3618                $path =~ s!$self->{path_strip}!! or
3619                  die "Failed to strip path '$path' ($self->{path_strip})\n";
3620        }
3621        $path;
3622}
3623
3624sub delete_entry {
3625        my ($self, $path, $rev, $pb) = @_;
3626        return undef if $self->is_path_ignored($path);
3627
3628        my $gpath = $self->git_path($path);
3629        return undef if ($gpath eq '');
3630
3631        # remove entire directories.
3632        my ($tree) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
3633                         =~ /\A040000 tree ([a-f\d]{40})\t\Q$gpath\E\0/);
3634        if ($tree) {
3635                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3636                                                     -r --name-only -z/,
3637                                                     $tree);
3638                local $/ = "\0";
3639                while (<$ls>) {
3640                        chomp;
3641                        my $rmpath = "$gpath/$_";
3642                        $self->{gii}->remove($rmpath);
3643                        print "\tD\t$rmpath\n" unless $::_q;
3644                }
3645                print "\tD\t$gpath/\n" unless $::_q;
3646                command_close_pipe($ls, $ctx);
3647                $self->{empty}->{$path} = 0
3648        } else {
3649                $self->{gii}->remove($gpath);
3650                print "\tD\t$gpath\n" unless $::_q;
3651        }
3652        undef;
3653}
3654
3655sub open_file {
3656        my ($self, $path, $pb, $rev) = @_;
3657        my ($mode, $blob);
3658
3659        goto out if $self->is_path_ignored($path);
3660
3661        my $gpath = $self->git_path($path);
3662        ($mode, $blob) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
3663                             =~ /\A(\d{6}) blob ([a-f\d]{40})\t\Q$gpath\E\0/);
3664        unless (defined $mode && defined $blob) {
3665                die "$path was not found in commit $self->{c} (r$rev)\n";
3666        }
3667        if ($mode eq '100644' && $self->{empty_symlinks}->{$path}) {
3668                $mode = '120000';
3669        }
3670out:
3671        { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
3672          pool => SVN::Pool->new, action => 'M' };
3673}
3674
3675sub add_file {
3676        my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
3677        my $mode;
3678
3679        if (!$self->is_path_ignored($path)) {
3680                my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3681                delete $self->{empty}->{$dir};
3682                $mode = '100644';
3683        }
3684        { path => $path, mode_a => $mode, mode_b => $mode,
3685          pool => SVN::Pool->new, action => 'A' };
3686}
3687
3688sub add_directory {
3689        my ($self, $path, $cp_path, $cp_rev) = @_;
3690        goto out if $self->is_path_ignored($path);
3691        my $gpath = $self->git_path($path);
3692        if ($gpath eq '') {
3693                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3694                                                     -r --name-only -z/,
3695                                                     $self->{c});
3696                local $/ = "\0";
3697                while (<$ls>) {
3698                        chomp;
3699                        $self->{gii}->remove($_);
3700                        print "\tD\t$_\n" unless $::_q;
3701                }
3702                command_close_pipe($ls, $ctx);
3703                $self->{empty}->{$path} = 0;
3704        }
3705        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3706        delete $self->{empty}->{$dir};
3707        $self->{empty}->{$path} = 1;
3708out:
3709        { path => $path };
3710}
3711
3712sub change_dir_prop {
3713        my ($self, $db, $prop, $value) = @_;
3714        return undef if $self->is_path_ignored($db->{path});
3715        $self->{dir_prop}->{$db->{path}} ||= {};
3716        $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
3717        undef;
3718}
3719
3720sub absent_directory {
3721        my ($self, $path, $pb) = @_;
3722        return undef if $self->is_path_ignored($path);
3723        $self->{absent_dir}->{$pb->{path}} ||= [];
3724        push @{$self->{absent_dir}->{$pb->{path}}}, $path;
3725        undef;
3726}
3727
3728sub absent_file {
3729        my ($self, $path, $pb) = @_;
3730        return undef if $self->is_path_ignored($path);
3731        $self->{absent_file}->{$pb->{path}} ||= [];
3732        push @{$self->{absent_file}->{$pb->{path}}}, $path;
3733        undef;
3734}
3735
3736sub change_file_prop {
3737        my ($self, $fb, $prop, $value) = @_;
3738        return undef if $self->is_path_ignored($fb->{path});
3739        if ($prop eq 'svn:executable') {
3740                if ($fb->{mode_b} != 120000) {
3741                        $fb->{mode_b} = defined $value ? 100755 : 100644;
3742                }
3743        } elsif ($prop eq 'svn:special') {
3744                $fb->{mode_b} = defined $value ? 120000 : 100644;
3745        } else {
3746                $self->{file_prop}->{$fb->{path}} ||= {};
3747                $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
3748        }
3749        undef;
3750}
3751
3752sub apply_textdelta {
3753        my ($self, $fb, $exp) = @_;
3754        return undef if $self->is_path_ignored($fb->{path});
3755        my $fh = $::_repository->temp_acquire('svn_delta');
3756        # $fh gets auto-closed() by SVN::TxDelta::apply(),
3757        # (but $base does not,) so dup() it for reading in close_file
3758        open my $dup, '<&', $fh or croak $!;
3759        my $base = $::_repository->temp_acquire('git_blob');
3760
3761        if ($fb->{blob}) {
3762                my ($base_is_link, $size);
3763
3764                if ($fb->{mode_a} eq '120000' &&
3765                    ! $self->{empty_symlinks}->{$fb->{path}}) {
3766                        print $base 'link ' or die "print $!\n";
3767                        $base_is_link = 1;
3768                }
3769        retry:
3770                $size = $::_repository->cat_blob($fb->{blob}, $base);
3771                die "Failed to read object $fb->{blob}" if ($size < 0);
3772
3773                if (defined $exp) {
3774                        seek $base, 0, 0 or croak $!;
3775                        my $got = ::md5sum($base);
3776                        if ($got ne $exp) {
3777                                my $err = "Checksum mismatch: ".
3778                                       "$fb->{path} $fb->{blob}\n" .
3779                                       "expected: $exp\n" .
3780                                       "     got: $got\n";
3781                                if ($base_is_link) {
3782                                        warn $err,
3783                                             "Retrying... (possibly ",
3784                                             "a bad symlink from SVN)\n";
3785                                        $::_repository->temp_reset($base);
3786                                        $base_is_link = 0;
3787                                        goto retry;
3788                                }
3789                                die $err;
3790                        }
3791                }
3792        }
3793        seek $base, 0, 0 or croak $!;
3794        $fb->{fh} = $fh;
3795        $fb->{base} = $base;
3796        [ SVN::TxDelta::apply($base, $dup, undef, $fb->{path}, $fb->{pool}) ];
3797}
3798
3799sub close_file {
3800        my ($self, $fb, $exp) = @_;
3801        return undef if $self->is_path_ignored($fb->{path});
3802
3803        my $hash;
3804        my $path = $self->git_path($fb->{path});
3805        if (my $fh = $fb->{fh}) {
3806                if (defined $exp) {
3807                        seek($fh, 0, 0) or croak $!;
3808                        my $got = ::md5sum($fh);
3809                        if ($got ne $exp) {
3810                                die "Checksum mismatch: $path\n",
3811                                    "expected: $exp\n    got: $got\n";
3812                        }
3813                }
3814                if ($fb->{mode_b} == 120000) {
3815                        sysseek($fh, 0, 0) or croak $!;
3816                        my $rd = sysread($fh, my $buf, 5);
3817
3818                        if (!defined $rd) {
3819                                croak "sysread: $!\n";
3820                        } elsif ($rd == 0) {
3821                                warn "$path has mode 120000",
3822                                     " but it points to nothing\n",
3823                                     "converting to an empty file with mode",
3824                                     " 100644\n";
3825                                $fb->{mode_b} = '100644';
3826                        } elsif ($buf ne 'link ') {
3827                                warn "$path has mode 120000",
3828                                     " but is not a link\n";
3829                        } else {
3830                                my $tmp_fh = $::_repository->temp_acquire(
3831                                        'svn_hash');
3832                                my $res;
3833                                while ($res = sysread($fh, my $str, 1024)) {
3834                                        my $out = syswrite($tmp_fh, $str, $res);
3835                                        defined($out) && $out == $res
3836                                                or croak("write ",
3837                                                        Git::temp_path($tmp_fh),
3838                                                        ": $!\n");
3839                                }
3840                                defined $res or croak $!;
3841
3842                                ($fh, $tmp_fh) = ($tmp_fh, $fh);
3843                                Git::temp_release($tmp_fh, 1);
3844                        }
3845                }
3846
3847                $hash = $::_repository->hash_and_insert_object(
3848                                Git::temp_path($fh));
3849                $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
3850
3851                Git::temp_release($fb->{base}, 1);
3852                Git::temp_release($fh, 1);
3853        } else {
3854                $hash = $fb->{blob} or die "no blob information\n";
3855        }
3856        $fb->{pool}->clear;
3857        $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
3858        print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
3859        undef;
3860}
3861
3862sub abort_edit {
3863        my $self = shift;
3864        $self->{nr} = $self->{gii}->{nr};
3865        delete $self->{gii};
3866        $self->SUPER::abort_edit(@_);
3867}
3868
3869sub close_edit {
3870        my $self = shift;
3871        $self->{git_commit_ok} = 1;
3872        $self->{nr} = $self->{gii}->{nr};
3873        delete $self->{gii};
3874        $self->SUPER::close_edit(@_);
3875}
3876
3877package SVN::Git::Editor;
3878use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
3879use strict;
3880use warnings;
3881use Carp qw/croak/;
3882use IO::File;
3883
3884sub new {
3885        my ($class, $opts) = @_;
3886        foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
3887                die "$_ required!\n" unless (defined $opts->{$_});
3888        }
3889
3890        my $pool = SVN::Pool->new;
3891        my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
3892        my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
3893                                     $opts->{r}, $mods);
3894
3895        # $opts->{ra} functions should not be used after this:
3896        my @ce  = $opts->{ra}->get_commit_editor($opts->{log},
3897                                                $opts->{editor_cb}, $pool);
3898        my $self = SVN::Delta::Editor->new(@ce, $pool);
3899        bless $self, $class;
3900        foreach (qw/svn_path r tree_a tree_b/) {
3901                $self->{$_} = $opts->{$_};
3902        }
3903        $self->{url} = $opts->{ra}->{url};
3904        $self->{mods} = $mods;
3905        $self->{types} = $types;
3906        $self->{pool} = $pool;
3907        $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
3908        $self->{rm} = { };
3909        $self->{path_prefix} = length $self->{svn_path} ?
3910                               "$self->{svn_path}/" : '';
3911        $self->{config} = $opts->{config};
3912        return $self;
3913}
3914
3915sub generate_diff {
3916        my ($tree_a, $tree_b) = @_;
3917        my @diff_tree = qw(diff-tree -z -r);
3918        if ($_cp_similarity) {
3919                push @diff_tree, "-C$_cp_similarity";
3920        } else {
3921                push @diff_tree, '-C';
3922        }
3923        push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
3924        push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
3925        push @diff_tree, $tree_a, $tree_b;
3926        my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
3927        local $/ = "\0";
3928        my $state = 'meta';
3929        my @mods;
3930        while (<$diff_fh>) {
3931                chomp $_; # this gets rid of the trailing "\0"
3932                if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
3933                                        ($::sha1)\s($::sha1)\s
3934                                        ([MTCRAD])\d*$/xo) {
3935                        push @mods, {   mode_a => $1, mode_b => $2,
3936                                        sha1_a => $3, sha1_b => $4,
3937                                        chg => $5 };
3938                        if ($5 =~ /^(?:C|R)$/) {
3939                                $state = 'file_a';
3940                        } else {
3941                                $state = 'file_b';
3942                        }
3943                } elsif ($state eq 'file_a') {
3944                        my $x = $mods[$#mods] or croak "Empty array\n";
3945                        if ($x->{chg} !~ /^(?:C|R)$/) {
3946                                croak "Error parsing $_, $x->{chg}\n";
3947                        }
3948                        $x->{file_a} = $_;
3949                        $state = 'file_b';
3950                } elsif ($state eq 'file_b') {
3951                        my $x = $mods[$#mods] or croak "Empty array\n";
3952                        if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
3953                                croak "Error parsing $_, $x->{chg}\n";
3954                        }
3955                        if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
3956                                croak "Error parsing $_, $x->{chg}\n";
3957                        }
3958                        $x->{file_b} = $_;
3959                        $state = 'meta';
3960                } else {
3961                        croak "Error parsing $_\n";
3962                }
3963        }
3964        command_close_pipe($diff_fh, $ctx);
3965        \@mods;
3966}
3967
3968sub check_diff_paths {
3969        my ($ra, $pfx, $rev, $mods) = @_;
3970        my %types;
3971        $pfx .= '/' if length $pfx;
3972
3973        sub type_diff_paths {
3974                my ($ra, $types, $path, $rev) = @_;
3975                my @p = split m#/+#, $path;
3976                my $c = shift @p;
3977                unless (defined $types->{$c}) {
3978                        $types->{$c} = $ra->check_path($c, $rev);
3979                }
3980                while (@p) {
3981                        $c .= '/' . shift @p;
3982                        next if defined $types->{$c};
3983                        $types->{$c} = $ra->check_path($c, $rev);
3984                }
3985        }
3986
3987        foreach my $m (@$mods) {
3988                foreach my $f (qw/file_a file_b/) {
3989                        next unless defined $m->{$f};
3990                        my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
3991                        if (length $pfx.$dir && ! defined $types{$dir}) {
3992                                type_diff_paths($ra, \%types, $pfx.$dir, $rev);
3993                        }
3994                }
3995        }
3996        \%types;
3997}
3998
3999sub split_path {
4000        return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
4001}
4002
4003sub repo_path {
4004        my ($self, $path) = @_;
4005        $self->{path_prefix}.(defined $path ? $path : '');
4006}
4007
4008sub url_path {
4009        my ($self, $path) = @_;
4010        if ($self->{url} =~ m#^https?://#) {
4011                $path =~ s!([^~a-zA-Z0-9_./-])!uc sprintf("%%%02x",ord($1))!eg;
4012        }
4013        $self->{url} . '/' . $self->repo_path($path);
4014}
4015
4016sub rmdirs {
4017        my ($self) = @_;
4018        my $rm = $self->{rm};
4019        delete $rm->{''}; # we never delete the url we're tracking
4020        return unless %$rm;
4021
4022        foreach (keys %$rm) {
4023                my @d = split m#/#, $_;
4024                my $c = shift @d;
4025                $rm->{$c} = 1;
4026                while (@d) {
4027                        $c .= '/' . shift @d;
4028                        $rm->{$c} = 1;
4029                }
4030        }
4031        delete $rm->{$self->{svn_path}};
4032        delete $rm->{''}; # we never delete the url we're tracking
4033        return unless %$rm;
4034
4035        my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
4036                                             $self->{tree_b});
4037        local $/ = "\0";
4038        while (<$fh>) {
4039                chomp;
4040                my @dn = split m#/#, $_;
4041                while (pop @dn) {
4042                        delete $rm->{join '/', @dn};
4043                }
4044                unless (%$rm) {
4045                        close $fh;
4046                        return;
4047                }
4048        }
4049        command_close_pipe($fh, $ctx);
4050
4051        my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
4052        foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
4053                $self->close_directory($bat->{$d}, $p);
4054                my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
4055                print "\tD+\t$d/\n" unless $::_q;
4056                $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
4057                delete $bat->{$d};
4058        }
4059}
4060
4061sub open_or_add_dir {
4062        my ($self, $full_path, $baton) = @_;
4063        my $t = $self->{types}->{$full_path};
4064        if (!defined $t) {
4065                die "$full_path not known in r$self->{r} or we have a bug!\n";
4066        }
4067        {
4068                no warnings 'once';
4069                # SVN::Node::none and SVN::Node::file are used only once,
4070                # so we're shutting up Perl's warnings about them.
4071                if ($t == $SVN::Node::none) {
4072                        return $self->add_directory($full_path, $baton,
4073                            undef, -1, $self->{pool});
4074                } elsif ($t == $SVN::Node::dir) {
4075                        return $self->open_directory($full_path, $baton,
4076                            $self->{r}, $self->{pool});
4077                } # no warnings 'once'
4078                print STDERR "$full_path already exists in repository at ",
4079                    "r$self->{r} and it is not a directory (",
4080                    ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
4081        } # no warnings 'once'
4082        exit 1;
4083}
4084
4085sub ensure_path {
4086        my ($self, $path) = @_;
4087        my $bat = $self->{bat};
4088        my $repo_path = $self->repo_path($path);
4089        return $bat->{''} unless (length $repo_path);
4090        my @p = split m#/+#, $repo_path;
4091        my $c = shift @p;
4092        $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
4093        while (@p) {
4094                my $c0 = $c;
4095                $c .= '/' . shift @p;
4096                $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
4097        }
4098        return $bat->{$c};
4099}
4100
4101# Subroutine to convert a globbing pattern to a regular expression.
4102# From perl cookbook.
4103sub glob2pat {
4104        my $globstr = shift;
4105        my %patmap = ('*' => '.*', '?' => '.', '[' => '[', ']' => ']');
4106        $globstr =~ s{(.)} { $patmap{$1} || "\Q$1" }ge;
4107        return '^' . $globstr . '$';
4108}
4109
4110sub check_autoprop {
4111        my ($self, $pattern, $properties, $file, $fbat) = @_;
4112        # Convert the globbing pattern to a regular expression.
4113        my $regex = glob2pat($pattern);
4114        # Check if the pattern matches the file name.
4115        if($file =~ m/($regex)/) {
4116                # Parse the list of properties to set.
4117                my @props = split(/;/, $properties);
4118                foreach my $prop (@props) {
4119                        # Parse 'name=value' syntax and set the property.
4120                        if ($prop =~ /([^=]+)=(.*)/) {
4121                                my ($n,$v) = ($1,$2);
4122                                for ($n, $v) {
4123                                        s/^\s+//; s/\s+$//;
4124                                }
4125                                $self->change_file_prop($fbat, $n, $v);
4126                        }
4127                }
4128        }
4129}
4130
4131sub apply_autoprops {
4132        my ($self, $file, $fbat) = @_;
4133        my $conf_t = ${$self->{config}}{'config'};
4134        no warnings 'once';
4135        # Check [miscellany]/enable-auto-props in svn configuration.
4136        if (SVN::_Core::svn_config_get_bool(
4137                $conf_t,
4138                $SVN::_Core::SVN_CONFIG_SECTION_MISCELLANY,
4139                $SVN::_Core::SVN_CONFIG_OPTION_ENABLE_AUTO_PROPS,
4140                0)) {
4141                # Auto-props are enabled.  Enumerate them to look for matches.
4142                my $callback = sub {
4143                        $self->check_autoprop($_[0], $_[1], $file, $fbat);
4144                };
4145                SVN::_Core::svn_config_enumerate(
4146                        $conf_t,
4147                        $SVN::_Core::SVN_CONFIG_SECTION_AUTO_PROPS,
4148                        $callback);
4149        }
4150}
4151
4152sub A {
4153        my ($self, $m) = @_;
4154        my ($dir, $file) = split_path($m->{file_b});
4155        my $pbat = $self->ensure_path($dir);
4156        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4157                                        undef, -1);
4158        print "\tA\t$m->{file_b}\n" unless $::_q;
4159        $self->apply_autoprops($file, $fbat);
4160        $self->chg_file($fbat, $m);
4161        $self->close_file($fbat,undef,$self->{pool});
4162}
4163
4164sub C {
4165        my ($self, $m) = @_;
4166        my ($dir, $file) = split_path($m->{file_b});
4167        my $pbat = $self->ensure_path($dir);
4168        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4169                                $self->url_path($m->{file_a}), $self->{r});
4170        print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
4171        $self->chg_file($fbat, $m);
4172        $self->close_file($fbat,undef,$self->{pool});
4173}
4174
4175sub delete_entry {
4176        my ($self, $path, $pbat) = @_;
4177        my $rpath = $self->repo_path($path);
4178        my ($dir, $file) = split_path($rpath);
4179        $self->{rm}->{$dir} = 1;
4180        $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
4181}
4182
4183sub R {
4184        my ($self, $m) = @_;
4185        my ($dir, $file) = split_path($m->{file_b});
4186        my $pbat = $self->ensure_path($dir);
4187        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4188                                $self->url_path($m->{file_a}), $self->{r});
4189        print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
4190        $self->apply_autoprops($file, $fbat);
4191        $self->chg_file($fbat, $m);
4192        $self->close_file($fbat,undef,$self->{pool});
4193
4194        ($dir, $file) = split_path($m->{file_a});
4195        $pbat = $self->ensure_path($dir);
4196        $self->delete_entry($m->{file_a}, $pbat);
4197}
4198
4199sub M {
4200        my ($self, $m) = @_;
4201        my ($dir, $file) = split_path($m->{file_b});
4202        my $pbat = $self->ensure_path($dir);
4203        my $fbat = $self->open_file($self->repo_path($m->{file_b}),
4204                                $pbat,$self->{r},$self->{pool});
4205        print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
4206        $self->chg_file($fbat, $m);
4207        $self->close_file($fbat,undef,$self->{pool});
4208}
4209
4210sub T { shift->M(@_) }
4211
4212sub change_file_prop {
4213        my ($self, $fbat, $pname, $pval) = @_;
4214        $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
4215}
4216
4217sub _chg_file_get_blob ($$$$) {
4218        my ($self, $fbat, $m, $which) = @_;
4219        my $fh = $::_repository->temp_acquire("git_blob_$which");
4220        if ($m->{"mode_$which"} =~ /^120/) {
4221                print $fh 'link ' or croak $!;
4222                $self->change_file_prop($fbat,'svn:special','*');
4223        } elsif ($m->{mode_a} =~ /^120/ && $m->{"mode_$which"} !~ /^120/) {
4224                $self->change_file_prop($fbat,'svn:special',undef);
4225        }
4226        my $blob = $m->{"sha1_$which"};
4227        return ($fh,) if ($blob =~ /^0{40}$/);
4228        my $size = $::_repository->cat_blob($blob, $fh);
4229        croak "Failed to read object $blob" if ($size < 0);
4230        $fh->flush == 0 or croak $!;
4231        seek $fh, 0, 0 or croak $!;
4232
4233        my $exp = ::md5sum($fh);
4234        seek $fh, 0, 0 or croak $!;
4235        return ($fh, $exp);
4236}
4237
4238sub chg_file {
4239        my ($self, $fbat, $m) = @_;
4240        if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
4241                $self->change_file_prop($fbat,'svn:executable','*');
4242        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
4243                $self->change_file_prop($fbat,'svn:executable',undef);
4244        }
4245        my ($fh_a, $exp_a) = _chg_file_get_blob $self, $fbat, $m, 'a';
4246        my ($fh_b, $exp_b) = _chg_file_get_blob $self, $fbat, $m, 'b';
4247        my $pool = SVN::Pool->new;
4248        my $atd = $self->apply_textdelta($fbat, $exp_a, $pool);
4249        if (-s $fh_a) {
4250                my $txstream = SVN::TxDelta::new ($fh_a, $fh_b, $pool);
4251                my $res = SVN::TxDelta::send_txstream($txstream, @$atd, $pool);
4252                if (defined $res) {
4253                        die "Unexpected result from send_txstream: $res\n",
4254                            "(SVN::Core::VERSION: $SVN::Core::VERSION)\n";
4255                }
4256        } else {
4257                my $got = SVN::TxDelta::send_stream($fh_b, @$atd, $pool);
4258                die "Checksum mismatch\nexpected: $exp_b\ngot: $got\n"
4259                    if ($got ne $exp_b);
4260        }
4261        Git::temp_release($fh_b, 1);
4262        Git::temp_release($fh_a, 1);
4263        $pool->clear;
4264}
4265
4266sub D {
4267        my ($self, $m) = @_;
4268        my ($dir, $file) = split_path($m->{file_b});
4269        my $pbat = $self->ensure_path($dir);
4270        print "\tD\t$m->{file_b}\n" unless $::_q;
4271        $self->delete_entry($m->{file_b}, $pbat);
4272}
4273
4274sub close_edit {
4275        my ($self) = @_;
4276        my ($p,$bat) = ($self->{pool}, $self->{bat});
4277        foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
4278                next if $_ eq '';
4279                $self->close_directory($bat->{$_}, $p);
4280        }
4281        $self->close_directory($bat->{''}, $p);
4282        $self->SUPER::close_edit($p);
4283        $p->clear;
4284}
4285
4286sub abort_edit {
4287        my ($self) = @_;
4288        $self->SUPER::abort_edit($self->{pool});
4289}
4290
4291sub DESTROY {
4292        my $self = shift;
4293        $self->SUPER::DESTROY(@_);
4294        $self->{pool}->clear;
4295}
4296
4297# this drives the editor
4298sub apply_diff {
4299        my ($self) = @_;
4300        my $mods = $self->{mods};
4301        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
4302        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
4303                my $f = $m->{chg};
4304                if (defined $o{$f}) {
4305                        $self->$f($m);
4306                } else {
4307                        fatal("Invalid change type: $f");
4308                }
4309        }
4310        $self->rmdirs if $_rmdir;
4311        if (@$mods == 0) {
4312                $self->abort_edit;
4313        } else {
4314                $self->close_edit;
4315        }
4316        return scalar @$mods;
4317}
4318
4319package Git::SVN::Ra;
4320use vars qw/@ISA $config_dir $_log_window_size/;
4321use strict;
4322use warnings;
4323my ($ra_invalid, $can_do_switch, %ignored_err, $RA);
4324
4325BEGIN {
4326        # enforce temporary pool usage for some simple functions
4327        no strict 'refs';
4328        for my $f (qw/rev_proplist get_latest_revnum get_uuid get_repos_root
4329                      get_file/) {
4330                my $SUPER = "SUPER::$f";
4331                *$f = sub {
4332                        my $self = shift;
4333                        my $pool = SVN::Pool->new;
4334                        my @ret = $self->$SUPER(@_,$pool);
4335                        $pool->clear;
4336                        wantarray ? @ret : $ret[0];
4337                };
4338        }
4339}
4340
4341sub _auth_providers () {
4342        [
4343          SVN::Client::get_simple_provider(),
4344          SVN::Client::get_ssl_server_trust_file_provider(),
4345          SVN::Client::get_simple_prompt_provider(
4346            \&Git::SVN::Prompt::simple, 2),
4347          SVN::Client::get_ssl_client_cert_file_provider(),
4348          SVN::Client::get_ssl_client_cert_prompt_provider(
4349            \&Git::SVN::Prompt::ssl_client_cert, 2),
4350          SVN::Client::get_ssl_client_cert_pw_file_provider(),
4351          SVN::Client::get_ssl_client_cert_pw_prompt_provider(
4352            \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
4353          SVN::Client::get_username_provider(),
4354          SVN::Client::get_ssl_server_trust_prompt_provider(
4355            \&Git::SVN::Prompt::ssl_server_trust),
4356          SVN::Client::get_username_prompt_provider(
4357            \&Git::SVN::Prompt::username, 2)
4358        ]
4359}
4360
4361sub escape_uri_only {
4362        my ($uri) = @_;
4363        my @tmp;
4364        foreach (split m{/}, $uri) {
4365                s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
4366                push @tmp, $_;
4367        }
4368        join('/', @tmp);
4369}
4370
4371sub escape_url {
4372        my ($url) = @_;
4373        if ($url =~ m#^(https?)://([^/]+)(.*)$#) {
4374                my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
4375                $url = "$scheme://$domain$uri";
4376        }
4377        $url;
4378}
4379
4380sub new {
4381        my ($class, $url) = @_;
4382        $url =~ s!/+$!!;
4383        return $RA if ($RA && $RA->{url} eq $url);
4384
4385        SVN::_Core::svn_config_ensure($config_dir, undef);
4386        my ($baton, $callbacks) = SVN::Core::auth_open_helper(_auth_providers);
4387        my $config = SVN::Core::config_get_config($config_dir);
4388        $RA = undef;
4389        my $dont_store_passwords = 1;
4390        my $conf_t = ${$config}{'config'};
4391        {
4392                no warnings 'once';
4393                # The usage of $SVN::_Core::SVN_CONFIG_* variables
4394                # produces warnings that variables are used only once.
4395                # I had not found the better way to shut them up, so
4396                # the warnings of type 'once' are disabled in this block.
4397                if (SVN::_Core::svn_config_get_bool($conf_t,
4398                    $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4399                    $SVN::_Core::SVN_CONFIG_OPTION_STORE_PASSWORDS,
4400                    1) == 0) {
4401                        SVN::_Core::svn_auth_set_parameter($baton,
4402                            $SVN::_Core::SVN_AUTH_PARAM_DONT_STORE_PASSWORDS,
4403                            bless (\$dont_store_passwords, "_p_void"));
4404                }
4405                if (SVN::_Core::svn_config_get_bool($conf_t,
4406                    $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4407                    $SVN::_Core::SVN_CONFIG_OPTION_STORE_AUTH_CREDS,
4408                    1) == 0) {
4409                        $Git::SVN::Prompt::_no_auth_cache = 1;
4410                }
4411        } # no warnings 'once'
4412        my $self = SVN::Ra->new(url => escape_url($url), auth => $baton,
4413                              config => $config,
4414                              pool => SVN::Pool->new,
4415                              auth_provider_callbacks => $callbacks);
4416        $self->{url} = $url;
4417        $self->{svn_path} = $url;
4418        $self->{repos_root} = $self->get_repos_root;
4419        $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
4420        $self->{cache} = { check_path => { r => 0, data => {} },
4421                           get_dir => { r => 0, data => {} } };
4422        $RA = bless $self, $class;
4423}
4424
4425sub check_path {
4426        my ($self, $path, $r) = @_;
4427        my $cache = $self->{cache}->{check_path};
4428        if ($r == $cache->{r} && exists $cache->{data}->{$path}) {
4429                return $cache->{data}->{$path};
4430        }
4431        my $pool = SVN::Pool->new;
4432        my $t = $self->SUPER::check_path($path, $r, $pool);
4433        $pool->clear;
4434        if ($r != $cache->{r}) {
4435                %{$cache->{data}} = ();
4436                $cache->{r} = $r;
4437        }
4438        $cache->{data}->{$path} = $t;
4439}
4440
4441sub get_dir {
4442        my ($self, $dir, $r) = @_;
4443        my $cache = $self->{cache}->{get_dir};
4444        if ($r == $cache->{r}) {
4445                if (my $x = $cache->{data}->{$dir}) {
4446                        return wantarray ? @$x : $x->[0];
4447                }
4448        }
4449        my $pool = SVN::Pool->new;
4450        my ($d, undef, $props) = $self->SUPER::get_dir($dir, $r, $pool);
4451        my %dirents = map { $_ => { kind => $d->{$_}->kind } } keys %$d;
4452        $pool->clear;
4453        if ($r != $cache->{r}) {
4454                %{$cache->{data}} = ();
4455                $cache->{r} = $r;
4456        }
4457        $cache->{data}->{$dir} = [ \%dirents, $r, $props ];
4458        wantarray ? (\%dirents, $r, $props) : \%dirents;
4459}
4460
4461sub DESTROY {
4462        # do not call the real DESTROY since we store ourselves in $RA
4463}
4464
4465# get_log(paths, start, end, limit,
4466#         discover_changed_paths, strict_node_history, receiver)
4467sub get_log {
4468        my ($self, @args) = @_;
4469        my $pool = SVN::Pool->new;
4470
4471        # svn_log_changed_path_t objects passed to get_log are likely to be
4472        # overwritten even if only the refs are copied to an external variable,
4473        # so we should dup the structures in their entirety.  Using an
4474        # externally passed pool (instead of our temporary and quickly cleared
4475        # pool in Git::SVN::Ra) does not help matters at all...
4476        my $receiver = pop @args;
4477        my $prefix = "/".$self->{svn_path};
4478        $prefix =~ s#/+($)##;
4479        my $prefix_regex = qr#^\Q$prefix\E#;
4480        push(@args, sub {
4481                my ($paths) = $_[0];
4482                return &$receiver(@_) unless $paths;
4483                $_[0] = ();
4484                foreach my $p (keys %$paths) {
4485                        my $i = $paths->{$p};
4486                        # Make path relative to our url, not repos_root
4487                        $p =~ s/$prefix_regex//;
4488                        my %s = map { $_ => $i->$_; }
4489                                qw/copyfrom_path copyfrom_rev action/;
4490                        if ($s{'copyfrom_path'}) {
4491                                $s{'copyfrom_path'} =~ s/$prefix_regex//;
4492                        }
4493                        $_[0]{$p} = \%s;
4494                }
4495                &$receiver(@_);
4496        });
4497
4498
4499        # the limit parameter was not supported in SVN 1.1.x, so we
4500        # drop it.  Therefore, the receiver callback passed to it
4501        # is made aware of this limitation by being wrapped if
4502        # the limit passed to is being wrapped.
4503        if ($SVN::Core::VERSION le '1.2.0') {
4504                my $limit = splice(@args, 3, 1);
4505                if ($limit > 0) {
4506                        my $receiver = pop @args;
4507                        push(@args, sub { &$receiver(@_) if (--$limit >= 0) });
4508                }
4509        }
4510        my $ret = $self->SUPER::get_log(@args, $pool);
4511        $pool->clear;
4512        $ret;
4513}
4514
4515sub trees_match {
4516        my ($self, $url1, $rev1, $url2, $rev2) = @_;
4517        my $ctx = SVN::Client->new(auth => _auth_providers);
4518        my $out = IO::File->new_tmpfile;
4519
4520        # older SVN (1.1.x) doesn't take $pool as the last parameter for
4521        # $ctx->diff(), so we'll create a default one
4522        my $pool = SVN::Pool->new_default_sub;
4523
4524        $ra_invalid = 1; # this will open a new SVN::Ra connection to $url1
4525        $ctx->diff([], $url1, $rev1, $url2, $rev2, 1, 1, 0, $out, $out);
4526        $out->flush;
4527        my $ret = (($out->stat)[7] == 0);
4528        close $out or croak $!;
4529
4530        $ret;
4531}
4532
4533sub get_commit_editor {
4534        my ($self, $log, $cb, $pool) = @_;
4535        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
4536        $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
4537}
4538
4539sub gs_do_update {
4540        my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
4541        my $new = ($rev_a == $rev_b);
4542        my $path = $gs->{path};
4543
4544        if ($new && -e $gs->{index}) {
4545                unlink $gs->{index} or die
4546                  "Couldn't unlink index: $gs->{index}: $!\n";
4547        }
4548        my $pool = SVN::Pool->new;
4549        $editor->set_path_strip($path);
4550        my (@pc) = split m#/#, $path;
4551        my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
4552                                        1, $editor, $pool);
4553        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
4554
4555        # Since we can't rely on svn_ra_reparent being available, we'll
4556        # just have to do some magic with set_path to make it so
4557        # we only want a partial path.
4558        my $sp = '';
4559        my $final = join('/', @pc);
4560        while (@pc) {
4561                $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
4562                $sp .= '/' if length $sp;
4563                $sp .= shift @pc;
4564        }
4565        die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
4566
4567        $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
4568
4569        $reporter->finish_report($pool);
4570        $pool->clear;
4571        $editor->{git_commit_ok};
4572}
4573
4574# this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
4575# svn_ra_reparent didn't work before 1.4)
4576sub gs_do_switch {
4577        my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
4578        my $path = $gs->{path};
4579        my $pool = SVN::Pool->new;
4580
4581        my $full_url = $self->{url};
4582        my $old_url = $full_url;
4583        $full_url .= '/' . $path if length $path;
4584        my ($ra, $reparented);
4585
4586        if ($old_url =~ m#^svn(\+ssh)?://# ||
4587            ($full_url =~ m#^https?://# &&
4588             escape_url($full_url) ne $full_url)) {
4589                $_[0] = undef;
4590                $self = undef;
4591                $RA = undef;
4592                $ra = Git::SVN::Ra->new($full_url);
4593                $ra_invalid = 1;
4594        } elsif ($old_url ne $full_url) {
4595                SVN::_Ra::svn_ra_reparent($self->{session}, $full_url, $pool);
4596                $self->{url} = $full_url;
4597                $reparented = 1;
4598        }
4599
4600        $ra ||= $self;
4601        $url_b = escape_url($url_b);
4602        my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
4603        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
4604        $reporter->set_path('', $rev_a, 0, @lock, $pool);
4605        $reporter->finish_report($pool);
4606
4607        if ($reparented) {
4608                SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
4609                $self->{url} = $old_url;
4610        }
4611
4612        $pool->clear;
4613        $editor->{git_commit_ok};
4614}
4615
4616sub longest_common_path {
4617        my ($gsv, $globs) = @_;
4618        my %common;
4619        my $common_max = scalar @$gsv;
4620
4621        foreach my $gs (@$gsv) {
4622                my @tmp = split m#/#, $gs->{path};
4623                my $p = '';
4624                foreach (@tmp) {
4625                        $p .= length($p) ? "/$_" : $_;
4626                        $common{$p} ||= 0;
4627                        $common{$p}++;
4628                }
4629        }
4630        $globs ||= [];
4631        $common_max += scalar @$globs;
4632        foreach my $glob (@$globs) {
4633                my @tmp = split m#/#, $glob->{path}->{left};
4634                my $p = '';
4635                foreach (@tmp) {
4636                        $p .= length($p) ? "/$_" : $_;
4637                        $common{$p} ||= 0;
4638                        $common{$p}++;
4639                }
4640        }
4641
4642        my $longest_path = '';
4643        foreach (sort {length $b <=> length $a} keys %common) {
4644                if ($common{$_} == $common_max) {
4645                        $longest_path = $_;
4646                        last;
4647                }
4648        }
4649        $longest_path;
4650}
4651
4652sub gs_fetch_loop_common {
4653        my ($self, $base, $head, $gsv, $globs) = @_;
4654        return if ($base > $head);
4655        my $inc = $_log_window_size;
4656        my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
4657        my $longest_path = longest_common_path($gsv, $globs);
4658        my $ra_url = $self->{url};
4659        my $find_trailing_edge;
4660        while (1) {
4661                my %revs;
4662                my $err;
4663                my $err_handler = $SVN::Error::handler;
4664                $SVN::Error::handler = sub {
4665                        ($err) = @_;
4666                        skip_unknown_revs($err);
4667                };
4668                sub _cb {
4669                        my ($paths, $r, $author, $date, $log) = @_;
4670                        [ $paths,
4671                          { author => $author, date => $date, log => $log } ];
4672                }
4673                $self->get_log([$longest_path], $min, $max, 0, 1, 1,
4674                               sub { $revs{$_[1]} = _cb(@_) });
4675                if ($err) {
4676                        print "Checked through r$max\r";
4677                } else {
4678                        $find_trailing_edge = 1;
4679                }
4680                if ($err and $find_trailing_edge) {
4681                        print STDERR "Path '$longest_path' ",
4682                                     "was probably deleted:\n",
4683                                     $err->expanded_message,
4684                                     "\nWill attempt to follow ",
4685                                     "revisions r$min .. r$max ",
4686                                     "committed before the deletion\n";
4687                        my $hi = $max;
4688                        while (--$hi >= $min) {
4689                                my $ok;
4690                                $self->get_log([$longest_path], $min, $hi,
4691                                               0, 1, 1, sub {
4692                                               $ok = $_[1];
4693                                               $revs{$_[1]} = _cb(@_) });
4694                                if ($ok) {
4695                                        print STDERR "r$min .. r$ok OK\n";
4696                                        last;
4697                                }
4698                        }
4699                        $find_trailing_edge = 0;
4700                }
4701                $SVN::Error::handler = $err_handler;
4702
4703                my %exists = map { $_->{path} => $_ } @$gsv;
4704                foreach my $r (sort {$a <=> $b} keys %revs) {
4705                        my ($paths, $logged) = @{$revs{$r}};
4706
4707                        foreach my $gs ($self->match_globs(\%exists, $paths,
4708                                                           $globs, $r)) {
4709                                if ($gs->rev_map_max >= $r) {
4710                                        next;
4711                                }
4712                                next unless $gs->match_paths($paths, $r);
4713                                $gs->{logged_rev_props} = $logged;
4714                                if (my $last_commit = $gs->last_commit) {
4715                                        $gs->assert_index_clean($last_commit);
4716                                }
4717                                my $log_entry = $gs->do_fetch($paths, $r);
4718                                if ($log_entry) {
4719                                        $gs->do_git_commit($log_entry);
4720                                }
4721                                $INDEX_FILES{$gs->{index}} = 1;
4722                        }
4723                        foreach my $g (@$globs) {
4724                                my $k = "svn-remote.$g->{remote}." .
4725                                        "$g->{t}-maxRev";
4726                                Git::SVN::tmp_config($k, $r);
4727                        }
4728                        if ($ra_invalid) {
4729                                $_[0] = undef;
4730                                $self = undef;
4731                                $RA = undef;
4732                                $self = Git::SVN::Ra->new($ra_url);
4733                                $ra_invalid = undef;
4734                        }
4735                }
4736                # pre-fill the .rev_db since it'll eventually get filled in
4737                # with '0' x40 if something new gets committed
4738                foreach my $gs (@$gsv) {
4739                        next if $gs->rev_map_max >= $max;
4740                        next if defined $gs->rev_map_get($max);
4741                        $gs->rev_map_set($max, 0 x40);
4742                }
4743                foreach my $g (@$globs) {
4744                        my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
4745                        Git::SVN::tmp_config($k, $max);
4746                }
4747                last if $max >= $head;
4748                $min = $max + 1;
4749                $max += $inc;
4750                $max = $head if ($max > $head);
4751        }
4752        Git::SVN::gc();
4753}
4754
4755sub get_dir_globbed {
4756        my ($self, $left, $depth, $r) = @_;
4757
4758        my @x = eval { $self->get_dir($left, $r) };
4759        return unless scalar @x == 3;
4760        my $dirents = $x[0];
4761        my @finalents;
4762        foreach my $de (keys %$dirents) {
4763                next if $dirents->{$de}->{kind} != $SVN::Node::dir;
4764                if ($depth > 1) {
4765                        my @args = ("$left/$de", $depth - 1, $r);
4766                        foreach my $dir ($self->get_dir_globbed(@args)) {
4767                                push @finalents, "$de/$dir";
4768                        }
4769                } else {
4770                        push @finalents, $de;
4771                }
4772        }
4773        @finalents;
4774}
4775
4776sub match_globs {
4777        my ($self, $exists, $paths, $globs, $r) = @_;
4778
4779        sub get_dir_check {
4780                my ($self, $exists, $g, $r) = @_;
4781
4782                my @dirs = $self->get_dir_globbed($g->{path}->{left},
4783                                                  $g->{path}->{depth},
4784                                                  $r);
4785
4786                foreach my $de (@dirs) {
4787                        my $p = $g->{path}->full_path($de);
4788                        next if $exists->{$p};
4789                        next if (length $g->{path}->{right} &&
4790                                 ($self->check_path($p, $r) !=
4791                                  $SVN::Node::dir));
4792                        $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
4793                                         $g->{ref}->full_path($de), 1);
4794                }
4795        }
4796        foreach my $g (@$globs) {
4797                if (my $path = $paths->{"/$g->{path}->{left}"}) {
4798                        if ($path->{action} =~ /^[AR]$/) {
4799                                get_dir_check($self, $exists, $g, $r);
4800                        }
4801                }
4802                foreach (keys %$paths) {
4803                        if (/$g->{path}->{left_regex}/ &&
4804                            !/$g->{path}->{regex}/) {
4805                                next if $paths->{$_}->{action} !~ /^[AR]$/;
4806                                get_dir_check($self, $exists, $g, $r);
4807                        }
4808                        next unless /$g->{path}->{regex}/;
4809                        my $p = $1;
4810                        my $pathname = $g->{path}->full_path($p);
4811                        next if $exists->{$pathname};
4812                        next if ($self->check_path($pathname, $r) !=
4813                                 $SVN::Node::dir);
4814                        $exists->{$pathname} = Git::SVN->init(
4815                                              $self->{url}, $pathname, undef,
4816                                              $g->{ref}->full_path($p), 1);
4817                }
4818                my $c = '';
4819                foreach (split m#/#, $g->{path}->{left}) {
4820                        $c .= "/$_";
4821                        next unless ($paths->{$c} &&
4822                                     ($paths->{$c}->{action} =~ /^[AR]$/));
4823                        get_dir_check($self, $exists, $g, $r);
4824                }
4825        }
4826        values %$exists;
4827}
4828
4829sub minimize_url {
4830        my ($self) = @_;
4831        return $self->{url} if ($self->{url} eq $self->{repos_root});
4832        my $url = $self->{repos_root};
4833        my @components = split(m!/!, $self->{svn_path});
4834        my $c = '';
4835        do {
4836                $url .= "/$c" if length $c;
4837                eval {
4838                        my $ra = (ref $self)->new($url);
4839                        my $latest = $ra->get_latest_revnum;
4840                        $ra->get_log("", $latest, 0, 1, 0, 1, sub {});
4841                };
4842        } while ($@ && ($c = shift @components));
4843        $url;
4844}
4845
4846sub can_do_switch {
4847        my $self = shift;
4848        unless (defined $can_do_switch) {
4849                my $pool = SVN::Pool->new;
4850                my $rep = eval {
4851                        $self->do_switch(1, '', 0, $self->{url},
4852                                         SVN::Delta::Editor->new, $pool);
4853                };
4854                if ($@) {
4855                        $can_do_switch = 0;
4856                } else {
4857                        $rep->abort_report($pool);
4858                        $can_do_switch = 1;
4859                }
4860                $pool->clear;
4861        }
4862        $can_do_switch;
4863}
4864
4865sub skip_unknown_revs {
4866        my ($err) = @_;
4867        my $errno = $err->apr_err();
4868        # Maybe the branch we're tracking didn't
4869        # exist when the repo started, so it's
4870        # not an error if it doesn't, just continue
4871        #
4872        # Wonderfully consistent library, eh?
4873        # 160013 - svn:// and file://
4874        # 175002 - http(s)://
4875        # 175007 - http(s):// (this repo required authorization, too...)
4876        #   More codes may be discovered later...
4877        if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
4878                my $err_key = $err->expanded_message;
4879                # revision numbers change every time, filter them out
4880                $err_key =~ s/\d+/\0/g;
4881                $err_key = "$errno\0$err_key";
4882                unless ($ignored_err{$err_key}) {
4883                        warn "W: Ignoring error from SVN, path probably ",
4884                             "does not exist: ($errno): ",
4885                             $err->expanded_message,"\n";
4886                        warn "W: Do not be alarmed at the above message ",
4887                             "git-svn is just searching aggressively for ",
4888                             "old history.\n",
4889                             "This may take a while on large repositories\n";
4890                        $ignored_err{$err_key} = 1;
4891                }
4892                return;
4893        }
4894        die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
4895}
4896
4897package Git::SVN::Log;
4898use strict;
4899use warnings;
4900use POSIX qw/strftime/;
4901use Time::Local;
4902use constant commit_log_separator => ('-' x 72) . "\n";
4903use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
4904            %rusers $show_commit $incremental/;
4905my $l_fmt;
4906
4907sub cmt_showable {
4908        my ($c) = @_;
4909        return 1 if defined $c->{r};
4910
4911        # big commit message got truncated by the 16k pretty buffer in rev-list
4912        if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
4913                                $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
4914                @{$c->{l}} = ();
4915                my @log = command(qw/cat-file commit/, $c->{c});
4916
4917                # shift off the headers
4918                shift @log while ($log[0] ne '');
4919                shift @log;
4920
4921                # TODO: make $c->{l} not have a trailing newline in the future
4922                @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
4923
4924                (undef, $c->{r}, undef) = ::extract_metadata(
4925                                (grep(/^git-svn-id: /, @log))[-1]);
4926        }
4927        return defined $c->{r};
4928}
4929
4930sub log_use_color {
4931        return $color || Git->repository->get_colorbool('color.diff');
4932}
4933
4934sub git_svn_log_cmd {
4935        my ($r_min, $r_max, @args) = @_;
4936        my $head = 'HEAD';
4937        my (@files, @log_opts);
4938        foreach my $x (@args) {
4939                if ($x eq '--' || @files) {
4940                        push @files, $x;
4941                } else {
4942                        if (::verify_ref("$x^0")) {
4943                                $head = $x;
4944                        } else {
4945                                push @log_opts, $x;
4946                        }
4947                }
4948        }
4949
4950        my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
4951        $gs ||= Git::SVN->_new;
4952        my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
4953                   $gs->refname);
4954        push @cmd, '-r' unless $non_recursive;
4955        push @cmd, qw/--raw --name-status/ if $verbose;
4956        push @cmd, '--color' if log_use_color();
4957        push @cmd, @log_opts;
4958        if (defined $r_max && $r_max == $r_min) {
4959                push @cmd, '--max-count=1';
4960                if (my $c = $gs->rev_map_get($r_max)) {
4961                        push @cmd, $c;
4962                }
4963        } elsif (defined $r_max) {
4964                if ($r_max < $r_min) {
4965                        ($r_min, $r_max) = ($r_max, $r_min);
4966                }
4967                my (undef, $c_max) = $gs->find_rev_before($r_max, 1, $r_min);
4968                my (undef, $c_min) = $gs->find_rev_after($r_min, 1, $r_max);
4969                # If there are no commits in the range, both $c_max and $c_min
4970                # will be undefined.  If there is at least 1 commit in the
4971                # range, both will be defined.
4972                return () if !defined $c_min || !defined $c_max;
4973                if ($c_min eq $c_max) {
4974                        push @cmd, '--max-count=1', $c_min;
4975                } else {
4976                        push @cmd, '--boundary', "$c_min..$c_max";
4977                }
4978        }
4979        return (@cmd, @files);
4980}
4981
4982# adapted from pager.c
4983sub config_pager {
4984        $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
4985        if (!defined $pager) {
4986                $pager = 'less';
4987        } elsif (length $pager == 0 || $pager eq 'cat') {
4988                $pager = undef;
4989        }
4990        $ENV{GIT_PAGER_IN_USE} = defined($pager);
4991}
4992
4993sub run_pager {
4994        return unless -t *STDOUT && defined $pager;
4995        pipe my ($rfd, $wfd) or return;
4996        defined(my $pid = fork) or ::fatal "Can't fork: $!";
4997        if (!$pid) {
4998                open STDOUT, '>&', $wfd or
4999                                     ::fatal "Can't redirect to stdout: $!";
5000                return;
5001        }
5002        open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!";
5003        $ENV{LESS} ||= 'FRSX';
5004        exec $pager or ::fatal "Can't run pager: $! ($pager)";
5005}
5006
5007sub format_svn_date {
5008        # some systmes don't handle or mishandle %z, so be creative.
5009        my $t = shift || time;
5010        my $gm = timelocal(gmtime($t));
5011        my $sign = qw( + + - )[ $t <=> $gm ];
5012        my $gmoff = sprintf("%s%02d%02d", $sign, (gmtime(abs($t - $gm)))[2,1]);
5013        return strftime("%Y-%m-%d %H:%M:%S $gmoff (%a, %d %b %Y)", localtime($t));
5014}
5015
5016sub parse_git_date {
5017        my ($t, $tz) = @_;
5018        # Date::Parse isn't in the standard Perl distro :(
5019        if ($tz =~ s/^\+//) {
5020                $t += tz_to_s_offset($tz);
5021        } elsif ($tz =~ s/^\-//) {
5022                $t -= tz_to_s_offset($tz);
5023        }
5024        return $t;
5025}
5026
5027sub set_local_timezone {
5028        if (defined $TZ) {
5029                $ENV{TZ} = $TZ;
5030        } else {
5031                delete $ENV{TZ};
5032        }
5033}
5034
5035sub tz_to_s_offset {
5036        my ($tz) = @_;
5037        $tz =~ s/(\d\d)$//;
5038        return ($1 * 60) + ($tz * 3600);
5039}
5040
5041sub get_author_info {
5042        my ($dest, $author, $t, $tz) = @_;
5043        $author =~ s/(?:^\s*|\s*$)//g;
5044        $dest->{a_raw} = $author;
5045        my $au;
5046        if ($::_authors) {
5047                $au = $rusers{$author} || undef;
5048        }
5049        if (!$au) {
5050                ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
5051        }
5052        $dest->{t} = $t;
5053        $dest->{tz} = $tz;
5054        $dest->{a} = $au;
5055        $dest->{t_utc} = parse_git_date($t, $tz);
5056}
5057
5058sub process_commit {
5059        my ($c, $r_min, $r_max, $defer) = @_;
5060        if (defined $r_min && defined $r_max) {
5061                if ($r_min == $c->{r} && $r_min == $r_max) {
5062                        show_commit($c);
5063                        return 0;
5064                }
5065                return 1 if $r_min == $r_max;
5066                if ($r_min < $r_max) {
5067                        # we need to reverse the print order
5068                        return 0 if (defined $limit && --$limit < 0);
5069                        push @$defer, $c;
5070                        return 1;
5071                }
5072                if ($r_min != $r_max) {
5073                        return 1 if ($r_min < $c->{r});
5074                        return 1 if ($r_max > $c->{r});
5075                }
5076        }
5077        return 0 if (defined $limit && --$limit < 0);
5078        show_commit($c);
5079        return 1;
5080}
5081
5082sub show_commit {
5083        my $c = shift;
5084        if ($oneline) {
5085                my $x = "\n";
5086                if (my $l = $c->{l}) {
5087                        while ($l->[0] =~ /^\s*$/) { shift @$l }
5088                        $x = $l->[0];
5089                }
5090                $l_fmt ||= 'A' . length($c->{r});
5091                print 'r',pack($l_fmt, $c->{r}),' | ';
5092                print "$c->{c} | " if $show_commit;
5093                print $x;
5094        } else {
5095                show_commit_normal($c);
5096        }
5097}
5098
5099sub show_commit_changed_paths {
5100        my ($c) = @_;
5101        return unless $c->{changed};
5102        print "Changed paths:\n", @{$c->{changed}};
5103}
5104
5105sub show_commit_normal {
5106        my ($c) = @_;
5107        print commit_log_separator, "r$c->{r} | ";
5108        print "$c->{c} | " if $show_commit;
5109        print "$c->{a} | ", format_svn_date($c->{t_utc}), ' | ';
5110        my $nr_line = 0;
5111
5112        if (my $l = $c->{l}) {
5113                while ($l->[$#$l] eq "\n" && $#$l > 0
5114                                          && $l->[($#$l - 1)] eq "\n") {
5115                        pop @$l;
5116                }
5117                $nr_line = scalar @$l;
5118                if (!$nr_line) {
5119                        print "1 line\n\n\n";
5120                } else {
5121                        if ($nr_line == 1) {
5122                                $nr_line = '1 line';
5123                        } else {
5124                                $nr_line .= ' lines';
5125                        }
5126                        print $nr_line, "\n";
5127                        show_commit_changed_paths($c);
5128                        print "\n";
5129                        print $_ foreach @$l;
5130                }
5131        } else {
5132                print "1 line\n";
5133                show_commit_changed_paths($c);
5134                print "\n";
5135
5136        }
5137        foreach my $x (qw/raw stat diff/) {
5138                if ($c->{$x}) {
5139                        print "\n";
5140                        print $_ foreach @{$c->{$x}}
5141                }
5142        }
5143}
5144
5145sub cmd_show_log {
5146        my (@args) = @_;
5147        my ($r_min, $r_max);
5148        my $r_last = -1; # prevent dupes
5149        set_local_timezone();
5150        if (defined $::_revision) {
5151                if ($::_revision =~ /^(\d+):(\d+)$/) {
5152                        ($r_min, $r_max) = ($1, $2);
5153                } elsif ($::_revision =~ /^\d+$/) {
5154                        $r_min = $r_max = $::_revision;
5155                } else {
5156                        ::fatal "-r$::_revision is not supported, use ",
5157                                "standard 'git log' arguments instead";
5158                }
5159        }
5160
5161        config_pager();
5162        @args = git_svn_log_cmd($r_min, $r_max, @args);
5163        if (!@args) {
5164                print commit_log_separator unless $incremental || $oneline;
5165                return;
5166        }
5167        my $log = command_output_pipe(@args);
5168        run_pager();
5169        my (@k, $c, $d, $stat);
5170        my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
5171        while (<$log>) {
5172                if (/^${esc_color}commit -?($::sha1_short)/o) {
5173                        my $cmt = $1;
5174                        if ($c && cmt_showable($c) && $c->{r} != $r_last) {
5175                                $r_last = $c->{r};
5176                                process_commit($c, $r_min, $r_max, \@k) or
5177                                                                goto out;
5178                        }
5179                        $d = undef;
5180                        $c = { c => $cmt };
5181                } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
5182                        get_author_info($c, $1, $2, $3);
5183                } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
5184                        # ignore
5185                } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
5186                        push @{$c->{raw}}, $_;
5187                } elsif (/^${esc_color}[ACRMDT]\t/) {
5188                        # we could add $SVN->{svn_path} here, but that requires
5189                        # remote access at the moment (repo_path_split)...
5190                        s#^(${esc_color})([ACRMDT])\t#$1   $2 #o;
5191                        push @{$c->{changed}}, $_;
5192                } elsif (/^${esc_color}diff /o) {
5193                        $d = 1;
5194                        push @{$c->{diff}}, $_;
5195                } elsif ($d) {
5196                        push @{$c->{diff}}, $_;
5197                } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
5198                          $esc_color*[\+\-]*$esc_color$/x) {
5199                        $stat = 1;
5200                        push @{$c->{stat}}, $_;
5201                } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
5202                        push @{$c->{stat}}, $_;
5203                        $stat = undef;
5204                } elsif (/^${esc_color}    (git-svn-id:.+)$/o) {
5205                        ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
5206                } elsif (s/^${esc_color}    //o) {
5207                        push @{$c->{l}}, $_;
5208                }
5209        }
5210        if ($c && defined $c->{r} && $c->{r} != $r_last) {
5211                $r_last = $c->{r};
5212                process_commit($c, $r_min, $r_max, \@k);
5213        }
5214        if (@k) {
5215                ($r_min, $r_max) = ($r_max, $r_min);
5216                process_commit($_, $r_min, $r_max) foreach reverse @k;
5217        }
5218out:
5219        close $log;
5220        print commit_log_separator unless $incremental || $oneline;
5221}
5222
5223sub cmd_blame {
5224        my $path = pop;
5225
5226        config_pager();
5227        run_pager();
5228
5229        my ($fh, $ctx, $rev);
5230
5231        if ($_git_format) {
5232                ($fh, $ctx) = command_output_pipe('blame', @_, $path);
5233                while (my $line = <$fh>) {
5234                        if ($line =~ /^\^?([[:xdigit:]]+)\s/) {
5235                                # Uncommitted edits show up as a rev ID of
5236                                # all zeros, which we can't look up with
5237                                # cmt_metadata
5238                                if ($1 !~ /^0+$/) {
5239                                        (undef, $rev, undef) =
5240                                                ::cmt_metadata($1);
5241                                        $rev = '0' if (!$rev);
5242                                } else {
5243                                        $rev = '0';
5244                                }
5245                                $rev = sprintf('%-10s', $rev);
5246                                $line =~ s/^\^?[[:xdigit:]]+(\s)/$rev$1/;
5247                        }
5248                        print $line;
5249                }
5250        } else {
5251                ($fh, $ctx) = command_output_pipe('blame', '-p', @_, 'HEAD',
5252                                                  '--', $path);
5253                my ($sha1);
5254                my %authors;
5255                my @buffer;
5256                my %dsha; #distinct sha keys
5257
5258                while (my $line = <$fh>) {
5259                        push @buffer, $line;
5260                        if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5261                                $dsha{$1} = 1;
5262                        }
5263                }
5264
5265                my $s2r = ::cmt_sha2rev_batch([keys %dsha]);
5266
5267                foreach my $line (@buffer) {
5268                        if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5269                                $rev = $s2r->{$1};
5270                                $rev = '0' if (!$rev)
5271                        }
5272                        elsif ($line =~ /^author (.*)/) {
5273                                $authors{$rev} = $1;
5274                                $authors{$rev} =~ s/\s/_/g;
5275                        }
5276                        elsif ($line =~ /^\t(.*)$/) {
5277                                printf("%6s %10s %s\n", $rev, $authors{$rev}, $1);
5278                        }
5279                }
5280        }
5281        command_close_pipe($fh, $ctx);
5282}
5283
5284package Git::SVN::Migration;
5285# these version numbers do NOT correspond to actual version numbers
5286# of git nor git-svn.  They are just relative.
5287#
5288# v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
5289#
5290# v1 layout: .git/$id/info/url, refs/remotes/$id
5291#
5292# v2 layout: .git/svn/$id/info/url, refs/remotes/$id
5293#
5294# v3 layout: .git/svn/$id, refs/remotes/$id
5295#            - info/url may remain for backwards compatibility
5296#            - this is what we migrate up to this layout automatically,
5297#            - this will be used by git svn init on single branches
5298# v3.1 layout (auto migrated):
5299#            - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
5300#              for backwards compatibility
5301#
5302# v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
5303#            - this is only created for newly multi-init-ed
5304#              repositories.  Similar in spirit to the
5305#              --use-separate-remotes option in git-clone (now default)
5306#            - we do not automatically migrate to this (following
5307#              the example set by core git)
5308#
5309# v5 layout: .rev_db.$UUID => .rev_map.$UUID
5310#            - newer, more-efficient format that uses 24-bytes per record
5311#              with no filler space.
5312#            - use xxd -c24 < .rev_map.$UUID to view and debug
5313#            - This is a one-way migration, repositories updated to the
5314#              new format will not be able to use old git-svn without
5315#              rebuilding the .rev_db.  Rebuilding the rev_db is not
5316#              possible if noMetadata or useSvmProps are set; but should
5317#              be no problem for users that use the (sensible) defaults.
5318use strict;
5319use warnings;
5320use Carp qw/croak/;
5321use File::Path qw/mkpath/;
5322use File::Basename qw/dirname basename/;
5323use vars qw/$_minimize/;
5324
5325sub migrate_from_v0 {
5326        my $git_dir = $ENV{GIT_DIR};
5327        return undef unless -d $git_dir;
5328        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5329        my $migrated = 0;
5330        while (<$fh>) {
5331                chomp;
5332                my ($id, $orig_ref) = ($_, $_);
5333                next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
5334                next unless -f "$git_dir/$id/info/url";
5335                my $new_ref = "refs/remotes/$id";
5336                if (::verify_ref("$new_ref^0")) {
5337                        print STDERR "W: $orig_ref is probably an old ",
5338                                     "branch used by an ancient version of ",
5339                                     "git-svn.\n",
5340                                     "However, $new_ref also exists.\n",
5341                                     "We will not be able ",
5342                                     "to use this branch until this ",
5343                                     "ambiguity is resolved.\n";
5344                        next;
5345                }
5346                print STDERR "Migrating from v0 layout...\n" if !$migrated;
5347                print STDERR "Renaming ref: $orig_ref => $new_ref\n";
5348                command_noisy('update-ref', $new_ref, $orig_ref);
5349                command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
5350                $migrated++;
5351        }
5352        command_close_pipe($fh, $ctx);
5353        print STDERR "Done migrating from v0 layout...\n" if $migrated;
5354        $migrated;
5355}
5356
5357sub migrate_from_v1 {
5358        my $git_dir = $ENV{GIT_DIR};
5359        my $migrated = 0;
5360        return $migrated unless -d $git_dir;
5361        my $svn_dir = "$git_dir/svn";
5362
5363        # just in case somebody used 'svn' as their $id at some point...
5364        return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
5365
5366        print STDERR "Migrating from a git-svn v1 layout...\n";
5367        mkpath([$svn_dir]);
5368        print STDERR "Data from a previous version of git-svn exists, but\n\t",
5369                     "$svn_dir\n\t(required for this version ",
5370                     "($::VERSION) of git-svn) does not exist.\n";
5371        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5372        while (<$fh>) {
5373                my $x = $_;
5374                next unless $x =~ s#^refs/remotes/##;
5375                chomp $x;
5376                next unless -f "$git_dir/$x/info/url";
5377                my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
5378                next unless $u;
5379                my $dn = dirname("$git_dir/svn/$x");
5380                mkpath([$dn]) unless -d $dn;
5381                if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
5382                        mkpath(["$git_dir/svn/svn"]);
5383                        print STDERR " - $git_dir/$x/info => ",
5384                                        "$git_dir/svn/$x/info\n";
5385                        rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
5386                               croak "$!: $x";
5387                        # don't worry too much about these, they probably
5388                        # don't exist with repos this old (save for index,
5389                        # and we can easily regenerate that)
5390                        foreach my $f (qw/unhandled.log index .rev_db/) {
5391                                rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
5392                        }
5393                } else {
5394                        print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
5395                        rename "$git_dir/$x", "$git_dir/svn/$x" or
5396                               croak "$!: $x";
5397                }
5398                $migrated++;
5399        }
5400        command_close_pipe($fh, $ctx);
5401        print STDERR "Done migrating from a git-svn v1 layout\n";
5402        $migrated;
5403}
5404
5405sub read_old_urls {
5406        my ($l_map, $pfx, $path) = @_;
5407        my @dir;
5408        foreach (<$path/*>) {
5409                if (-r "$_/info/url") {
5410                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
5411                        my $ref_id = $pfx . basename $_;
5412                        my $url = ::file_to_s("$_/info/url");
5413                        $l_map->{$ref_id} = $url;
5414                } elsif (-d $_) {
5415                        push @dir, $_;
5416                }
5417        }
5418        foreach (@dir) {
5419                my $x = $_;
5420                $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
5421                read_old_urls($l_map, $x, $_);
5422        }
5423}
5424
5425sub migrate_from_v2 {
5426        my @cfg = command(qw/config -l/);
5427        return if grep /^svn-remote\..+\.url=/, @cfg;
5428        my %l_map;
5429        read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
5430        my $migrated = 0;
5431
5432        foreach my $ref_id (sort keys %l_map) {
5433                eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
5434                if ($@) {
5435                        Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
5436                }
5437                $migrated++;
5438        }
5439        $migrated;
5440}
5441
5442sub minimize_connections {
5443        my $r = Git::SVN::read_all_remotes();
5444        my $new_urls = {};
5445        my $root_repos = {};
5446        foreach my $repo_id (keys %$r) {
5447                my $url = $r->{$repo_id}->{url} or next;
5448                my $fetch = $r->{$repo_id}->{fetch} or next;
5449                my $ra = Git::SVN::Ra->new($url);
5450
5451                # skip existing cases where we already connect to the root
5452                if (($ra->{url} eq $ra->{repos_root}) ||
5453                    ($ra->{repos_root} eq $repo_id)) {
5454                        $root_repos->{$ra->{url}} = $repo_id;
5455                        next;
5456                }
5457
5458                my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
5459                my $root_path = $ra->{url};
5460                $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
5461                foreach my $path (keys %$fetch) {
5462                        my $ref_id = $fetch->{$path};
5463                        my $gs = Git::SVN->new($ref_id, $repo_id, $path);
5464
5465                        # make sure we can read when connecting to
5466                        # a higher level of a repository
5467                        my ($last_rev, undef) = $gs->last_rev_commit;
5468                        if (!defined $last_rev) {
5469                                $last_rev = eval {
5470                                        $root_ra->get_latest_revnum;
5471                                };
5472                                next if $@;
5473                        }
5474                        my $new = $root_path;
5475                        $new .= length $path ? "/$path" : '';
5476                        eval {
5477                                $root_ra->get_log([$new], $last_rev, $last_rev,
5478                                                  0, 0, 1, sub { });
5479                        };
5480                        next if $@;
5481                        $new_urls->{$ra->{repos_root}}->{$new} =
5482                                { ref_id => $ref_id,
5483                                  old_repo_id => $repo_id,
5484                                  old_path => $path };
5485                }
5486        }
5487
5488        my @emptied;
5489        foreach my $url (keys %$new_urls) {
5490                # see if we can re-use an existing [svn-remote "repo_id"]
5491                # instead of creating a(n ugly) new section:
5492                my $repo_id = $root_repos->{$url} || $url;
5493
5494                my $fetch = $new_urls->{$url};
5495                foreach my $path (keys %$fetch) {
5496                        my $x = $fetch->{$path};
5497                        Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
5498                        my $pfx = "svn-remote.$x->{old_repo_id}";
5499
5500                        my $old_fetch = quotemeta("$x->{old_path}:".
5501                                                  "refs/remotes/$x->{ref_id}");
5502                        command_noisy(qw/config --unset/,
5503                                      "$pfx.fetch", '^'. $old_fetch . '$');
5504                        delete $r->{$x->{old_repo_id}}->
5505                               {fetch}->{$x->{old_path}};
5506                        if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
5507                                command_noisy(qw/config --unset/,
5508                                              "$pfx.url");
5509                                push @emptied, $x->{old_repo_id}
5510                        }
5511                }
5512        }
5513        if (@emptied) {
5514                my $file = $ENV{GIT_CONFIG} || "$ENV{GIT_DIR}/config";
5515                print STDERR <<EOF;
5516The following [svn-remote] sections in your config file ($file) are empty
5517and can be safely removed:
5518EOF
5519                print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
5520        }
5521}
5522
5523sub migration_check {
5524        migrate_from_v0();
5525        migrate_from_v1();
5526        migrate_from_v2();
5527        minimize_connections() if $_minimize;
5528}
5529
5530package Git::IndexInfo;
5531use strict;
5532use warnings;
5533use Git qw/command_input_pipe command_close_pipe/;
5534
5535sub new {
5536        my ($class) = @_;
5537        my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
5538        bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
5539}
5540
5541sub remove {
5542        my ($self, $path) = @_;
5543        if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
5544                return ++$self->{nr};
5545        }
5546        undef;
5547}
5548
5549sub update {
5550        my ($self, $mode, $hash, $path) = @_;
5551        if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
5552                return ++$self->{nr};
5553        }
5554        undef;
5555}
5556
5557sub DESTROY {
5558        my ($self) = @_;
5559        command_close_pipe($self->{gui}, $self->{ctx});
5560}
5561
5562package Git::SVN::GlobSpec;
5563use strict;
5564use warnings;
5565
5566sub new {
5567        my ($class, $glob) = @_;
5568        my $re = $glob;
5569        $re =~ s!/+$!!g; # no need for trailing slashes
5570        $re =~ m!^([^*]*)(\*(?:/\*)*)([^*]*)$!;
5571        my $temp = $re;
5572        my ($left, $right) = ($1, $3);
5573        $re = $2;
5574        my $depth = $re =~ tr/*/*/;
5575        if ($depth != $temp =~ tr/*/*/) {
5576                die "Only one set of wildcard directories " .
5577                        "(e.g. '*' or '*/*/*') is supported: '$glob'\n";
5578        }
5579        if ($depth == 0) {
5580                die "One '*' is needed for glob: '$glob'\n";
5581        }
5582        $re =~ s!\*!\[^/\]*!g;
5583        $re = quotemeta($left) . "($re)" . quotemeta($right);
5584        if (length $left && !($left =~ s!/+$!!g)) {
5585                die "Missing trailing '/' on left side of: '$glob' ($left)\n";
5586        }
5587        if (length $right && !($right =~ s!^/+!!g)) {
5588                die "Missing leading '/' on right side of: '$glob' ($right)\n";
5589        }
5590        my $left_re = qr/^\/\Q$left\E(\/|$)/;
5591        bless { left => $left, right => $right, left_regex => $left_re,
5592                regex => qr/$re/, glob => $glob, depth => $depth }, $class;
5593}
5594
5595sub full_path {
5596        my ($self, $path) = @_;
5597        return (length $self->{left} ? "$self->{left}/" : '') .
5598               $path . (length $self->{right} ? "/$self->{right}" : '');
5599}
5600
5601__END__
5602
5603Data structures:
5604
5605
5606$remotes = { # returned by read_all_remotes()
5607        'svn' => {
5608                # svn-remote.svn.url=https://svn.musicpd.org
5609                url => 'https://svn.musicpd.org',
5610                # svn-remote.svn.fetch=mpd/trunk:trunk
5611                fetch => {
5612                        'mpd/trunk' => 'trunk',
5613                },
5614                # svn-remote.svn.tags=mpd/tags/*:tags/*
5615                tags => {
5616                        path => {
5617                                left => 'mpd/tags',
5618                                right => '',
5619                                regex => qr!mpd/tags/([^/]+)$!,
5620                                glob => 'tags/*',
5621                        },
5622                        ref => {
5623                                left => 'tags',
5624                                right => '',
5625                                regex => qr!tags/([^/]+)$!,
5626                                glob => 'tags/*',
5627                        },
5628                }
5629        }
5630};
5631
5632$log_entry hashref as returned by libsvn_log_entry()
5633{
5634        log => 'whitespace-formatted log entry
5635',                                              # trailing newline is preserved
5636        revision => '8',                        # integer
5637        date => '2004-02-24T17:01:44.108345Z',  # commit date
5638        author => 'committer name'
5639};
5640
5641
5642# this is generated by generate_diff();
5643@mods = array of diff-index line hashes, each element represents one line
5644        of diff-index output
5645
5646diff-index line ($m hash)
5647{
5648        mode_a => first column of diff-index output, no leading ':',
5649        mode_b => second column of diff-index output,
5650        sha1_b => sha1sum of the final blob,
5651        chg => change type [MCRADT],
5652        file_a => original file name of a file (iff chg is 'C' or 'R')
5653        file_b => new/current file name of a file (any chg)
5654}
5655;
5656
5657# retval of read_url_paths{,_all}();
5658$l_map = {
5659        # repository root url
5660        'https://svn.musicpd.org' => {
5661                # repository path               # GIT_SVN_ID
5662                'mpd/trunk'             =>      'trunk',
5663                'mpd/tags/0.11.5'       =>      'tags/0.11.5',
5664        },
5665}
5666
5667Notes:
5668        I don't trust the each() function on unless I created %hash myself
5669        because the internal iterator may not have started at base.