Andrew's git - gitweb.git/blob

   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
  24if (! exists $ENV{SVN_SSH}) {
  25        if (exists $ENV{GIT_SSH}) {
  26                $ENV{SVN_SSH} = $ENV{GIT_SSH};
  27                if ($^O eq 'msys') {
  28                        $ENV{SVN_SSH} =~ s/\\/\\\\/g;
  29                }
  30        }
  31}
  32
  33$Git::SVN::Log::TZ = $ENV{TZ};
  34$ENV{TZ} = 'UTC';
  35$| = 1; # unbuffer STDOUT
  36
  37sub fatal (@) { print STDERR "@_\n"; exit 1 }
  38require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
  39require SVN::Ra;
  40require SVN::Delta;
  41if ($SVN::Core::VERSION lt '1.1.0') {
  42        fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)";
  43}
  44my $can_compress = eval { require Compress::Zlib; 1};
  45push @Git::SVN::Ra::ISA, 'SVN::Ra';
  46push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
  47push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
  48use Carp qw/croak/;
  49use Digest::MD5;
  50use IO::File qw//;
  51use File::Basename qw/dirname basename/;
  52use File::Path qw/mkpath/;
  53use File::Spec;
  54use File::Find;
  55use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev/;
  56use IPC::Open3;
  57use Git;
  58
  59BEGIN {
  60        # import functions from Git into our packages, en masse
  61        no strict 'refs';
  62        foreach (qw/command command_oneline command_noisy command_output_pipe
  63                    command_input_pipe command_close_pipe
  64                    command_bidi_pipe command_close_bidi_pipe/) {
  65                for my $package ( qw(SVN::Git::Editor SVN::Git::Fetcher
  66                        Git::SVN::Migration Git::SVN::Log Git::SVN),
  67                        __PACKAGE__) {
  68                        *{"${package}::$_"} = \&{"Git::$_"};
  69                }
  70        }
  71}
  72
  73my ($SVN);
  74
  75$sha1 = qr/[a-f\d]{40}/;
  76$sha1_short = qr/[a-f\d]{4,40}/;
  77my ($_stdin, $_help, $_edit,
  78        $_message, $_file, $_branch_dest,
  79        $_template, $_shared,
  80        $_version, $_fetch_all, $_no_rebase, $_fetch_parent,
  81        $_merge, $_strategy, $_dry_run, $_local,
  82        $_prefix, $_no_checkout, $_url, $_verbose,
  83        $_git_format, $_commit_url, $_tag);
  84$Git::SVN::_follow_parent = 1;
  85$_q ||= 0;
  86my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
  87                    'config-dir=s' => \$Git::SVN::Ra::config_dir,
  88                    'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache,
  89                    'ignore-paths=s' => \$SVN::Git::Fetcher::_ignore_regex );
  90my %fc_opts = ( 'follow-parent|follow!' => \$Git::SVN::_follow_parent,
  91                'authors-file|A=s' => \$_authors,
  92                'authors-prog=s' => \$_authors_prog,
  93                'repack:i' => \$Git::SVN::_repack,
  94                'noMetadata' => \$Git::SVN::_no_metadata,
  95                'useSvmProps' => \$Git::SVN::_use_svm_props,
  96                'useSvnsyncProps' => \$Git::SVN::_use_svnsync_props,
  97                'log-window-size=i' => \$Git::SVN::Ra::_log_window_size,
  98                'no-checkout' => \$_no_checkout,
  99                'quiet|q+' => \$_q,
 100                'repack-flags|repack-args|repack-opts=s' =>
 101                   \$Git::SVN::_repack_flags,
 102                'use-log-author' => \$Git::SVN::_use_log_author,
 103                'add-author-from' => \$Git::SVN::_add_author_from,
 104                'localtime' => \$Git::SVN::_localtime,
 105                %remote_opts );
 106
 107my ($_trunk, @_tags, @_branches, $_stdlayout);
 108my %icv;
 109my %init_opts = ( 'template=s' => \$_template, 'shared:s' => \$_shared,
 110                  'trunk|T=s' => \$_trunk, 'tags|t=s@' => \@_tags,
 111                  'branches|b=s@' => \@_branches, 'prefix=s' => \$_prefix,
 112                  'stdlayout|s' => \$_stdlayout,
 113                  'minimize-url|m!' => \$Git::SVN::_minimize_url,
 114                  'no-metadata' => sub { $icv{noMetadata} = 1 },
 115                  'use-svm-props' => sub { $icv{useSvmProps} = 1 },
 116                  'use-svnsync-props' => sub { $icv{useSvnsyncProps} = 1 },
 117                  'rewrite-root=s' => sub { $icv{rewriteRoot} = $_[1] },
 118                  %remote_opts );
 119my %cmt_opts = ( 'edit|e' => \$_edit,
 120                'rmdir' => \$SVN::Git::Editor::_rmdir,
 121                'find-copies-harder' => \$SVN::Git::Editor::_find_copies_harder,
 122                'l=i' => \$SVN::Git::Editor::_rename_limit,
 123                'copy-similarity|C=i'=> \$SVN::Git::Editor::_cp_similarity
 124);
 125
 126my %cmd = (
 127        fetch => [ \&cmd_fetch, "Download new revisions from SVN",
 128                        { 'revision|r=s' => \$_revision,
 129                          'fetch-all|all' => \$_fetch_all,
 130                          'parent|p' => \$_fetch_parent,
 131                           %fc_opts } ],
 132        clone => [ \&cmd_clone, "Initialize and fetch revisions",
 133                        { 'revision|r=s' => \$_revision,
 134                           %fc_opts, %init_opts } ],
 135        init => [ \&cmd_init, "Initialize a repo for tracking" .
 136                          " (requires URL argument)",
 137                          \%init_opts ],
 138        'multi-init' => [ \&cmd_multi_init,
 139                          "Deprecated alias for ".
 140                          "'$0 init -T<trunk> -b<branches> -t<tags>'",
 141                          \%init_opts ],
 142        dcommit => [ \&cmd_dcommit,
 143                     'Commit several diffs to merge with upstream',
 144                        { 'merge|m|M' => \$_merge,
 145                          'strategy|s=s' => \$_strategy,
 146                          'verbose|v' => \$_verbose,
 147                          'dry-run|n' => \$_dry_run,
 148                          'fetch-all|all' => \$_fetch_all,
 149                          'commit-url=s' => \$_commit_url,
 150                          'revision|r=i' => \$_revision,
 151                          'no-rebase' => \$_no_rebase,
 152                        %cmt_opts, %fc_opts } ],
 153        branch => [ \&cmd_branch,
 154                    'Create a branch in the SVN repository',
 155                    { 'message|m=s' => \$_message,
 156                      'destination|d=s' => \$_branch_dest,
 157                      'dry-run|n' => \$_dry_run,
 158                      'tag|t' => \$_tag } ],
 159        tag => [ sub { $_tag = 1; cmd_branch(@_) },
 160                 'Create a tag in the SVN repository',
 161                 { 'message|m=s' => \$_message,
 162                   'destination|d=s' => \$_branch_dest,
 163                   'dry-run|n' => \$_dry_run } ],
 164        'set-tree' => [ \&cmd_set_tree,
 165                        "Set an SVN repository to a git tree-ish",
 166                        { 'stdin' => \$_stdin, %cmt_opts, %fc_opts, } ],
 167        'create-ignore' => [ \&cmd_create_ignore,
 168                             'Create a .gitignore per svn:ignore',
 169                             { 'revision|r=i' => \$_revision
 170                             } ],
 171        'mkdirs' => [ \&cmd_mkdirs ,
 172                      "recreate empty directories after a checkout",
 173                      { 'revision|r=i' => \$_revision } ],
 174        'propget' => [ \&cmd_propget,
 175                       'Print the value of a property on a file or directory',
 176                       { 'revision|r=i' => \$_revision } ],
 177        'proplist' => [ \&cmd_proplist,
 178                       'List all properties of a file or directory',
 179                       { 'revision|r=i' => \$_revision } ],
 180        'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
 181                        { 'revision|r=i' => \$_revision
 182                        } ],
 183        'show-externals' => [ \&cmd_show_externals, "Show svn:externals listings",
 184                        { 'revision|r=i' => \$_revision
 185                        } ],
 186        'multi-fetch' => [ \&cmd_multi_fetch,
 187                           "Deprecated alias for $0 fetch --all",
 188                           { 'revision|r=s' => \$_revision, %fc_opts } ],
 189        'migrate' => [ sub { },
 190                       # no-op, we automatically run this anyways,
 191                       'Migrate configuration/metadata/layout from
 192                        previous versions of git-svn',
 193                       { 'minimize' => \$Git::SVN::Migration::_minimize,
 194                         %remote_opts } ],
 195        'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
 196                        { 'limit=i' => \$Git::SVN::Log::limit,
 197                          'revision|r=s' => \$_revision,
 198                          'verbose|v' => \$Git::SVN::Log::verbose,
 199                          'incremental' => \$Git::SVN::Log::incremental,
 200                          'oneline' => \$Git::SVN::Log::oneline,
 201                          'show-commit' => \$Git::SVN::Log::show_commit,
 202                          'non-recursive' => \$Git::SVN::Log::non_recursive,
 203                          'authors-file|A=s' => \$_authors,
 204                          'color' => \$Git::SVN::Log::color,
 205                          'pager=s' => \$Git::SVN::Log::pager
 206                        } ],
 207        'find-rev' => [ \&cmd_find_rev,
 208                        "Translate between SVN revision numbers and tree-ish",
 209                        {} ],
 210        'rebase' => [ \&cmd_rebase, "Fetch and rebase your working directory",
 211                        { 'merge|m|M' => \$_merge,
 212                          'verbose|v' => \$_verbose,
 213                          'strategy|s=s' => \$_strategy,
 214                          'local|l' => \$_local,
 215                          'fetch-all|all' => \$_fetch_all,
 216                          'dry-run|n' => \$_dry_run,
 217                          %fc_opts } ],
 218        'commit-diff' => [ \&cmd_commit_diff,
 219                           'Commit a diff between two trees',
 220                        { 'message|m=s' => \$_message,
 221                          'file|F=s' => \$_file,
 222                          'revision|r=s' => \$_revision,
 223                        %cmt_opts } ],
 224        'info' => [ \&cmd_info,
 225                    "Show info about the latest SVN revision
 226                     on the current branch",
 227                    { 'url' => \$_url, } ],
 228        'blame' => [ \&Git::SVN::Log::cmd_blame,
 229                    "Show what revision and author last modified each line of a file",
 230                    { 'git-format' => \$_git_format } ],
 231        'reset' => [ \&cmd_reset,
 232                     "Undo fetches back to the specified SVN revision",
 233                     { 'revision|r=s' => \$_revision,
 234                       'parent|p' => \$_fetch_parent } ],
 235        'gc' => [ \&cmd_gc,
 236                  "Compress unhandled.log files in .git/svn and remove " .
 237                  "index files in .git/svn",
 238                {} ],
 239);
 240
 241my $cmd;
 242for (my $i = 0; $i < @ARGV; $i++) {
 243        if (defined $cmd{$ARGV[$i]}) {
 244                $cmd = $ARGV[$i];
 245                splice @ARGV, $i, 1;
 246                last;
 247        } elsif ($ARGV[$i] eq 'help') {
 248                $cmd = $ARGV[$i+1];
 249                usage(0);
 250        }
 251};
 252
 253# make sure we're always running at the top-level working directory
 254unless ($cmd && $cmd =~ /(?:clone|init|multi-init)$/) {
 255        unless (-d $ENV{GIT_DIR}) {
 256                if ($git_dir_user_set) {
 257                        die "GIT_DIR=$ENV{GIT_DIR} explicitly set, ",
 258                            "but it is not a directory\n";
 259                }
 260                my $git_dir = delete $ENV{GIT_DIR};
 261                my $cdup = undef;
 262                git_cmd_try {
 263                        $cdup = command_oneline(qw/rev-parse --show-cdup/);
 264                        $git_dir = '.' unless ($cdup);
 265                        chomp $cdup if ($cdup);
 266                        $cdup = "." unless ($cdup && length $cdup);
 267                } "Already at toplevel, but $git_dir not found\n";
 268                chdir $cdup or die "Unable to chdir up to '$cdup'\n";
 269                unless (-d $git_dir) {
 270                        die "$git_dir still not found after going to ",
 271                            "'$cdup'\n";
 272                }
 273                $ENV{GIT_DIR} = $git_dir;
 274        }
 275        $_repository = Git->repository(Repository => $ENV{GIT_DIR});
 276}
 277
 278my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
 279
 280read_git_config(\%opts);
 281if ($cmd && ($cmd eq 'log' || $cmd eq 'blame')) {
 282        Getopt::Long::Configure('pass_through');
 283}
 284my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
 285                    'minimize-connections' => \$Git::SVN::Migration::_minimize,
 286                    'id|i=s' => \$Git::SVN::default_ref_id,
 287                    'svn-remote|remote|R=s' => sub {
 288                       $Git::SVN::no_reuse_existing = 1;
 289                       $Git::SVN::default_repo_id = $_[1] });
 290exit 1 if (!$rv && $cmd && $cmd ne 'log');
 291
 292usage(0) if $_help;
 293version() if $_version;
 294usage(1) unless defined $cmd;
 295load_authors() if $_authors;
 296if (defined $_authors_prog) {
 297        $_authors_prog = "'" . File::Spec->rel2abs($_authors_prog) . "'";
 298}
 299
 300unless ($cmd =~ /^(?:clone|init|multi-init|commit-diff)$/) {
 301        Git::SVN::Migration::migration_check();
 302}
 303Git::SVN::init_vars();
 304eval {
 305        Git::SVN::verify_remotes_sanity();
 306        $cmd{$cmd}->[0]->(@ARGV);
 307};
 308fatal $@ if $@;
 309post_fetch_checkout();
 310exit 0;
 311
 312####################### primary functions ######################
 313sub usage {
 314        my $exit = shift || 0;
 315        my $fd = $exit ? \*STDERR : \*STDOUT;
 316        print $fd <<"";
 317git-svn - bidirectional operations between a single Subversion tree and git
 318Usage: git svn <command> [options] [arguments]\n
 319
 320        print $fd "Available commands:\n" unless $cmd;
 321
 322        foreach (sort keys %cmd) {
 323                next if $cmd && $cmd ne $_;
 324                next if /^multi-/; # don't show deprecated commands
 325                print $fd '  ',pack('A17',$_),$cmd{$_}->[1],"\n";
 326                foreach (sort keys %{$cmd{$_}->[2]}) {
 327                        # mixed-case options are for .git/config only
 328                        next if /[A-Z]/ && /^[a-z]+$/i;
 329                        # prints out arguments as they should be passed:
 330                        my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
 331                        print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
 332                                                        "--$_" : "-$_" }
 333                                                split /\|/,$_)," $x\n";
 334                }
 335        }
 336        print $fd <<"";
 337\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
 338arbitrary identifier if you're tracking multiple SVN branches/repositories in
 339one git repository and want to keep them separate.  See git-svn(1) for more
 340information.
 341
 342        exit $exit;
 343}
 344
 345sub version {
 346        print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
 347        exit 0;
 348}
 349
 350sub do_git_init_db {
 351        unless (-d $ENV{GIT_DIR}) {
 352                my @init_db = ('init');
 353                push @init_db, "--template=$_template" if defined $_template;
 354                if (defined $_shared) {
 355                        if ($_shared =~ /[a-z]/) {
 356                                push @init_db, "--shared=$_shared";
 357                        } else {
 358                                push @init_db, "--shared";
 359                        }
 360                }
 361                command_noisy(@init_db);
 362                $_repository = Git->repository(Repository => ".git");
 363        }
 364        command_noisy('config', 'core.autocrlf', 'false');
 365        my $set;
 366        my $pfx = "svn-remote.$Git::SVN::default_repo_id";
 367        foreach my $i (keys %icv) {
 368                die "'$set' and '$i' cannot both be set\n" if $set;
 369                next unless defined $icv{$i};
 370                command_noisy('config', "$pfx.$i", $icv{$i});
 371                $set = $i;
 372        }
 373        my $ignore_regex = \$SVN::Git::Fetcher::_ignore_regex;
 374        command_noisy('config', "$pfx.ignore-paths", $$ignore_regex)
 375                if defined $$ignore_regex;
 376}
 377
 378sub init_subdir {
 379        my $repo_path = shift or return;
 380        mkpath([$repo_path]) unless -d $repo_path;
 381        chdir $repo_path or die "Couldn't chdir to $repo_path: $!\n";
 382        $ENV{GIT_DIR} = '.git';
 383        $_repository = Git->repository(Repository => $ENV{GIT_DIR});
 384}
 385
 386sub cmd_clone {
 387        my ($url, $path) = @_;
 388        if (!defined $path &&
 389            (defined $_trunk || @_branches || @_tags ||
 390             defined $_stdlayout) &&
 391            $url !~ m#^[a-z\+]+://#) {
 392                $path = $url;
 393        }
 394        $path = basename($url) if !defined $path || !length $path;
 395        my $authors_absolute = $_authors ? File::Spec->rel2abs($_authors) : "";
 396        cmd_init($url, $path);
 397        command_oneline('config', 'svn.authorsfile', $authors_absolute)
 398            if $_authors;
 399        Git::SVN::fetch_all($Git::SVN::default_repo_id);
 400}
 401
 402sub cmd_init {
 403        if (defined $_stdlayout) {
 404                $_trunk = 'trunk' if (!defined $_trunk);
 405                @_tags = 'tags' if (! @_tags);
 406                @_branches = 'branches' if (! @_branches);
 407        }
 408        if (defined $_trunk || @_branches || @_tags) {
 409                return cmd_multi_init(@_);
 410        }
 411        my $url = shift or die "SVN repository location required ",
 412                               "as a command-line argument\n";
 413        $url = canonicalize_url($url);
 414        init_subdir(@_);
 415        do_git_init_db();
 416
 417        if ($Git::SVN::_minimize_url eq 'unset') {
 418                $Git::SVN::_minimize_url = 0;
 419        }
 420
 421        Git::SVN->init($url);
 422}
 423
 424sub cmd_fetch {
 425        if (grep /^\d+=./, @_) {
 426                die "'<rev>=<commit>' fetch arguments are ",
 427                    "no longer supported.\n";
 428        }
 429        my ($remote) = @_;
 430        if (@_ > 1) {
 431                die "Usage: $0 fetch [--all] [--parent] [svn-remote]\n";
 432        }
 433        $Git::SVN::no_reuse_existing = undef;
 434        if ($_fetch_parent) {
 435                my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 436                unless ($gs) {
 437                        die "Unable to determine upstream SVN information from ",
 438                            "working tree history\n";
 439                }
 440                # just fetch, don't checkout.
 441                $_no_checkout = 'true';
 442                $_fetch_all ? $gs->fetch_all : $gs->fetch;
 443        } elsif ($_fetch_all) {
 444                cmd_multi_fetch();
 445        } else {
 446                $remote ||= $Git::SVN::default_repo_id;
 447                Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
 448        }
 449}
 450
 451sub cmd_set_tree {
 452        my (@commits) = @_;
 453        if ($_stdin || !@commits) {
 454                print "Reading from stdin...\n";
 455                @commits = ();
 456                while (<STDIN>) {
 457                        if (/\b($sha1_short)\b/o) {
 458                                unshift @commits, $1;
 459                        }
 460                }
 461        }
 462        my @revs;
 463        foreach my $c (@commits) {
 464                my @tmp = command('rev-parse',$c);
 465                if (scalar @tmp == 1) {
 466                        push @revs, $tmp[0];
 467                } elsif (scalar @tmp > 1) {
 468                        push @revs, reverse(command('rev-list',@tmp));
 469                } else {
 470                        fatal "Failed to rev-parse $c";
 471                }
 472        }
 473        my $gs = Git::SVN->new;
 474        my ($r_last, $cmt_last) = $gs->last_rev_commit;
 475        $gs->fetch;
 476        if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
 477                fatal "There are new revisions that were fetched ",
 478                      "and need to be merged (or acknowledged) ",
 479                      "before committing.\nlast rev: $r_last\n",
 480                      " current: $gs->{last_rev}";
 481        }
 482        $gs->set_tree($_) foreach @revs;
 483        print "Done committing ",scalar @revs," revisions to SVN\n";
 484        unlink $gs->{index};
 485}
 486
 487sub cmd_dcommit {
 488        my $head = shift;
 489        git_cmd_try { command_oneline(qw/diff-index --quiet HEAD/) }
 490                'Cannot dcommit with a dirty index.  Commit your changes first, '
 491                . "or stash them with `git stash'.\n";
 492        $head ||= 'HEAD';
 493
 494        my $old_head;
 495        if ($head ne 'HEAD') {
 496                $old_head = eval {
 497                        command_oneline([qw/symbolic-ref -q HEAD/])
 498                };
 499                if ($old_head) {
 500                        $old_head =~ s{^refs/heads/}{};
 501                } else {
 502                        $old_head = eval { command_oneline(qw/rev-parse HEAD/) };
 503                }
 504                command(['checkout', $head], STDERR => 0);
 505        }
 506
 507        my @refs;
 508        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD', \@refs);
 509        unless ($gs) {
 510                die "Unable to determine upstream SVN information from ",
 511                    "$head history.\nPerhaps the repository is empty.";
 512        }
 513
 514        if (defined $_commit_url) {
 515                $url = $_commit_url;
 516        } else {
 517                $url = eval { command_oneline('config', '--get',
 518                              "svn-remote.$gs->{repo_id}.commiturl") };
 519                if (!$url) {
 520                        $url = $gs->full_url
 521                }
 522        }
 523
 524        my $last_rev = $_revision if defined $_revision;
 525        if ($url) {
 526                print "Committing to $url ...\n";
 527        }
 528        my ($linear_refs, $parents) = linearize_history($gs, \@refs);
 529        if ($_no_rebase && scalar(@$linear_refs) > 1) {
 530                warn "Attempting to commit more than one change while ",
 531                     "--no-rebase is enabled.\n",
 532                     "If these changes depend on each other, re-running ",
 533                     "without --no-rebase may be required."
 534        }
 535        my $expect_url = $url;
 536        Git::SVN::remove_username($expect_url);
 537        while (1) {
 538                my $d = shift @$linear_refs or last;
 539                unless (defined $last_rev) {
 540                        (undef, $last_rev, undef) = cmt_metadata("$d~1");
 541                        unless (defined $last_rev) {
 542                                fatal "Unable to extract revision information ",
 543                                      "from commit $d~1";
 544                        }
 545                }
 546                if ($_dry_run) {
 547                        print "diff-tree $d~1 $d\n";
 548                } else {
 549                        my $cmt_rev;
 550                        my %ed_opts = ( r => $last_rev,
 551                                        log => get_commit_entry($d)->{log},
 552                                        ra => Git::SVN::Ra->new($url),
 553                                        config => SVN::Core::config_get_config(
 554                                                $Git::SVN::Ra::config_dir
 555                                        ),
 556                                        tree_a => "$d~1",
 557                                        tree_b => $d,
 558                                        editor_cb => sub {
 559                                               print "Committed r$_[0]\n";
 560                                               $cmt_rev = $_[0];
 561                                        },
 562                                        svn_path => '');
 563                        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
 564                                print "No changes\n$d~1 == $d\n";
 565                        } elsif ($parents->{$d} && @{$parents->{$d}}) {
 566                                $gs->{inject_parents_dcommit}->{$cmt_rev} =
 567                                                               $parents->{$d};
 568                        }
 569                        $_fetch_all ? $gs->fetch_all : $gs->fetch;
 570                        $last_rev = $cmt_rev;
 571                        next if $_no_rebase;
 572
 573                        # we always want to rebase against the current HEAD,
 574                        # not any head that was passed to us
 575                        my @diff = command('diff-tree', $d,
 576                                           $gs->refname, '--');
 577                        my @finish;
 578                        if (@diff) {
 579                                @finish = rebase_cmd();
 580                                print STDERR "W: $d and ", $gs->refname,
 581                                             " differ, using @finish:\n",
 582                                             join("\n", @diff), "\n";
 583                        } else {
 584                                print "No changes between current HEAD and ",
 585                                      $gs->refname,
 586                                      "\nResetting to the latest ",
 587                                      $gs->refname, "\n";
 588                                @finish = qw/reset --mixed/;
 589                        }
 590                        command_noisy(@finish, $gs->refname);
 591                        if (@diff) {
 592                                @refs = ();
 593                                my ($url_, $rev_, $uuid_, $gs_) =
 594                                              working_head_info('HEAD', \@refs);
 595                                my ($linear_refs_, $parents_) =
 596                                              linearize_history($gs_, \@refs);
 597                                if (scalar(@$linear_refs) !=
 598                                    scalar(@$linear_refs_)) {
 599                                        fatal "# of revisions changed ",
 600                                          "\nbefore:\n",
 601                                          join("\n", @$linear_refs),
 602                                          "\n\nafter:\n",
 603                                          join("\n", @$linear_refs_), "\n",
 604                                          'If you are attempting to commit ',
 605                                          "merges, try running:\n\t",
 606                                          'git rebase --interactive',
 607                                          '--preserve-merges ',
 608                                          $gs->refname,
 609                                          "\nBefore dcommitting";
 610                                }
 611                                if ($url_ ne $expect_url) {
 612                                        if ($url_ eq $gs->metadata_url) {
 613                                                print
 614                                                  "Accepting rewritten URL:",
 615                                                  " $url_\n";
 616                                        } else {
 617                                                fatal
 618                                                  "URL mismatch after rebase:",
 619                                                  " $url_ != $expect_url";
 620                                        }
 621                                }
 622                                if ($uuid_ ne $uuid) {
 623                                        fatal "uuid mismatch after rebase: ",
 624                                              "$uuid_ != $uuid";
 625                                }
 626                                # remap parents
 627                                my (%p, @l, $i);
 628                                for ($i = 0; $i < scalar @$linear_refs; $i++) {
 629                                        my $new = $linear_refs_->[$i] or next;
 630                                        $p{$new} =
 631                                                $parents->{$linear_refs->[$i]};
 632                                        push @l, $new;
 633                                }
 634                                $parents = \%p;
 635                                $linear_refs = \@l;
 636                        }
 637                }
 638        }
 639
 640        if ($old_head) {
 641                my $new_head = command_oneline(qw/rev-parse HEAD/);
 642                my $new_is_symbolic = eval {
 643                        command_oneline(qw/symbolic-ref -q HEAD/);
 644                };
 645                if ($new_is_symbolic) {
 646                        print "dcommitted the branch ", $head, "\n";
 647                } else {
 648                        print "dcommitted on a detached HEAD because you gave ",
 649                              "a revision argument.\n",
 650                              "The rewritten commit is: ", $new_head, "\n";
 651                }
 652                command(['checkout', $old_head], STDERR => 0);
 653        }
 654
 655        unlink $gs->{index};
 656}
 657
 658sub cmd_branch {
 659        my ($branch_name, $head) = @_;
 660
 661        unless (defined $branch_name && length $branch_name) {
 662                die(($_tag ? "tag" : "branch") . " name required\n");
 663        }
 664        $head ||= 'HEAD';
 665
 666        my (undef, $rev, undef, $gs) = working_head_info($head);
 667        my $src = $gs->full_url;
 668
 669        my $remote = Git::SVN::read_all_remotes()->{$gs->{repo_id}};
 670        my $allglobs = $remote->{ $_tag ? 'tags' : 'branches' };
 671        my $glob;
 672        if ($#{$allglobs} == 0) {
 673                $glob = $allglobs->[0];
 674        } else {
 675                unless(defined $_branch_dest) {
 676                        die "Multiple ",
 677                            $_tag ? "tag" : "branch",
 678                            " paths defined for Subversion repository.\n",
 679                            "You must specify where you want to create the ",
 680                            $_tag ? "tag" : "branch",
 681                            " with the --destination argument.\n";
 682                }
 683                foreach my $g (@{$allglobs}) {
 684                        # SVN::Git::Editor could probably be moved to Git.pm..
 685                        my $re = SVN::Git::Editor::glob2pat($g->{path}->{left});
 686                        if ($_branch_dest =~ /$re/) {
 687                                $glob = $g;
 688                                last;
 689                        }
 690                }
 691                unless (defined $glob) {
 692                        my $dest_re = qr/\b\Q$_branch_dest\E\b/;
 693                        foreach my $g (@{$allglobs}) {
 694                                $g->{path}->{left} =~ /$dest_re/ or next;
 695                                if (defined $glob) {
 696                                        die "Ambiguous destination: ",
 697                                            $_branch_dest, "\nmatches both '",
 698                                            $glob->{path}->{left}, "' and '",
 699                                            $g->{path}->{left}, "'\n";
 700                                }
 701                                $glob = $g;
 702                        }
 703                        unless (defined $glob) {
 704                                die "Unknown ",
 705                                    $_tag ? "tag" : "branch",
 706                                    " destination $_branch_dest\n";
 707                        }
 708                }
 709        }
 710        my ($lft, $rgt) = @{ $glob->{path} }{qw/left right/};
 711        my $dst = join '/', $remote->{url}, $lft, $branch_name, ($rgt || ());
 712
 713        my $ctx = SVN::Client->new(
 714                auth    => Git::SVN::Ra::_auth_providers(),
 715                log_msg => sub {
 716                        ${ $_[0] } = defined $_message
 717                                ? $_message
 718                                : 'Create ' . ($_tag ? 'tag ' : 'branch ' )
 719                                . $branch_name;
 720                },
 721        );
 722
 723        eval {
 724                $ctx->ls($dst, 'HEAD', 0);
 725        } and die "branch ${branch_name} already exists\n";
 726
 727        print "Copying ${src} at r${rev} to ${dst}...\n";
 728        $ctx->copy($src, $rev, $dst)
 729                unless $_dry_run;
 730
 731        $gs->fetch_all;
 732}
 733
 734sub cmd_find_rev {
 735        my $revision_or_hash = shift or die "SVN or git revision required ",
 736                                            "as a command-line argument\n";
 737        my $result;
 738        if ($revision_or_hash =~ /^r\d+$/) {
 739                my $head = shift;
 740                $head ||= 'HEAD';
 741                my @refs;
 742                my (undef, undef, $uuid, $gs) = working_head_info($head, \@refs);
 743                unless ($gs) {
 744                        die "Unable to determine upstream SVN information from ",
 745                            "$head history\n";
 746                }
 747                my $desired_revision = substr($revision_or_hash, 1);
 748                $result = $gs->rev_map_get($desired_revision, $uuid);
 749        } else {
 750                my (undef, $rev, undef) = cmt_metadata($revision_or_hash);
 751                $result = $rev;
 752        }
 753        print "$result\n" if $result;
 754}
 755
 756sub cmd_rebase {
 757        command_noisy(qw/update-index --refresh/);
 758        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 759        unless ($gs) {
 760                die "Unable to determine upstream SVN information from ",
 761                    "working tree history\n";
 762        }
 763        if ($_dry_run) {
 764                print "Remote Branch: " . $gs->refname . "\n";
 765                print "SVN URL: " . $url . "\n";
 766                return;
 767        }
 768        if (command(qw/diff-index HEAD --/)) {
 769                print STDERR "Cannot rebase with uncommited changes:\n";
 770                command_noisy('status');
 771                exit 1;
 772        }
 773        unless ($_local) {
 774                # rebase will checkout for us, so no need to do it explicitly
 775                $_no_checkout = 'true';
 776                $_fetch_all ? $gs->fetch_all : $gs->fetch;
 777        }
 778        command_noisy(rebase_cmd(), $gs->refname);
 779        $gs->mkemptydirs;
 780}
 781
 782sub cmd_show_ignore {
 783        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 784        $gs ||= Git::SVN->new;
 785        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 786        $gs->prop_walk($gs->{path}, $r, sub {
 787                my ($gs, $path, $props) = @_;
 788                print STDOUT "\n# $path\n";
 789                my $s = $props->{'svn:ignore'} or return;
 790                $s =~ s/[\r\n]+/\n/g;
 791                $s =~ s/^\n+//;
 792                chomp $s;
 793                $s =~ s#^#$path#gm;
 794                print STDOUT "$s\n";
 795        });
 796}
 797
 798sub cmd_show_externals {
 799        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 800        $gs ||= Git::SVN->new;
 801        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 802        $gs->prop_walk($gs->{path}, $r, sub {
 803                my ($gs, $path, $props) = @_;
 804                print STDOUT "\n# $path\n";
 805                my $s = $props->{'svn:externals'} or return;
 806                $s =~ s/[\r\n]+/\n/g;
 807                chomp $s;
 808                $s =~ s#^#$path#gm;
 809                print STDOUT "$s\n";
 810        });
 811}
 812
 813sub cmd_create_ignore {
 814        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 815        $gs ||= Git::SVN->new;
 816        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 817        $gs->prop_walk($gs->{path}, $r, sub {
 818                my ($gs, $path, $props) = @_;
 819                # $path is of the form /path/to/dir/
 820                $path = '.' . $path;
 821                # SVN can have attributes on empty directories,
 822                # which git won't track
 823                mkpath([$path]) unless -d $path;
 824                my $ignore = $path . '.gitignore';
 825                my $s = $props->{'svn:ignore'} or return;
 826                open(GITIGNORE, '>', $ignore)
 827                  or fatal("Failed to open `$ignore' for writing: $!");
 828                $s =~ s/[\r\n]+/\n/g;
 829                $s =~ s/^\n+//;
 830                chomp $s;
 831                # Prefix all patterns so that the ignore doesn't apply
 832                # to sub-directories.
 833                $s =~ s#^#/#gm;
 834                print GITIGNORE "$s\n";
 835                close(GITIGNORE)
 836                  or fatal("Failed to close `$ignore': $!");
 837                command_noisy('add', '-f', $ignore);
 838        });
 839}
 840
 841sub cmd_mkdirs {
 842        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 843        $gs ||= Git::SVN->new;
 844        $gs->mkemptydirs($_revision);
 845}
 846
 847sub canonicalize_path {
 848        my ($path) = @_;
 849        my $dot_slash_added = 0;
 850        if (substr($path, 0, 1) ne "/") {
 851                $path = "./" . $path;
 852                $dot_slash_added = 1;
 853        }
 854        # File::Spec->canonpath doesn't collapse x/../y into y (for a
 855        # good reason), so let's do this manually.
 856        $path =~ s#/+#/#g;
 857        $path =~ s#/\.(?:/|$)#/#g;
 858        $path =~ s#/[^/]+/\.\.##g;
 859        $path =~ s#/$##g;
 860        $path =~ s#^\./## if $dot_slash_added;
 861        $path =~ s#^/##;
 862        $path =~ s#^\.$##;
 863        return $path;
 864}
 865
 866sub canonicalize_url {
 867        my ($url) = @_;
 868        $url =~ s#^([^:]+://[^/]*/)(.*)$#$1 . canonicalize_path($2)#e;
 869        return $url;
 870}
 871
 872# get_svnprops(PATH)
 873# ------------------
 874# Helper for cmd_propget and cmd_proplist below.
 875sub get_svnprops {
 876        my $path = shift;
 877        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 878        $gs ||= Git::SVN->new;
 879
 880        # prefix THE PATH by the sub-directory from which the user
 881        # invoked us.
 882        $path = $cmd_dir_prefix . $path;
 883        fatal("No such file or directory: $path") unless -e $path;
 884        my $is_dir = -d $path ? 1 : 0;
 885        $path = $gs->{path} . '/' . $path;
 886
 887        # canonicalize the path (otherwise libsvn will abort or fail to
 888        # find the file)
 889        $path = canonicalize_path($path);
 890
 891        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 892        my $props;
 893        if ($is_dir) {
 894                (undef, undef, $props) = $gs->ra->get_dir($path, $r);
 895        }
 896        else {
 897                (undef, $props) = $gs->ra->get_file($path, $r, undef);
 898        }
 899        return $props;
 900}
 901
 902# cmd_propget (PROP, PATH)
 903# ------------------------
 904# Print the SVN property PROP for PATH.
 905sub cmd_propget {
 906        my ($prop, $path) = @_;
 907        $path = '.' if not defined $path;
 908        usage(1) if not defined $prop;
 909        my $props = get_svnprops($path);
 910        if (not defined $props->{$prop}) {
 911                fatal("`$path' does not have a `$prop' SVN property.");
 912        }
 913        print $props->{$prop} . "\n";
 914}
 915
 916# cmd_proplist (PATH)
 917# -------------------
 918# Print the list of SVN properties for PATH.
 919sub cmd_proplist {
 920        my $path = shift;
 921        $path = '.' if not defined $path;
 922        my $props = get_svnprops($path);
 923        print "Properties on '$path':\n";
 924        foreach (sort keys %{$props}) {
 925                print "  $_\n";
 926        }
 927}
 928
 929sub cmd_multi_init {
 930        my $url = shift;
 931        unless (defined $_trunk || @_branches || @_tags) {
 932                usage(1);
 933        }
 934
 935        $_prefix = '' unless defined $_prefix;
 936        if (defined $url) {
 937                $url = canonicalize_url($url);
 938                init_subdir(@_);
 939        }
 940        do_git_init_db();
 941        if (defined $_trunk) {
 942                my $trunk_ref = 'refs/remotes/' . $_prefix . 'trunk';
 943                # try both old-style and new-style lookups:
 944                my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
 945                unless ($gs_trunk) {
 946                        my ($trunk_url, $trunk_path) =
 947                                              complete_svn_url($url, $_trunk);
 948                        $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
 949                                                   undef, $trunk_ref);
 950                }
 951        }
 952        return unless @_branches || @_tags;
 953        my $ra = $url ? Git::SVN::Ra->new($url) : undef;
 954        foreach my $path (@_branches) {
 955                complete_url_ls_init($ra, $path, '--branches/-b', $_prefix);
 956        }
 957        foreach my $path (@_tags) {
 958                complete_url_ls_init($ra, $path, '--tags/-t', $_prefix.'tags/');
 959        }
 960}
 961
 962sub cmd_multi_fetch {
 963        $Git::SVN::no_reuse_existing = undef;
 964        my $remotes = Git::SVN::read_all_remotes();
 965        foreach my $repo_id (sort keys %$remotes) {
 966                if ($remotes->{$repo_id}->{url}) {
 967                        Git::SVN::fetch_all($repo_id, $remotes);
 968                }
 969        }
 970}
 971
 972# this command is special because it requires no metadata
 973sub cmd_commit_diff {
 974        my ($ta, $tb, $url) = @_;
 975        my $usage = "Usage: $0 commit-diff -r<revision> ".
 976                    "<tree-ish> <tree-ish> [<URL>]";
 977        fatal($usage) if (!defined $ta || !defined $tb);
 978        my $svn_path = '';
 979        if (!defined $url) {
 980                my $gs = eval { Git::SVN->new };
 981                if (!$gs) {
 982                        fatal("Needed URL or usable git-svn --id in ",
 983                              "the command-line\n", $usage);
 984                }
 985                $url = $gs->{url};
 986                $svn_path = $gs->{path};
 987        }
 988        unless (defined $_revision) {
 989                fatal("-r|--revision is a required argument\n", $usage);
 990        }
 991        if (defined $_message && defined $_file) {
 992                fatal("Both --message/-m and --file/-F specified ",
 993                      "for the commit message.\n",
 994                      "I have no idea what you mean");
 995        }
 996        if (defined $_file) {
 997                $_message = file_to_s($_file);
 998        } else {
 999                $_message ||= get_commit_entry($tb)->{log};
1000        }
1001        my $ra ||= Git::SVN::Ra->new($url);
1002        my $r = $_revision;
1003        if ($r eq 'HEAD') {
1004                $r = $ra->get_latest_revnum;
1005        } elsif ($r !~ /^\d+$/) {
1006                die "revision argument: $r not understood by git-svn\n";
1007        }
1008        my %ed_opts = ( r => $r,
1009                        log => $_message,
1010                        ra => $ra,
1011                        tree_a => $ta,
1012                        tree_b => $tb,
1013                        editor_cb => sub { print "Committed r$_[0]\n" },
1014                        svn_path => $svn_path );
1015        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
1016                print "No changes\n$ta == $tb\n";
1017        }
1018}
1019
1020sub escape_uri_only {
1021        my ($uri) = @_;
1022        my @tmp;
1023        foreach (split m{/}, $uri) {
1024                s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
1025                push @tmp, $_;
1026        }
1027        join('/', @tmp);
1028}
1029
1030sub escape_url {
1031        my ($url) = @_;
1032        if ($url =~ m#^([^:]+)://([^/]*)(.*)$#) {
1033                my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
1034                $url = "$scheme://$domain$uri";
1035        }
1036        $url;
1037}
1038
1039sub cmd_info {
1040        my $path = canonicalize_path(defined($_[0]) ? $_[0] : ".");
1041        my $fullpath = canonicalize_path($cmd_dir_prefix . $path);
1042        if (exists $_[1]) {
1043                die "Too many arguments specified\n";
1044        }
1045
1046        my ($file_type, $diff_status) = find_file_type_and_diff_status($path);
1047
1048        if (!$file_type && !$diff_status) {
1049                print STDERR "svn: '$path' is not under version control\n";
1050                exit 1;
1051        }
1052
1053        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1054        unless ($gs) {
1055                die "Unable to determine upstream SVN information from ",
1056                    "working tree history\n";
1057        }
1058
1059        # canonicalize_path() will return "" to make libsvn 1.5.x happy,
1060        $path = "." if $path eq "";
1061
1062        my $full_url = $url . ($fullpath eq "" ? "" : "/$fullpath");
1063
1064        if ($_url) {
1065                print escape_url($full_url), "\n";
1066                return;
1067        }
1068
1069        my $result = "Path: $path\n";
1070        $result .= "Name: " . basename($path) . "\n" if $file_type ne "dir";
1071        $result .= "URL: " . escape_url($full_url) . "\n";
1072
1073        eval {
1074                my $repos_root = $gs->repos_root;
1075                Git::SVN::remove_username($repos_root);
1076                $result .= "Repository Root: " . escape_url($repos_root) . "\n";
1077        };
1078        if ($@) {
1079                $result .= "Repository Root: (offline)\n";
1080        }
1081        $result .= "Repository UUID: $uuid\n" unless $diff_status eq "A" &&
1082                ($SVN::Core::VERSION le '1.5.4' || $file_type ne "dir");
1083        $result .= "Revision: " . ($diff_status eq "A" ? 0 : $rev) . "\n";
1084
1085        $result .= "Node Kind: " .
1086                   ($file_type eq "dir" ? "directory" : "file") . "\n";
1087
1088        my $schedule = $diff_status eq "A"
1089                       ? "add"
1090                       : ($diff_status eq "D" ? "delete" : "normal");
1091        $result .= "Schedule: $schedule\n";
1092
1093        if ($diff_status eq "A") {
1094                print $result, "\n";
1095                return;
1096        }
1097
1098        my ($lc_author, $lc_rev, $lc_date_utc);
1099        my @args = Git::SVN::Log::git_svn_log_cmd($rev, $rev, "--", $fullpath);
1100        my $log = command_output_pipe(@args);
1101        my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
1102        while (<$log>) {
1103                if (/^${esc_color}author (.+) <[^>]+> (\d+) ([\-\+]?\d+)$/o) {
1104                        $lc_author = $1;
1105                        $lc_date_utc = Git::SVN::Log::parse_git_date($2, $3);
1106                } elsif (/^${esc_color}    (git-svn-id:.+)$/o) {
1107                        (undef, $lc_rev, undef) = ::extract_metadata($1);
1108                }
1109        }
1110        close $log;
1111
1112        Git::SVN::Log::set_local_timezone();
1113
1114        $result .= "Last Changed Author: $lc_author\n";
1115        $result .= "Last Changed Rev: $lc_rev\n";
1116        $result .= "Last Changed Date: " .
1117                   Git::SVN::Log::format_svn_date($lc_date_utc) . "\n";
1118
1119        if ($file_type ne "dir") {
1120                my $text_last_updated_date =
1121                    ($diff_status eq "D" ? $lc_date_utc : (stat $path)[9]);
1122                $result .=
1123                    "Text Last Updated: " .
1124                    Git::SVN::Log::format_svn_date($text_last_updated_date) .
1125                    "\n";
1126                my $checksum;
1127                if ($diff_status eq "D") {
1128                        my ($fh, $ctx) =
1129                            command_output_pipe(qw(cat-file blob), "HEAD:$path");
1130                        if ($file_type eq "link") {
1131                                my $file_name = <$fh>;
1132                                $checksum = md5sum("link $file_name");
1133                        } else {
1134                                $checksum = md5sum($fh);
1135                        }
1136                        command_close_pipe($fh, $ctx);
1137                } elsif ($file_type eq "link") {
1138                        my $file_name =
1139                            command(qw(cat-file blob), "HEAD:$path");
1140                        $checksum =
1141                            md5sum("link " . $file_name);
1142                } else {
1143                        open FILE, "<", $path or die $!;
1144                        $checksum = md5sum(\*FILE);
1145                        close FILE or die $!;
1146                }
1147                $result .= "Checksum: " . $checksum . "\n";
1148        }
1149
1150        print $result, "\n";
1151}
1152
1153sub cmd_reset {
1154        my $target = shift || $_revision or die "SVN revision required\n";
1155        $target = $1 if $target =~ /^r(\d+)$/;
1156        $target =~ /^\d+$/ or die "Numeric SVN revision expected\n";
1157        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1158        unless ($gs) {
1159                die "Unable to determine upstream SVN information from ".
1160                    "history\n";
1161        }
1162        my ($r, $c) = $gs->find_rev_before($target, not $_fetch_parent);
1163        $gs->rev_map_set($r, $c, 'reset', $uuid);
1164        print "r$r = $c ($gs->{ref_id})\n";
1165}
1166
1167sub cmd_gc {
1168        if (!$can_compress) {
1169                warn "Compress::Zlib could not be found; unhandled.log " .
1170                     "files will not be compressed.\n";
1171        }
1172        find({ wanted => \&gc_directory, no_chdir => 1}, "$ENV{GIT_DIR}/svn");
1173}
1174
1175########################### utility functions #########################
1176
1177sub rebase_cmd {
1178        my @cmd = qw/rebase/;
1179        push @cmd, '-v' if $_verbose;
1180        push @cmd, qw/--merge/ if $_merge;
1181        push @cmd, "--strategy=$_strategy" if $_strategy;
1182        @cmd;
1183}
1184
1185sub post_fetch_checkout {
1186        return if $_no_checkout;
1187        my $gs = $Git::SVN::_head or return;
1188        return if verify_ref('refs/heads/master^0');
1189
1190        # look for "trunk" ref if it exists
1191        my $remote = Git::SVN::read_all_remotes()->{$gs->{repo_id}};
1192        my $fetch = $remote->{fetch};
1193        if ($fetch) {
1194                foreach my $p (keys %$fetch) {
1195                        basename($fetch->{$p}) eq 'trunk' or next;
1196                        $gs = Git::SVN->new($fetch->{$p}, $gs->{repo_id}, $p);
1197                        last;
1198                }
1199        }
1200
1201        my $valid_head = verify_ref('HEAD^0');
1202        command_noisy(qw(update-ref refs/heads/master), $gs->refname);
1203        return if ($valid_head || !verify_ref('HEAD^0'));
1204
1205        return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
1206        my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
1207        return if -f $index;
1208
1209        return if command_oneline(qw/rev-parse --is-inside-work-tree/) eq 'false';
1210        return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
1211        command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
1212        print STDERR "Checked out HEAD:\n  ",
1213                     $gs->full_url, " r", $gs->last_rev, "\n";
1214        $gs->mkemptydirs($gs->last_rev);
1215}
1216
1217sub complete_svn_url {
1218        my ($url, $path) = @_;
1219        $path =~ s#/+$##;
1220        if ($path !~ m#^[a-z\+]+://#) {
1221                if (!defined $url || $url !~ m#^[a-z\+]+://#) {
1222                        fatal("E: '$path' is not a complete URL ",
1223                              "and a separate URL is not specified");
1224                }
1225                return ($url, $path);
1226        }
1227        return ($path, '');
1228}
1229
1230sub complete_url_ls_init {
1231        my ($ra, $repo_path, $switch, $pfx) = @_;
1232        unless ($repo_path) {
1233                print STDERR "W: $switch not specified\n";
1234                return;
1235        }
1236        $repo_path =~ s#/+$##;
1237        if ($repo_path =~ m#^[a-z\+]+://#) {
1238                $ra = Git::SVN::Ra->new($repo_path);
1239                $repo_path = '';
1240        } else {
1241                $repo_path =~ s#^/+##;
1242                unless ($ra) {
1243                        fatal("E: '$repo_path' is not a complete URL ",
1244                              "and a separate URL is not specified");
1245                }
1246        }
1247        my $url = $ra->{url};
1248        my $gs = Git::SVN->init($url, undef, undef, undef, 1);
1249        my $k = "svn-remote.$gs->{repo_id}.url";
1250        my $orig_url = eval { command_oneline(qw/config --get/, $k) };
1251        if ($orig_url && ($orig_url ne $gs->{url})) {
1252                die "$k already set: $orig_url\n",
1253                    "wanted to set to: $gs->{url}\n";
1254        }
1255        command_oneline('config', $k, $gs->{url}) unless $orig_url;
1256        my $remote_path = "$gs->{path}/$repo_path";
1257        $remote_path =~ s{%([0-9A-F]{2})}{chr hex($1)}ieg;
1258        $remote_path =~ s#/+#/#g;
1259        $remote_path =~ s#^/##g;
1260        $remote_path .= "/*" if $remote_path !~ /\*/;
1261        my ($n) = ($switch =~ /^--(\w+)/);
1262        if (length $pfx && $pfx !~ m#/$#) {
1263                die "--prefix='$pfx' must have a trailing slash '/'\n";
1264        }
1265        command_noisy('config',
1266                      '--add',
1267                      "svn-remote.$gs->{repo_id}.$n",
1268                      "$remote_path:refs/remotes/$pfx*" .
1269                        ('/*' x (($remote_path =~ tr/*/*/) - 1)) );
1270}
1271
1272sub verify_ref {
1273        my ($ref) = @_;
1274        eval { command_oneline([ 'rev-parse', '--verify', $ref ],
1275                               { STDERR => 0 }); };
1276}
1277
1278sub get_tree_from_treeish {
1279        my ($treeish) = @_;
1280        # $treeish can be a symbolic ref, too:
1281        my $type = command_oneline(qw/cat-file -t/, $treeish);
1282        my $expected;
1283        while ($type eq 'tag') {
1284                ($treeish, $type) = command(qw/cat-file tag/, $treeish);
1285        }
1286        if ($type eq 'commit') {
1287                $expected = (grep /^tree /, command(qw/cat-file commit/,
1288                                                    $treeish))[0];
1289                ($expected) = ($expected =~ /^tree ($sha1)$/o);
1290                die "Unable to get tree from $treeish\n" unless $expected;
1291        } elsif ($type eq 'tree') {
1292                $expected = $treeish;
1293        } else {
1294                die "$treeish is a $type, expected tree, tag or commit\n";
1295        }
1296        return $expected;
1297}
1298
1299sub get_commit_entry {
1300        my ($treeish) = shift;
1301        my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
1302        my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
1303        my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
1304        open my $log_fh, '>', $commit_editmsg or croak $!;
1305
1306        my $type = command_oneline(qw/cat-file -t/, $treeish);
1307        if ($type eq 'commit' || $type eq 'tag') {
1308                my ($msg_fh, $ctx) = command_output_pipe('cat-file',
1309                                                         $type, $treeish);
1310                my $in_msg = 0;
1311                my $author;
1312                my $saw_from = 0;
1313                my $msgbuf = "";
1314                while (<$msg_fh>) {
1315                        if (!$in_msg) {
1316                                $in_msg = 1 if (/^\s*$/);
1317                                $author = $1 if (/^author (.*>)/);
1318                        } elsif (/^git-svn-id: /) {
1319                                # skip this for now, we regenerate the
1320                                # correct one on re-fetch anyways
1321                                # TODO: set *:merge properties or like...
1322                        } else {
1323                                if (/^From:/ || /^Signed-off-by:/) {
1324                                        $saw_from = 1;
1325                                }
1326                                $msgbuf .= $_;
1327                        }
1328                }
1329                $msgbuf =~ s/\s+$//s;
1330                if ($Git::SVN::_add_author_from && defined($author)
1331                    && !$saw_from) {
1332                        $msgbuf .= "\n\nFrom: $author";
1333                }
1334                print $log_fh $msgbuf or croak $!;
1335                command_close_pipe($msg_fh, $ctx);
1336        }
1337        close $log_fh or croak $!;
1338
1339        if ($_edit || ($type eq 'tree')) {
1340                chomp(my $editor = command_oneline(qw(var GIT_EDITOR)));
1341                system('sh', '-c', $editor.' "$@"', $editor, $commit_editmsg);
1342        }
1343        rename $commit_editmsg, $commit_msg or croak $!;
1344        {
1345                require Encode;
1346                # SVN requires messages to be UTF-8 when entering the repo
1347                local $/;
1348                open $log_fh, '<', $commit_msg or croak $!;
1349                binmode $log_fh;
1350                chomp($log_entry{log} = <$log_fh>);
1351
1352                my $enc = Git::config('i18n.commitencoding') || 'UTF-8';
1353                my $msg = $log_entry{log};
1354
1355                eval { $msg = Encode::decode($enc, $msg, 1) };
1356                if ($@) {
1357                        die "Could not decode as $enc:\n", $msg,
1358                            "\nPerhaps you need to set i18n.commitencoding\n";
1359                }
1360
1361                eval { $msg = Encode::encode('UTF-8', $msg, 1) };
1362                die "Could not encode as UTF-8:\n$msg\n" if $@;
1363
1364                $log_entry{log} = $msg;
1365
1366                close $log_fh or croak $!;
1367        }
1368        unlink $commit_msg;
1369        \%log_entry;
1370}
1371
1372sub s_to_file {
1373        my ($str, $file, $mode) = @_;
1374        open my $fd,'>',$file or croak $!;
1375        print $fd $str,"\n" or croak $!;
1376        close $fd or croak $!;
1377        chmod ($mode &~ umask, $file) if (defined $mode);
1378}
1379
1380sub file_to_s {
1381        my $file = shift;
1382        open my $fd,'<',$file or croak "$!: file: $file\n";
1383        local $/;
1384        my $ret = <$fd>;
1385        close $fd or croak $!;
1386        $ret =~ s/\s*$//s;
1387        return $ret;
1388}
1389
1390# '<svn username> = real-name <email address>' mapping based on git-svnimport:
1391sub load_authors {
1392        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1393        my $log = $cmd eq 'log';
1394        while (<$authors>) {
1395                chomp;
1396                next unless /^(.+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
1397                my ($user, $name, $email) = ($1, $2, $3);
1398                if ($log) {
1399                        $Git::SVN::Log::rusers{"$name <$email>"} = $user;
1400                } else {
1401                        $users{$user} = [$name, $email];
1402                }
1403        }
1404        close $authors or croak $!;
1405}
1406
1407# convert GetOpt::Long specs for use by git-config
1408sub read_git_config {
1409        my $opts = shift;
1410        my @config_only;
1411        foreach my $o (keys %$opts) {
1412                # if we have mixedCase and a long option-only, then
1413                # it's a config-only variable that we don't need for
1414                # the command-line.
1415                push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
1416                my $v = $opts->{$o};
1417                my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
1418                $key =~ s/-//g;
1419                my $arg = 'git config';
1420                $arg .= ' --int' if ($o =~ /[:=]i$/);
1421                $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
1422                if (ref $v eq 'ARRAY') {
1423                        chomp(my @tmp = `$arg --get-all svn.$key`);
1424                        @$v = @tmp if @tmp;
1425                } else {
1426                        chomp(my $tmp = `$arg --get svn.$key`);
1427                        if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
1428                                $$v = $tmp;
1429                        }
1430                }
1431        }
1432        delete @$opts{@config_only} if @config_only;
1433}
1434
1435sub extract_metadata {
1436        my $id = shift or return (undef, undef, undef);
1437        my ($url, $rev, $uuid) = ($id =~ /^\s*git-svn-id:\s+(.*)\@(\d+)
1438                                                        \s([a-f\d\-]+)$/ix);
1439        if (!defined $rev || !$uuid || !$url) {
1440                # some of the original repositories I made had
1441                # identifiers like this:
1442                ($rev, $uuid) = ($id =~/^\s*git-svn-id:\s(\d+)\@([a-f\d\-]+)/i);
1443        }
1444        return ($url, $rev, $uuid);
1445}
1446
1447sub cmt_metadata {
1448        return extract_metadata((grep(/^git-svn-id: /,
1449                command(qw/cat-file commit/, shift)))[-1]);
1450}
1451
1452sub cmt_sha2rev_batch {
1453        my %s2r;
1454        my ($pid, $in, $out, $ctx) = command_bidi_pipe(qw/cat-file --batch/);
1455        my $list = shift;
1456
1457        foreach my $sha (@{$list}) {
1458                my $first = 1;
1459                my $size = 0;
1460                print $out $sha, "\n";
1461
1462                while (my $line = <$in>) {
1463                        if ($first && $line =~ /^[[:xdigit:]]{40}\smissing$/) {
1464                                last;
1465                        } elsif ($first &&
1466                               $line =~ /^[[:xdigit:]]{40}\scommit\s(\d+)$/) {
1467                                $first = 0;
1468                                $size = $1;
1469                                next;
1470                        } elsif ($line =~ /^(git-svn-id: )/) {
1471                                my (undef, $rev, undef) =
1472                                                      extract_metadata($line);
1473                                $s2r{$sha} = $rev;
1474                        }
1475
1476                        $size -= length($line);
1477                        last if ($size == 0);
1478                }
1479        }
1480
1481        command_close_bidi_pipe($pid, $in, $out, $ctx);
1482
1483        return \%s2r;
1484}
1485
1486sub working_head_info {
1487        my ($head, $refs) = @_;
1488        my @args = ('log', '--no-color', '--first-parent', '--pretty=medium');
1489        my ($fh, $ctx) = command_output_pipe(@args, $head);
1490        my $hash;
1491        my %max;
1492        while (<$fh>) {
1493                if ( m{^commit ($::sha1)$} ) {
1494                        unshift @$refs, $hash if $hash and $refs;
1495                        $hash = $1;
1496                        next;
1497                }
1498                next unless s{^\s*(git-svn-id:)}{$1};
1499                my ($url, $rev, $uuid) = extract_metadata($_);
1500                if (defined $url && defined $rev) {
1501                        next if $max{$url} and $max{$url} < $rev;
1502                        if (my $gs = Git::SVN->find_by_url($url)) {
1503                                my $c = $gs->rev_map_get($rev, $uuid);
1504                                if ($c && $c eq $hash) {
1505                                        close $fh; # break the pipe
1506                                        return ($url, $rev, $uuid, $gs);
1507                                } else {
1508                                        $max{$url} ||= $gs->rev_map_max;
1509                                }
1510                        }
1511                }
1512        }
1513        command_close_pipe($fh, $ctx);
1514        (undef, undef, undef, undef);
1515}
1516
1517sub read_commit_parents {
1518        my ($parents, $c) = @_;
1519        chomp(my $p = command_oneline(qw/rev-list --parents -1/, $c));
1520        $p =~ s/^($c)\s*// or die "rev-list --parents -1 $c failed!\n";
1521        @{$parents->{$c}} = split(/ /, $p);
1522}
1523
1524sub linearize_history {
1525        my ($gs, $refs) = @_;
1526        my %parents;
1527        foreach my $c (@$refs) {
1528                read_commit_parents(\%parents, $c);
1529        }
1530
1531        my @linear_refs;
1532        my %skip = ();
1533        my $last_svn_commit = $gs->last_commit;
1534        foreach my $c (reverse @$refs) {
1535                next if $c eq $last_svn_commit;
1536                last if $skip{$c};
1537
1538                unshift @linear_refs, $c;
1539                $skip{$c} = 1;
1540
1541                # we only want the first parent to diff against for linear
1542                # history, we save the rest to inject when we finalize the
1543                # svn commit
1544                my $fp_a = verify_ref("$c~1");
1545                my $fp_b = shift @{$parents{$c}} if $parents{$c};
1546                if (!$fp_a || !$fp_b) {
1547                        die "Commit $c\n",
1548                            "has no parent commit, and therefore ",
1549                            "nothing to diff against.\n",
1550                            "You should be working from a repository ",
1551                            "originally created by git-svn\n";
1552                }
1553                if ($fp_a ne $fp_b) {
1554                        die "$c~1 = $fp_a, however parsing commit $c ",
1555                            "revealed that:\n$c~1 = $fp_b\nBUG!\n";
1556                }
1557
1558                foreach my $p (@{$parents{$c}}) {
1559                        $skip{$p} = 1;
1560                }
1561        }
1562        (\@linear_refs, \%parents);
1563}
1564
1565sub find_file_type_and_diff_status {
1566        my ($path) = @_;
1567        return ('dir', '') if $path eq '';
1568
1569        my $diff_output =
1570            command_oneline(qw(diff --cached --name-status --), $path) || "";
1571        my $diff_status = (split(' ', $diff_output))[0] || "";
1572
1573        my $ls_tree = command_oneline(qw(ls-tree HEAD), $path) || "";
1574
1575        return (undef, undef) if !$diff_status && !$ls_tree;
1576
1577        if ($diff_status eq "A") {
1578                return ("link", $diff_status) if -l $path;
1579                return ("dir", $diff_status) if -d $path;
1580                return ("file", $diff_status);
1581        }
1582
1583        my $mode = (split(' ', $ls_tree))[0] || "";
1584
1585        return ("link", $diff_status) if $mode eq "120000";
1586        return ("dir", $diff_status) if $mode eq "040000";
1587        return ("file", $diff_status);
1588}
1589
1590sub md5sum {
1591        my $arg = shift;
1592        my $ref = ref $arg;
1593        my $md5 = Digest::MD5->new();
1594        if ($ref eq 'GLOB' || $ref eq 'IO::File' || $ref eq 'File::Temp') {
1595                $md5->addfile($arg) or croak $!;
1596        } elsif ($ref eq 'SCALAR') {
1597                $md5->add($$arg) or croak $!;
1598        } elsif (!$ref) {
1599                $md5->add($arg) or croak $!;
1600        } else {
1601                ::fatal "Can't provide MD5 hash for unknown ref type: '", $ref, "'";
1602        }
1603        return $md5->hexdigest();
1604}
1605
1606sub gc_directory {
1607        if ($can_compress && -f $_ && basename($_) eq "unhandled.log") {
1608                my $out_filename = $_ . ".gz";
1609                open my $in_fh, "<", $_ or die "Unable to open $_: $!\n";
1610                binmode $in_fh;
1611                my $gz = Compress::Zlib::gzopen($out_filename, "ab") or
1612                                die "Unable to open $out_filename: $!\n";
1613
1614                my $res;
1615                while ($res = sysread($in_fh, my $str, 1024)) {
1616                        $gz->gzwrite($str) or
1617                                die "Unable to write: ".$gz->gzerror()."!\n";
1618                }
1619                unlink $_ or die "unlink $File::Find::name: $!\n";
1620        } elsif (-f $_ && basename($_) eq "index") {
1621                unlink $_ or die "unlink $_: $!\n";
1622        }
1623}
1624
1625package Git::SVN;
1626use strict;
1627use warnings;
1628use Fcntl qw/:DEFAULT :seek/;
1629use constant rev_map_fmt => 'NH40';
1630use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
1631            $_repack $_repack_flags $_use_svm_props $_head
1632            $_use_svnsync_props $no_reuse_existing $_minimize_url
1633            $_use_log_author $_add_author_from $_localtime/;
1634use Carp qw/croak/;
1635use File::Path qw/mkpath/;
1636use File::Copy qw/copy/;
1637use IPC::Open3;
1638use Memoize;  # core since 5.8.0, Jul 2002
1639
1640my ($_gc_nr, $_gc_period);
1641
1642# properties that we do not log:
1643my %SKIP_PROP;
1644BEGIN {
1645        %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
1646                                        svn:special svn:executable
1647                                        svn:entry:committed-rev
1648                                        svn:entry:last-author
1649                                        svn:entry:uuid
1650                                        svn:entry:committed-date/;
1651
1652        # some options are read globally, but can be overridden locally
1653        # per [svn-remote "..."] section.  Command-line options will *NOT*
1654        # override options set in an [svn-remote "..."] section
1655        no strict 'refs';
1656        for my $option (qw/follow_parent no_metadata use_svm_props
1657                           use_svnsync_props/) {
1658                my $key = $option;
1659                $key =~ tr/_//d;
1660                my $prop = "-$option";
1661                *$option = sub {
1662                        my ($self) = @_;
1663                        return $self->{$prop} if exists $self->{$prop};
1664                        my $k = "svn-remote.$self->{repo_id}.$key";
1665                        eval { command_oneline(qw/config --get/, $k) };
1666                        if ($@) {
1667                                $self->{$prop} = ${"Git::SVN::_$option"};
1668                        } else {
1669                                my $v = command_oneline(qw/config --bool/,$k);
1670                                $self->{$prop} = $v eq 'false' ? 0 : 1;
1671                        }
1672                        return $self->{$prop};
1673                }
1674        }
1675}
1676
1677
1678my (%LOCKFILES, %INDEX_FILES);
1679END {
1680        unlink keys %LOCKFILES if %LOCKFILES;
1681        unlink keys %INDEX_FILES if %INDEX_FILES;
1682}
1683
1684sub resolve_local_globs {
1685        my ($url, $fetch, $glob_spec) = @_;
1686        return unless defined $glob_spec;
1687        my $ref = $glob_spec->{ref};
1688        my $path = $glob_spec->{path};
1689        foreach (command(qw#for-each-ref --format=%(refname) refs/#)) {
1690                next unless m#^$ref->{regex}$#;
1691                my $p = $1;
1692                my $pathname = desanitize_refname($path->full_path($p));
1693                my $refname = desanitize_refname($ref->full_path($p));
1694                if (my $existing = $fetch->{$pathname}) {
1695                        if ($existing ne $refname) {
1696                                die "Refspec conflict:\n",
1697                                    "existing: $existing\n",
1698                                    " globbed: $refname\n";
1699                        }
1700                        my $u = (::cmt_metadata("$refname"))[0];
1701                        $u =~ s!^\Q$url\E(/|$)!! or die
1702                          "$refname: '$url' not found in '$u'\n";
1703                        if ($pathname ne $u) {
1704                                warn "W: Refspec glob conflict ",
1705                                     "(ref: $refname):\n",
1706                                     "expected path: $pathname\n",
1707                                     "    real path: $u\n",
1708                                     "Continuing ahead with $u\n";
1709                                next;
1710                        }
1711                } else {
1712                        $fetch->{$pathname} = $refname;
1713                }
1714        }
1715}
1716
1717sub parse_revision_argument {
1718        my ($base, $head) = @_;
1719        if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
1720                return ($base, $head);
1721        }
1722        return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
1723        return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
1724        return ($head, $head) if ($::_revision eq 'HEAD');
1725        return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
1726        return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
1727        die "revision argument: $::_revision not understood by git-svn\n";
1728}
1729
1730sub fetch_all {
1731        my ($repo_id, $remotes) = @_;
1732        if (ref $repo_id) {
1733                my $gs = $repo_id;
1734                $repo_id = undef;
1735                $repo_id = $gs->{repo_id};
1736        }
1737        $remotes ||= read_all_remotes();
1738        my $remote = $remotes->{$repo_id} or
1739                     die "[svn-remote \"$repo_id\"] unknown\n";
1740        my $fetch = $remote->{fetch};
1741        my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
1742        my (@gs, @globs);
1743        my $ra = Git::SVN::Ra->new($url);
1744        my $uuid = $ra->get_uuid;
1745        my $head = $ra->get_latest_revnum;
1746
1747        # ignore errors, $head revision may not even exist anymore
1748        eval { $ra->get_log("", $head, 0, 1, 0, 1, sub { $head = $_[1] }) };
1749        warn "W: $@\n" if $@;
1750
1751        my $base = defined $fetch ? $head : 0;
1752
1753        # read the max revs for wildcard expansion (branches/*, tags/*)
1754        foreach my $t (qw/branches tags/) {
1755                defined $remote->{$t} or next;
1756                push @globs, @{$remote->{$t}};
1757
1758                my $max_rev = eval { tmp_config(qw/--int --get/,
1759                                         "svn-remote.$repo_id.${t}-maxRev") };
1760                if (defined $max_rev && ($max_rev < $base)) {
1761                        $base = $max_rev;
1762                } elsif (!defined $max_rev) {
1763                        $base = 0;
1764                }
1765        }
1766
1767        if ($fetch) {
1768                foreach my $p (sort keys %$fetch) {
1769                        my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
1770                        my $lr = $gs->rev_map_max;
1771                        if (defined $lr) {
1772                                $base = $lr if ($lr < $base);
1773                        }
1774                        push @gs, $gs;
1775                }
1776        }
1777
1778        ($base, $head) = parse_revision_argument($base, $head);
1779        $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
1780}
1781
1782sub read_all_remotes {
1783        my $r = {};
1784        my $use_svm_props = eval { command_oneline(qw/config --bool
1785            svn.useSvmProps/) };
1786        $use_svm_props = $use_svm_props eq 'true' if $use_svm_props;
1787        my $svn_refspec = qr{\s*(.*?)\s*:\s*(.+?)\s*};
1788        foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
1789                if (m!^(.+)\.fetch=$svn_refspec$!) {
1790                        my ($remote, $local_ref, $remote_ref) = ($1, $2, $3);
1791                        die("svn-remote.$remote: remote ref '$remote_ref' "
1792                            . "must start with 'refs/'\n")
1793                                unless $remote_ref =~ m{^refs/};
1794                        $r->{$remote}->{fetch}->{$local_ref} = $remote_ref;
1795                        $r->{$remote}->{svm} = {} if $use_svm_props;
1796                } elsif (m!^(.+)\.usesvmprops=\s*(.*)\s*$!) {
1797                        $r->{$1}->{svm} = {};
1798                } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
1799                        $r->{$1}->{url} = $2;
1800                } elsif (m!^(.+)\.(branches|tags)=$svn_refspec$!) {
1801                        my ($remote, $t, $local_ref, $remote_ref) =
1802                                                             ($1, $2, $3, $4);
1803                        die("svn-remote.$remote: remote ref '$remote_ref' ($t) "
1804                            . "must start with 'refs/'\n")
1805                                unless $remote_ref =~ m{^refs/};
1806                        my $rs = {
1807                            t => $t,
1808                            remote => $remote,
1809                            path => Git::SVN::GlobSpec->new($local_ref),
1810                            ref => Git::SVN::GlobSpec->new($remote_ref) };
1811                        if (length($rs->{ref}->{right}) != 0) {
1812                                die "The '*' glob character must be the last ",
1813                                    "character of '$remote_ref'\n";
1814                        }
1815                        push @{ $r->{$remote}->{$t} }, $rs;
1816                }
1817        }
1818
1819        map {
1820                if (defined $r->{$_}->{svm}) {
1821                        my $svm;
1822                        eval {
1823                                my $section = "svn-remote.$_";
1824                                $svm = {
1825                                        source => tmp_config('--get',
1826                                            "$section.svm-source"),
1827                                        replace => tmp_config('--get',
1828                                            "$section.svm-replace"),
1829                                }
1830                        };
1831                        $r->{$_}->{svm} = $svm;
1832                }
1833        } keys %$r;
1834
1835        $r;
1836}
1837
1838sub init_vars {
1839        $_gc_nr = $_gc_period = 1000;
1840        if (defined $_repack || defined $_repack_flags) {
1841               warn "Repack options are obsolete; they have no effect.\n";
1842        }
1843}
1844
1845sub verify_remotes_sanity {
1846        return unless -d $ENV{GIT_DIR};
1847        my %seen;
1848        foreach (command(qw/config -l/)) {
1849                if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
1850                        if ($seen{$1}) {
1851                                die "Remote ref refs/remote/$1 is tracked by",
1852                                    "\n  \"$_\"\nand\n  \"$seen{$1}\"\n",
1853                                    "Please resolve this ambiguity in ",
1854                                    "your git configuration file before ",
1855                                    "continuing\n";
1856                        }
1857                        $seen{$1} = $_;
1858                }
1859        }
1860}
1861
1862sub find_existing_remote {
1863        my ($url, $remotes) = @_;
1864        return undef if $no_reuse_existing;
1865        my $existing;
1866        foreach my $repo_id (keys %$remotes) {
1867                my $u = $remotes->{$repo_id}->{url} or next;
1868                next if $u ne $url;
1869                $existing = $repo_id;
1870                last;
1871        }
1872        $existing;
1873}
1874
1875sub init_remote_config {
1876        my ($self, $url, $no_write) = @_;
1877        $url =~ s!/+$!!; # strip trailing slash
1878        my $r = read_all_remotes();
1879        my $existing = find_existing_remote($url, $r);
1880        if ($existing) {
1881                unless ($no_write) {
1882                        print STDERR "Using existing ",
1883                                     "[svn-remote \"$existing\"]\n";
1884                }
1885                $self->{repo_id} = $existing;
1886        } elsif ($_minimize_url) {
1887                my $min_url = Git::SVN::Ra->new($url)->minimize_url;
1888                $existing = find_existing_remote($min_url, $r);
1889                if ($existing) {
1890                        unless ($no_write) {
1891                                print STDERR "Using existing ",
1892                                             "[svn-remote \"$existing\"]\n";
1893                        }
1894                        $self->{repo_id} = $existing;
1895                }
1896                if ($min_url ne $url) {
1897                        unless ($no_write) {
1898                                print STDERR "Using higher level of URL: ",
1899                                             "$url => $min_url\n";
1900                        }
1901                        my $old_path = $self->{path};
1902                        $self->{path} = $url;
1903                        $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
1904                        if (length $old_path) {
1905                                $self->{path} .= "/$old_path";
1906                        }
1907                        $url = $min_url;
1908                }
1909        }
1910        my $orig_url;
1911        if (!$existing) {
1912                # verify that we aren't overwriting anything:
1913                $orig_url = eval {
1914                        command_oneline('config', '--get',
1915                                        "svn-remote.$self->{repo_id}.url")
1916                };
1917                if ($orig_url && ($orig_url ne $url)) {
1918                        die "svn-remote.$self->{repo_id}.url already set: ",
1919                            "$orig_url\nwanted to set to: $url\n";
1920                }
1921        }
1922        my ($xrepo_id, $xpath) = find_ref($self->refname);
1923        if (!$no_write && defined $xpath) {
1924                die "svn-remote.$xrepo_id.fetch already set to track ",
1925                    "$xpath:", $self->refname, "\n";
1926        }
1927        unless ($no_write) {
1928                command_noisy('config',
1929                              "svn-remote.$self->{repo_id}.url", $url);
1930                $self->{path} =~ s{^/}{};
1931                $self->{path} =~ s{%([0-9A-F]{2})}{chr hex($1)}ieg;
1932                command_noisy('config', '--add',
1933                              "svn-remote.$self->{repo_id}.fetch",
1934                              "$self->{path}:".$self->refname);
1935        }
1936        $self->{url} = $url;
1937}
1938
1939sub find_by_url { # repos_root and, path are optional
1940        my ($class, $full_url, $repos_root, $path) = @_;
1941
1942        return undef unless defined $full_url;
1943        remove_username($full_url);
1944        remove_username($repos_root) if defined $repos_root;
1945        my $remotes = read_all_remotes();
1946        if (defined $full_url && defined $repos_root && !defined $path) {
1947                $path = $full_url;
1948                $path =~ s#^\Q$repos_root\E(?:/|$)##;
1949        }
1950        foreach my $repo_id (keys %$remotes) {
1951                my $u = $remotes->{$repo_id}->{url} or next;
1952                remove_username($u);
1953                next if defined $repos_root && $repos_root ne $u;
1954
1955                my $fetch = $remotes->{$repo_id}->{fetch} || {};
1956                foreach my $t (qw/branches tags/) {
1957                        foreach my $globspec (@{$remotes->{$repo_id}->{$t}}) {
1958                                resolve_local_globs($u, $fetch, $globspec);
1959                        }
1960                }
1961                my $p = $path;
1962                my $rwr = rewrite_root({repo_id => $repo_id});
1963                my $svm = $remotes->{$repo_id}->{svm}
1964                        if defined $remotes->{$repo_id}->{svm};
1965                unless (defined $p) {
1966                        $p = $full_url;
1967                        my $z = $u;
1968                        my $prefix = '';
1969                        if ($rwr) {
1970                                $z = $rwr;
1971                                remove_username($z);
1972                        } elsif (defined $svm) {
1973                                $z = $svm->{source};
1974                                $prefix = $svm->{replace};
1975                                $prefix =~ s#^\Q$u\E(?:/|$)##;
1976                                $prefix =~ s#/$##;
1977                        }
1978                        $p =~ s#^\Q$z\E(?:/|$)#$prefix# or next;
1979                }
1980                foreach my $f (keys %$fetch) {
1981                        next if $f ne $p;
1982                        return Git::SVN->new($fetch->{$f}, $repo_id, $f);
1983                }
1984        }
1985        undef;
1986}
1987
1988sub init {
1989        my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
1990        my $self = _new($class, $repo_id, $ref_id, $path);
1991        if (defined $url) {
1992                $self->init_remote_config($url, $no_write);
1993        }
1994        $self;
1995}
1996
1997sub find_ref {
1998        my ($ref_id) = @_;
1999        foreach (command(qw/config -l/)) {
2000                next unless m!^svn-remote\.(.+)\.fetch=
2001                              \s*(.*?)\s*:\s*(.+?)\s*$!x;
2002                my ($repo_id, $path, $ref) = ($1, $2, $3);
2003                if ($ref eq $ref_id) {
2004                        $path = '' if ($path =~ m#^\./?#);
2005                        return ($repo_id, $path);
2006                }
2007        }
2008        (undef, undef, undef);
2009}
2010
2011sub new {
2012        my ($class, $ref_id, $repo_id, $path) = @_;
2013        if (defined $ref_id && !defined $repo_id && !defined $path) {
2014                ($repo_id, $path) = find_ref($ref_id);
2015                if (!defined $repo_id) {
2016                        die "Could not find a \"svn-remote.*.fetch\" key ",
2017                            "in the repository configuration matching: ",
2018                            "$ref_id\n";
2019                }
2020        }
2021        my $self = _new($class, $repo_id, $ref_id, $path);
2022        if (!defined $self->{path} || !length $self->{path}) {
2023                my $fetch = command_oneline('config', '--get',
2024                                            "svn-remote.$repo_id.fetch",
2025                                            ":$ref_id\$") or
2026                     die "Failed to read \"svn-remote.$repo_id.fetch\" ",
2027                         "\":$ref_id\$\" in config\n";
2028                ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
2029        }
2030        $self->{url} = command_oneline('config', '--get',
2031                                       "svn-remote.$repo_id.url") or
2032                  die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
2033        $self->rebuild;
2034        $self;
2035}
2036
2037sub refname {
2038        my ($refname) = $_[0]->{ref_id} ;
2039
2040        # It cannot end with a slash /, we'll throw up on this because
2041        # SVN can't have directories with a slash in their name, either:
2042        if ($refname =~ m{/$}) {
2043                die "ref: '$refname' ends with a trailing slash, this is ",
2044                    "not permitted by git nor Subversion\n";
2045        }
2046
2047        # It cannot have ASCII control character space, tilde ~, caret ^,
2048        # colon :, question-mark ?, asterisk *, space, or open bracket [
2049        # anywhere.
2050        #
2051        # Additionally, % must be escaped because it is used for escaping
2052        # and we want our escaped refname to be reversible
2053        $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
2054
2055        # no slash-separated component can begin with a dot .
2056        # /.* becomes /%2E*
2057        $refname =~ s{/\.}{/%2E}g;
2058
2059        # It cannot have two consecutive dots .. anywhere
2060        # .. becomes %2E%2E
2061        $refname =~ s{\.\.}{%2E%2E}g;
2062
2063        return $refname;
2064}
2065
2066sub desanitize_refname {
2067        my ($refname) = @_;
2068        $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
2069        return $refname;
2070}
2071
2072sub svm_uuid {
2073        my ($self) = @_;
2074        return $self->{svm}->{uuid} if $self->svm;
2075        $self->ra;
2076        unless ($self->{svm}) {
2077                die "SVM UUID not cached, and reading remotely failed\n";
2078        }
2079        $self->{svm}->{uuid};
2080}
2081
2082sub svm {
2083        my ($self) = @_;
2084        return $self->{svm} if $self->{svm};
2085        my $svm;
2086        # see if we have it in our config, first:
2087        eval {
2088                my $section = "svn-remote.$self->{repo_id}";
2089                $svm = {
2090                  source => tmp_config('--get', "$section.svm-source"),
2091                  uuid => tmp_config('--get', "$section.svm-uuid"),
2092                  replace => tmp_config('--get', "$section.svm-replace"),
2093                }
2094        };
2095        if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
2096                $self->{svm} = $svm;
2097        }
2098        $self->{svm};
2099}
2100
2101sub _set_svm_vars {
2102        my ($self, $ra) = @_;
2103        return $ra if $self->svm;
2104
2105        my @err = ( "useSvmProps set, but failed to read SVM properties\n",
2106                    "(svm:source, svm:uuid) ",
2107                    "from the following URLs:\n" );
2108        sub read_svm_props {
2109                my ($self, $ra, $path, $r) = @_;
2110                my $props = ($ra->get_dir($path, $r))[2];
2111                my $src = $props->{'svm:source'};
2112                my $uuid = $props->{'svm:uuid'};
2113                return undef if (!$src || !$uuid);
2114
2115                chomp($src, $uuid);
2116
2117                $uuid =~ m{^[0-9a-f\-]{30,}$}i
2118                    or die "doesn't look right - svm:uuid is '$uuid'\n";
2119
2120                # the '!' is used to mark the repos_root!/relative/path
2121                $src =~ s{/?!/?}{/};
2122                $src =~ s{/+$}{}; # no trailing slashes please
2123                # username is of no interest
2124                $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
2125
2126                my $replace = $ra->{url};
2127                $replace .= "/$path" if length $path;
2128
2129                my $section = "svn-remote.$self->{repo_id}";
2130                tmp_config("$section.svm-source", $src);
2131                tmp_config("$section.svm-replace", $replace);
2132                tmp_config("$section.svm-uuid", $uuid);
2133                $self->{svm} = {
2134                        source => $src,
2135                        uuid => $uuid,
2136                        replace => $replace
2137                };
2138        }
2139
2140        my $r = $ra->get_latest_revnum;
2141        my $path = $self->{path};
2142        my %tried;
2143        while (length $path) {
2144                unless ($tried{"$self->{url}/$path"}) {
2145                        return $ra if $self->read_svm_props($ra, $path, $r);
2146                        $tried{"$self->{url}/$path"} = 1;
2147                }
2148                $path =~ s#/?[^/]+$##;
2149        }
2150        die "Path: '$path' should be ''\n" if $path ne '';
2151        return $ra if $self->read_svm_props($ra, $path, $r);
2152        $tried{"$self->{url}/$path"} = 1;
2153
2154        if ($ra->{repos_root} eq $self->{url}) {
2155                die @err, (map { "  $_\n" } keys %tried), "\n";
2156        }
2157
2158        # nope, make sure we're connected to the repository root:
2159        my $ok;
2160        my @tried_b;
2161        $path = $ra->{svn_path};
2162        $ra = Git::SVN::Ra->new($ra->{repos_root});
2163        while (length $path) {
2164                unless ($tried{"$ra->{url}/$path"}) {
2165                        $ok = $self->read_svm_props($ra, $path, $r);
2166                        last if $ok;
2167                        $tried{"$ra->{url}/$path"} = 1;
2168                }
2169                $path =~ s#/?[^/]+$##;
2170        }
2171        die "Path: '$path' should be ''\n" if $path ne '';
2172        $ok ||= $self->read_svm_props($ra, $path, $r);
2173        $tried{"$ra->{url}/$path"} = 1;
2174        if (!$ok) {
2175                die @err, (map { "  $_\n" } keys %tried), "\n";
2176        }
2177        Git::SVN::Ra->new($self->{url});
2178}
2179
2180sub svnsync {
2181        my ($self) = @_;
2182        return $self->{svnsync} if $self->{svnsync};
2183
2184        if ($self->no_metadata) {
2185                die "Can't have both 'noMetadata' and ",
2186                    "'useSvnsyncProps' options set!\n";
2187        }
2188        if ($self->rewrite_root) {
2189                die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
2190                    "options set!\n";
2191        }
2192
2193        my $svnsync;
2194        # see if we have it in our config, first:
2195        eval {
2196                my $section = "svn-remote.$self->{repo_id}";
2197
2198                my $url = tmp_config('--get', "$section.svnsync-url");
2199                ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2200                   die "doesn't look right - svn:sync-from-url is '$url'\n";
2201
2202                my $uuid = tmp_config('--get', "$section.svnsync-uuid");
2203                ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2204                   die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2205
2206                $svnsync = { url => $url, uuid => $uuid }
2207        };
2208        if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
2209                return $self->{svnsync} = $svnsync;
2210        }
2211
2212        my $err = "useSvnsyncProps set, but failed to read " .
2213                  "svnsync property: svn:sync-from-";
2214        my $rp = $self->ra->rev_proplist(0);
2215
2216        my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
2217        ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2218                   die "doesn't look right - svn:sync-from-url is '$url'\n";
2219
2220        my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
2221        ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2222                   die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2223
2224        my $section = "svn-remote.$self->{repo_id}";
2225        tmp_config('--add', "$section.svnsync-uuid", $uuid);
2226        tmp_config('--add', "$section.svnsync-url", $url);
2227        return $self->{svnsync} = { url => $url, uuid => $uuid };
2228}
2229
2230# this allows us to memoize our SVN::Ra UUID locally and avoid a
2231# remote lookup (useful for 'git svn log').
2232sub ra_uuid {
2233        my ($self) = @_;
2234        unless ($self->{ra_uuid}) {
2235                my $key = "svn-remote.$self->{repo_id}.uuid";
2236                my $uuid = eval { tmp_config('--get', $key) };
2237                if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/i) {
2238                        $self->{ra_uuid} = $uuid;
2239                } else {
2240                        die "ra_uuid called without URL\n" unless $self->{url};
2241                        $self->{ra_uuid} = $self->ra->get_uuid;
2242                        tmp_config('--add', $key, $self->{ra_uuid});
2243                }
2244        }
2245        $self->{ra_uuid};
2246}
2247
2248sub _set_repos_root {
2249        my ($self, $repos_root) = @_;
2250        my $k = "svn-remote.$self->{repo_id}.reposRoot";
2251        $repos_root ||= $self->ra->{repos_root};
2252        tmp_config($k, $repos_root);
2253        $repos_root;
2254}
2255
2256sub repos_root {
2257        my ($self) = @_;
2258        my $k = "svn-remote.$self->{repo_id}.reposRoot";
2259        eval { tmp_config('--get', $k) } || $self->_set_repos_root;
2260}
2261
2262sub ra {
2263        my ($self) = shift;
2264        my $ra = Git::SVN::Ra->new($self->{url});
2265        $self->_set_repos_root($ra->{repos_root});
2266        if ($self->use_svm_props && !$self->{svm}) {
2267                if ($self->no_metadata) {
2268                        die "Can't have both 'noMetadata' and ",
2269                            "'useSvmProps' options set!\n";
2270                } elsif ($self->use_svnsync_props) {
2271                        die "Can't have both 'useSvnsyncProps' and ",
2272                            "'useSvmProps' options set!\n";
2273                }
2274                $ra = $self->_set_svm_vars($ra);
2275                $self->{-want_revprops} = 1;
2276        }
2277        $ra;
2278}
2279
2280# prop_walk(PATH, REV, SUB)
2281# -------------------------
2282# Recursively traverse PATH at revision REV and invoke SUB for each
2283# directory that contains a SVN property.  SUB will be invoked as
2284# follows:  &SUB(gs, path, props);  where `gs' is this instance of
2285# Git::SVN, `path' the path to the directory where the properties
2286# `props' were found.  The `path' will be relative to point of checkout,
2287# that is, if url://repo/trunk is the current Git branch, and that
2288# directory contains a sub-directory `d', SUB will be invoked with `/d/'
2289# as `path' (note the trailing `/').
2290sub prop_walk {
2291        my ($self, $path, $rev, $sub) = @_;
2292
2293        $path =~ s#^/##;
2294        my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev);
2295        $path =~ s#^/*#/#g;
2296        my $p = $path;
2297        # Strip the irrelevant part of the path.
2298        $p =~ s#^/+\Q$self->{path}\E(/|$)#/#;
2299        # Ensure the path is terminated by a `/'.
2300        $p =~ s#/*$#/#;
2301
2302        # The properties contain all the internal SVN stuff nobody
2303        # (usually) cares about.
2304        my $interesting_props = 0;
2305        foreach (keys %{$props}) {
2306                # If it doesn't start with `svn:', it must be a
2307                # user-defined property.
2308                ++$interesting_props and next if $_ !~ /^svn:/;
2309                # FIXME: Fragile, if SVN adds new public properties,
2310                # this needs to be updated.
2311                ++$interesting_props if /^svn:(?:ignore|keywords|executable
2312                                                 |eol-style|mime-type
2313                                                 |externals|needs-lock)$/x;
2314        }
2315        &$sub($self, $p, $props) if $interesting_props;
2316
2317        foreach (sort keys %$dirent) {
2318                next if $dirent->{$_}->{kind} != $SVN::Node::dir;
2319                $self->prop_walk($self->{path} . $p . $_, $rev, $sub);
2320        }
2321}
2322
2323sub last_rev { ($_[0]->last_rev_commit)[0] }
2324sub last_commit { ($_[0]->last_rev_commit)[1] }
2325
2326# returns the newest SVN revision number and newest commit SHA1
2327sub last_rev_commit {
2328        my ($self) = @_;
2329        if (defined $self->{last_rev} && defined $self->{last_commit}) {
2330                return ($self->{last_rev}, $self->{last_commit});
2331        }
2332        my $c = ::verify_ref($self->refname.'^0');
2333        if ($c && !$self->use_svm_props && !$self->no_metadata) {
2334                my $rev = (::cmt_metadata($c))[1];
2335                if (defined $rev) {
2336                        ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
2337                        return ($rev, $c);
2338                }
2339        }
2340        my $map_path = $self->map_path;
2341        unless (-e $map_path) {
2342                ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
2343                return (undef, undef);
2344        }
2345        my ($rev, $commit) = $self->rev_map_max(1);
2346        ($self->{last_rev}, $self->{last_commit}) = ($rev, $commit);
2347        return ($rev, $commit);
2348}
2349
2350sub get_fetch_range {
2351        my ($self, $min, $max) = @_;
2352        $max ||= $self->ra->get_latest_revnum;
2353        $min ||= $self->rev_map_max;
2354        (++$min, $max);
2355}
2356
2357sub tmp_config {
2358        my (@args) = @_;
2359        my $old_def_config = "$ENV{GIT_DIR}/svn/config";
2360        my $config = "$ENV{GIT_DIR}/svn/.metadata";
2361        if (! -f $config && -f $old_def_config) {
2362                rename $old_def_config, $config or
2363                       die "Failed rename $old_def_config => $config: $!\n";
2364        }
2365        my $old_config = $ENV{GIT_CONFIG};
2366        $ENV{GIT_CONFIG} = $config;
2367        $@ = undef;
2368        my @ret = eval {
2369                unless (-f $config) {
2370                        mkfile($config);
2371                        open my $fh, '>', $config or
2372                            die "Can't open $config: $!\n";
2373                        print $fh "; This file is used internally by ",
2374                                  "git-svn\n" or die
2375                                  "Couldn't write to $config: $!\n";
2376                        print $fh "; You should not have to edit it\n" or
2377                              die "Couldn't write to $config: $!\n";
2378                        close $fh or die "Couldn't close $config: $!\n";
2379                }
2380                command('config', @args);
2381        };
2382        my $err = $@;
2383        if (defined $old_config) {
2384                $ENV{GIT_CONFIG} = $old_config;
2385        } else {
2386                delete $ENV{GIT_CONFIG};
2387        }
2388        die $err if $err;
2389        wantarray ? @ret : $ret[0];
2390}
2391
2392sub tmp_index_do {
2393        my ($self, $sub) = @_;
2394        my $old_index = $ENV{GIT_INDEX_FILE};
2395        $ENV{GIT_INDEX_FILE} = $self->{index};
2396        $@ = undef;
2397        my @ret = eval {
2398                my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
2399                mkpath([$dir]) unless -d $dir;
2400                &$sub;
2401        };
2402        my $err = $@;
2403        if (defined $old_index) {
2404                $ENV{GIT_INDEX_FILE} = $old_index;
2405        } else {
2406                delete $ENV{GIT_INDEX_FILE};
2407        }
2408        die $err if $err;
2409        wantarray ? @ret : $ret[0];
2410}
2411
2412sub assert_index_clean {
2413        my ($self, $treeish) = @_;
2414
2415        $self->tmp_index_do(sub {
2416                command_noisy('read-tree', $treeish) unless -e $self->{index};
2417                my $x = command_oneline('write-tree');
2418                my ($y) = (command(qw/cat-file commit/, $treeish) =~
2419                           /^tree ($::sha1)/mo);
2420                return if $y eq $x;
2421
2422                warn "Index mismatch: $y != $x\nrereading $treeish\n";
2423                unlink $self->{index} or die "unlink $self->{index}: $!\n";
2424                command_noisy('read-tree', $treeish);
2425                $x = command_oneline('write-tree');
2426                if ($y ne $x) {
2427                        ::fatal "trees ($treeish) $y != $x\n",
2428                                "Something is seriously wrong...";
2429                }
2430        });
2431}
2432
2433sub get_commit_parents {
2434        my ($self, $log_entry) = @_;
2435        my (%seen, @ret, @tmp);
2436        # legacy support for 'set-tree'; this is only used by set_tree_cb:
2437        if (my $ip = $self->{inject_parents}) {
2438                if (my $commit = delete $ip->{$log_entry->{revision}}) {
2439                        push @tmp, $commit;
2440                }
2441        }
2442        if (my $cur = ::verify_ref($self->refname.'^0')) {
2443                push @tmp, $cur;
2444        }
2445        if (my $ipd = $self->{inject_parents_dcommit}) {
2446                if (my $commit = delete $ipd->{$log_entry->{revision}}) {
2447                        push @tmp, @$commit;
2448                }
2449        }
2450        push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
2451        while (my $p = shift @tmp) {
2452                next if $seen{$p};
2453                $seen{$p} = 1;
2454                push @ret, $p;
2455        }
2456        @ret;
2457}
2458
2459sub rewrite_root {
2460        my ($self) = @_;
2461        return $self->{-rewrite_root} if exists $self->{-rewrite_root};
2462        my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
2463        my $rwr = eval { command_oneline(qw/config --get/, $k) };
2464        if ($rwr) {
2465                $rwr =~ s#/+$##;
2466                if ($rwr !~ m#^[a-z\+]+://#) {
2467                        die "$rwr is not a valid URL (key: $k)\n";
2468                }
2469        }
2470        $self->{-rewrite_root} = $rwr;
2471}
2472
2473sub metadata_url {
2474        my ($self) = @_;
2475        ($self->rewrite_root || $self->{url}) .
2476           (length $self->{path} ? '/' . $self->{path} : '');
2477}
2478
2479sub full_url {
2480        my ($self) = @_;
2481        $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
2482}
2483
2484
2485sub set_commit_header_env {
2486        my ($log_entry) = @_;
2487        my %env;
2488        foreach my $ned (qw/NAME EMAIL DATE/) {
2489                foreach my $ac (qw/AUTHOR COMMITTER/) {
2490                        $env{"GIT_${ac}_${ned}"} = $ENV{"GIT_${ac}_${ned}"};
2491                }
2492        }
2493
2494        $ENV{GIT_AUTHOR_NAME} = $log_entry->{name};
2495        $ENV{GIT_AUTHOR_EMAIL} = $log_entry->{email};
2496        $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
2497
2498        $ENV{GIT_COMMITTER_NAME} = (defined $log_entry->{commit_name})
2499                                                ? $log_entry->{commit_name}
2500                                                : $log_entry->{name};
2501        $ENV{GIT_COMMITTER_EMAIL} = (defined $log_entry->{commit_email})
2502                                                ? $log_entry->{commit_email}
2503                                                : $log_entry->{email};
2504        \%env;
2505}
2506
2507sub restore_commit_header_env {
2508        my ($env) = @_;
2509        foreach my $ned (qw/NAME EMAIL DATE/) {
2510                foreach my $ac (qw/AUTHOR COMMITTER/) {
2511                        my $k = "GIT_${ac}_${ned}";
2512                        if (defined $env->{$k}) {
2513                                $ENV{$k} = $env->{$k};
2514                        } else {
2515                                delete $ENV{$k};
2516                        }
2517                }
2518        }
2519}
2520
2521sub gc {
2522        command_noisy('gc', '--auto');
2523};
2524
2525sub do_git_commit {
2526        my ($self, $log_entry) = @_;
2527        my $lr = $self->last_rev;
2528        if (defined $lr && $lr >= $log_entry->{revision}) {
2529                die "Last fetched revision of ", $self->refname,
2530                    " was r$lr, but we are about to fetch: ",
2531                    "r$log_entry->{revision}!\n";
2532        }
2533        if (my $c = $self->rev_map_get($log_entry->{revision})) {
2534                croak "$log_entry->{revision} = $c already exists! ",
2535                      "Why are we refetching it?\n";
2536        }
2537        my $old_env = set_commit_header_env($log_entry);
2538        my $tree = $log_entry->{tree};
2539        if (!defined $tree) {
2540                $tree = $self->tmp_index_do(sub {
2541                                            command_oneline('write-tree') });
2542        }
2543        die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
2544
2545        my @exec = ('git', 'commit-tree', $tree);
2546        foreach ($self->get_commit_parents($log_entry)) {
2547                push @exec, '-p', $_;
2548        }
2549        defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
2550                                                                   or croak $!;
2551        binmode $msg_fh;
2552
2553        # we always get UTF-8 from SVN, but we may want our commits in
2554        # a different encoding.
2555        if (my $enc = Git::config('i18n.commitencoding')) {
2556                require Encode;
2557                Encode::from_to($log_entry->{log}, 'UTF-8', $enc);
2558        }
2559        print $msg_fh $log_entry->{log} or croak $!;
2560        restore_commit_header_env($old_env);
2561        unless ($self->no_metadata) {
2562                print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
2563                              or croak $!;
2564        }
2565        $msg_fh->flush == 0 or croak $!;
2566        close $msg_fh or croak $!;
2567        chomp(my $commit = do { local $/; <$out_fh> });
2568        close $out_fh or croak $!;
2569        waitpid $pid, 0;
2570        croak $? if $?;
2571        if ($commit !~ /^$::sha1$/o) {
2572                die "Failed to commit, invalid sha1: $commit\n";
2573        }
2574
2575        $self->rev_map_set($log_entry->{revision}, $commit, 1);
2576
2577        $self->{last_rev} = $log_entry->{revision};
2578        $self->{last_commit} = $commit;
2579        print "r$log_entry->{revision}" unless $::_q > 1;
2580        if (defined $log_entry->{svm_revision}) {
2581                 print " (\@$log_entry->{svm_revision})" unless $::_q > 1;
2582                 $self->rev_map_set($log_entry->{svm_revision}, $commit,
2583                                   0, $self->svm_uuid);
2584        }
2585        print " = $commit ($self->{ref_id})\n" unless $::_q > 1;
2586        if (--$_gc_nr == 0) {
2587                $_gc_nr = $_gc_period;
2588                gc();
2589        }
2590        return $commit;
2591}
2592
2593sub match_paths {
2594        my ($self, $paths, $r) = @_;
2595        return 1 if $self->{path} eq '';
2596        if (my $path = $paths->{"/$self->{path}"}) {
2597                return ($path->{action} eq 'D') ? 0 : 1;
2598        }
2599        $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
2600        if (grep /$self->{path_regex}/, keys %$paths) {
2601                return 1;
2602        }
2603        my $c = '';
2604        foreach (split m#/#, $self->{path}) {
2605                $c .= "/$_";
2606                next unless ($paths->{$c} &&
2607                             ($paths->{$c}->{action} =~ /^[AR]$/));
2608                if ($self->ra->check_path($self->{path}, $r) ==
2609                    $SVN::Node::dir) {
2610                        return 1;
2611                }
2612        }
2613        return 0;
2614}
2615
2616sub find_parent_branch {
2617        my ($self, $paths, $rev) = @_;
2618        return undef unless $self->follow_parent;
2619        unless (defined $paths) {
2620                my $err_handler = $SVN::Error::handler;
2621                $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
2622                $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1,
2623                                   sub { $paths = $_[0] });
2624                $SVN::Error::handler = $err_handler;
2625        }
2626        return undef unless defined $paths;
2627
2628        # look for a parent from another branch:
2629        my @b_path_components = split m#/#, $self->{path};
2630        my @a_path_components;
2631        my $i;
2632        while (@b_path_components) {
2633                $i = $paths->{'/'.join('/', @b_path_components)};
2634                last if $i && defined $i->{copyfrom_path};
2635                unshift(@a_path_components, pop(@b_path_components));
2636        }
2637        return undef unless defined $i && defined $i->{copyfrom_path};
2638        my $branch_from = $i->{copyfrom_path};
2639        if (@a_path_components) {
2640                print STDERR "branch_from: $branch_from => ";
2641                $branch_from .= '/'.join('/', @a_path_components);
2642                print STDERR $branch_from, "\n";
2643        }
2644        my $r = $i->{copyfrom_rev};
2645        my $repos_root = $self->ra->{repos_root};
2646        my $url = $self->ra->{url};
2647        my $new_url = $url . $branch_from;
2648        print STDERR  "Found possible branch point: ",
2649                      "$new_url => ", $self->full_url, ", $r\n"
2650                      unless $::_q > 1;
2651        $branch_from =~ s#^/##;
2652        my $gs = $self->other_gs($new_url, $url,
2653                                 $branch_from, $r, $self->{ref_id});
2654        my ($r0, $parent) = $gs->find_rev_before($r, 1);
2655        {
2656                my ($base, $head);
2657                if (!defined $r0 || !defined $parent) {
2658                        ($base, $head) = parse_revision_argument(0, $r);
2659                } else {
2660                        if ($r0 < $r) {
2661                                $gs->ra->get_log([$gs->{path}], $r0 + 1, $r, 1,
2662                                        0, 1, sub { $base = $_[1] - 1 });
2663                        }
2664                }
2665                if (defined $base && $base <= $r) {
2666                        $gs->fetch($base, $r);
2667                }
2668                ($r0, $parent) = $gs->find_rev_before($r, 1);
2669        }
2670        if (defined $r0 && defined $parent) {
2671                print STDERR "Found branch parent: ($self->{ref_id}) $parent\n"
2672                             unless $::_q > 1;
2673                my $ed;
2674                if ($self->ra->can_do_switch) {
2675                        $self->assert_index_clean($parent);
2676                        print STDERR "Following parent with do_switch\n"
2677                                     unless $::_q > 1;
2678                        # do_switch works with svn/trunk >= r22312, but that
2679                        # is not included with SVN 1.4.3 (the latest version
2680                        # at the moment), so we can't rely on it
2681                        $self->{last_rev} = $r0;
2682                        $self->{last_commit} = $parent;
2683                        $ed = SVN::Git::Fetcher->new($self, $gs->{path});
2684                        $gs->ra->gs_do_switch($r0, $rev, $gs,
2685                                              $self->full_url, $ed)
2686                          or die "SVN connection failed somewhere...\n";
2687                } elsif ($self->ra->trees_match($new_url, $r0,
2688                                                $self->full_url, $rev)) {
2689                        print STDERR "Trees match:\n",
2690                                     "  $new_url\@$r0\n",
2691                                     "  ${\$self->full_url}\@$rev\n",
2692                                     "Following parent with no changes\n"
2693                                     unless $::_q > 1;
2694                        $self->tmp_index_do(sub {
2695                            command_noisy('read-tree', $parent);
2696                        });
2697                        $self->{last_commit} = $parent;
2698                } else {
2699                        print STDERR "Following parent with do_update\n"
2700                                     unless $::_q > 1;
2701                        $ed = SVN::Git::Fetcher->new($self);
2702                        $self->ra->gs_do_update($rev, $rev, $self, $ed)
2703                          or die "SVN connection failed somewhere...\n";
2704                }
2705                print STDERR "Successfully followed parent\n" unless $::_q > 1;
2706                return $self->make_log_entry($rev, [$parent], $ed);
2707        }
2708        return undef;
2709}
2710
2711sub do_fetch {
2712        my ($self, $paths, $rev) = @_;
2713        my $ed;
2714        my ($last_rev, @parents);
2715        if (my $lc = $self->last_commit) {
2716                # we can have a branch that was deleted, then re-added
2717                # under the same name but copied from another path, in
2718                # which case we'll have multiple parents (we don't
2719                # want to break the original ref, nor lose copypath info):
2720                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2721                        push @{$log_entry->{parents}}, $lc;
2722                        return $log_entry;
2723                }
2724                $ed = SVN::Git::Fetcher->new($self);
2725                $last_rev = $self->{last_rev};
2726                $ed->{c} = $lc;
2727                @parents = ($lc);
2728        } else {
2729                $last_rev = $rev;
2730                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2731                        return $log_entry;
2732                }
2733                $ed = SVN::Git::Fetcher->new($self);
2734        }
2735        unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
2736                die "SVN connection failed somewhere...\n";
2737        }
2738        $self->make_log_entry($rev, \@parents, $ed);
2739}
2740
2741sub mkemptydirs {
2742        my ($self, $r) = @_;
2743
2744        sub scan {
2745                my ($r, $empty_dirs, $line) = @_;
2746                if (defined $r && $line =~ /^r(\d+)$/) {
2747                        return 0 if $1 > $r;
2748                } elsif ($line =~ /^  \+empty_dir: (.+)$/) {
2749                        $empty_dirs->{$1} = 1;
2750                } elsif ($line =~ /^  \-empty_dir: (.+)$/) {
2751                        my @d = grep {m[^\Q$1\E(/|$)]} (keys %$empty_dirs);
2752                        delete @$empty_dirs{@d};
2753                }
2754                1; # continue
2755        };
2756
2757        my %empty_dirs = ();
2758        my $gz_file = "$self->{dir}/unhandled.log.gz";
2759        if (-f $gz_file) {
2760                if (!$can_compress) {
2761                        warn "Compress::Zlib could not be found; ",
2762                             "empty directories in $gz_file will not be read\n";
2763                } else {
2764                        my $gz = Compress::Zlib::gzopen($gz_file, "rb") or
2765                                die "Unable to open $gz_file: $!\n";
2766                        my $line;
2767                        while ($gz->gzreadline($line) > 0) {
2768                                scan($r, \%empty_dirs, $line) or last;
2769                        }
2770                        $gz->gzclose;
2771                }
2772        }
2773
2774        if (open my $fh, '<', "$self->{dir}/unhandled.log") {
2775                binmode $fh or croak "binmode: $!";
2776                while (<$fh>) {
2777                        scan($r, \%empty_dirs, $_) or last;
2778                }
2779                close $fh;
2780        }
2781
2782        my $strip = qr/\A\Q$self->{path}\E(?:\/|$)/;
2783        foreach my $d (sort keys %empty_dirs) {
2784                $d = uri_decode($d);
2785                $d =~ s/$strip//;
2786                next if -d $d;
2787                if (-e _) {
2788                        warn "$d exists but is not a directory\n";
2789                } else {
2790                        print "creating empty directory: $d\n";
2791                        mkpath([$d]);
2792                }
2793        }
2794}
2795
2796sub get_untracked {
2797        my ($self, $ed) = @_;
2798        my @out;
2799        my $h = $ed->{empty};
2800        foreach (sort keys %$h) {
2801                my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
2802                push @out, "  $act: " . uri_encode($_);
2803                warn "W: $act: $_\n";
2804        }
2805        foreach my $t (qw/dir_prop file_prop/) {
2806                $h = $ed->{$t} or next;
2807                foreach my $path (sort keys %$h) {
2808                        my $ppath = $path eq '' ? '.' : $path;
2809                        foreach my $prop (sort keys %{$h->{$path}}) {
2810                                next if $SKIP_PROP{$prop};
2811                                my $v = $h->{$path}->{$prop};
2812                                my $t_ppath_prop = "$t: " .
2813                                                    uri_encode($ppath) . ' ' .
2814                                                    uri_encode($prop);
2815                                if (defined $v) {
2816                                        push @out, "  +$t_ppath_prop " .
2817                                                   uri_encode($v);
2818                                } else {
2819                                        push @out, "  -$t_ppath_prop";
2820                                }
2821                        }
2822                }
2823        }
2824        foreach my $t (qw/absent_file absent_directory/) {
2825                $h = $ed->{$t} or next;
2826                foreach my $parent (sort keys %$h) {
2827                        foreach my $path (sort @{$h->{$parent}}) {
2828                                push @out, "  $t: " .
2829                                           uri_encode("$parent/$path");
2830                                warn "W: $t: $parent/$path ",
2831                                     "Insufficient permissions?\n";
2832                        }
2833                }
2834        }
2835        \@out;
2836}
2837
2838# parse_svn_date(DATE)
2839# --------------------
2840# Given a date (in UTC) from Subversion, return a string in the format
2841# "<TZ Offset> <local date/time>" that Git will use.
2842#
2843# By default the parsed date will be in UTC; if $Git::SVN::_localtime
2844# is true we'll convert it to the local timezone instead.
2845sub parse_svn_date {
2846        my $date = shift || return '+0000 1970-01-01 00:00:00';
2847        my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2848                                            (\d\d)\:(\d\d)\:(\d\d)\.\d*Z$/x) or
2849                                         croak "Unable to parse date: $date\n";
2850        my $parsed_date;    # Set next.
2851
2852        if ($Git::SVN::_localtime) {
2853                # Translate the Subversion datetime to an epoch time.
2854                # Begin by switching ourselves to $date's timezone, UTC.
2855                my $old_env_TZ = $ENV{TZ};
2856                $ENV{TZ} = 'UTC';
2857
2858                my $epoch_in_UTC =
2859                    POSIX::strftime('%s', $S, $M, $H, $d, $m - 1, $Y - 1900);
2860
2861                # Determine our local timezone (including DST) at the
2862                # time of $epoch_in_UTC.  $Git::SVN::Log::TZ stored the
2863                # value of TZ, if any, at the time we were run.
2864                if (defined $Git::SVN::Log::TZ) {
2865                        $ENV{TZ} = $Git::SVN::Log::TZ;
2866                } else {
2867                        delete $ENV{TZ};
2868                }
2869
2870                my $our_TZ =
2871                    POSIX::strftime('%Z', $S, $M, $H, $d, $m - 1, $Y - 1900);
2872
2873                # This converts $epoch_in_UTC into our local timezone.
2874                my ($sec, $min, $hour, $mday, $mon, $year,
2875                    $wday, $yday, $isdst) = localtime($epoch_in_UTC);
2876
2877                $parsed_date = sprintf('%s %04d-%02d-%02d %02d:%02d:%02d',
2878                                       $our_TZ, $year + 1900, $mon + 1,
2879                                       $mday, $hour, $min, $sec);
2880
2881                # Reset us to the timezone in effect when we entered
2882                # this routine.
2883                if (defined $old_env_TZ) {
2884                        $ENV{TZ} = $old_env_TZ;
2885                } else {
2886                        delete $ENV{TZ};
2887                }
2888        } else {
2889                $parsed_date = "+0000 $Y-$m-$d $H:$M:$S";
2890        }
2891
2892        return $parsed_date;
2893}
2894
2895sub other_gs {
2896        my ($self, $new_url, $url,
2897            $branch_from, $r, $old_ref_id) = @_;
2898        my $gs = Git::SVN->find_by_url($new_url, $url, $branch_from);
2899        unless ($gs) {
2900                my $ref_id = $old_ref_id;
2901                $ref_id =~ s/\@\d+$//;
2902                $ref_id .= "\@$r";
2903                # just grow a tail if we're not unique enough :x
2904                $ref_id .= '-' while find_ref($ref_id);
2905                print STDERR "Initializing parent: $ref_id\n" unless $::_q > 1;
2906                my ($u, $p, $repo_id) = ($new_url, '', $ref_id);
2907                if ($u =~ s#^\Q$url\E(/|$)##) {
2908                        $p = $u;
2909                        $u = $url;
2910                        $repo_id = $self->{repo_id};
2911                }
2912                $gs = Git::SVN->init($u, $p, $repo_id, $ref_id, 1);
2913        }
2914        $gs
2915}
2916
2917sub call_authors_prog {
2918        my ($orig_author) = @_;
2919        $orig_author = command_oneline('rev-parse', '--sq-quote', $orig_author);
2920        my $author = `$::_authors_prog $orig_author`;
2921        if ($? != 0) {
2922                die "$::_authors_prog failed with exit code $?\n"
2923        }
2924        if ($author =~ /^\s*(.+?)\s*<(.*)>\s*$/) {
2925                my ($name, $email) = ($1, $2);
2926                $email = undef if length $2 == 0;
2927                return [$name, $email];
2928        } else {
2929                die "Author: $orig_author: $::_authors_prog returned "
2930                        . "invalid author format: $author\n";
2931        }
2932}
2933
2934sub check_author {
2935        my ($author) = @_;
2936        if (!defined $author || length $author == 0) {
2937                $author = '(no author)';
2938        }
2939        if (!defined $::users{$author}) {
2940                if (defined $::_authors_prog) {
2941                        $::users{$author} = call_authors_prog($author);
2942                } elsif (defined $::_authors) {
2943                        die "Author: $author not defined in $::_authors file\n";
2944                }
2945        }
2946        $author;
2947}
2948
2949sub find_extra_svk_parents {
2950        my ($self, $ed, $tickets, $parents) = @_;
2951        # aha!  svk:merge property changed...
2952        my @tickets = split "\n", $tickets;
2953        my @known_parents;
2954        for my $ticket ( @tickets ) {
2955                my ($uuid, $path, $rev) = split /:/, $ticket;
2956                if ( $uuid eq $self->ra_uuid ) {
2957                        my $url = $self->rewrite_root || $self->{url};
2958                        my $repos_root = $url;
2959                        my $branch_from = $path;
2960                        $branch_from =~ s{^/}{};
2961                        my $gs = $self->other_gs($repos_root."/".$branch_from,
2962                                                 $url,
2963                                                 $branch_from,
2964                                                 $rev,
2965                                                 $self->{ref_id});
2966                        if ( my $commit = $gs->rev_map_get($rev, $uuid) ) {
2967                                # wahey!  we found it, but it might be
2968                                # an old one (!)
2969                                push @known_parents, [ $rev, $commit ];
2970                        }
2971                }
2972        }
2973        # Ordering matters; highest-numbered commit merge tickets
2974        # first, as they may account for later merge ticket additions
2975        # or changes.
2976        @known_parents = map {$_->[1]} sort {$b->[0] <=> $a->[0]} @known_parents;
2977        for my $parent ( @known_parents ) {
2978                my @cmd = ('rev-list', $parent, map { "^$_" } @$parents );
2979                my ($msg_fh, $ctx) = command_output_pipe(@cmd);
2980                my $new;
2981                while ( <$msg_fh> ) {
2982                        $new=1;last;
2983                }
2984                command_close_pipe($msg_fh, $ctx);
2985                if ( $new ) {
2986                        print STDERR
2987                            "Found merge parent (svk:merge ticket): $parent\n";
2988                        push @$parents, $parent;
2989                }
2990        }
2991}
2992
2993sub lookup_svn_merge {
2994        my $uuid = shift;
2995        my $url = shift;
2996        my $merge = shift;
2997
2998        my ($source, $revs) = split ":", $merge;
2999        my $path = $source;
3000        $path =~ s{^/}{};
3001        my $gs = Git::SVN->find_by_url($url.$source, $url, $path);
3002        if ( !$gs ) {
3003                warn "Couldn't find revmap for $url$source\n";
3004                return;
3005        }
3006        my @ranges = split ",", $revs;
3007        my ($tip, $tip_commit);
3008        my @merged_commit_ranges;
3009        # find the tip
3010        for my $range ( @ranges ) {
3011                my ($bottom, $top) = split "-", $range;
3012                $top ||= $bottom;
3013                my $bottom_commit = $gs->find_rev_after( $bottom, 1, $top );
3014                my $top_commit = $gs->find_rev_before( $top, 1, $bottom );
3015
3016                unless ($top_commit and $bottom_commit) {
3017                        warn "W:unknown path/rev in svn:mergeinfo "
3018                                ."dirprop: $source:$range\n";
3019                        next;
3020                }
3021
3022                push @merged_commit_ranges,
3023                        "$bottom_commit^..$top_commit";
3024
3025                if ( !defined $tip or $top > $tip ) {
3026                        $tip = $top;
3027                        $tip_commit = $top_commit;
3028                }
3029        }
3030        return ($tip_commit, @merged_commit_ranges);
3031}
3032
3033sub _rev_list {
3034        my ($msg_fh, $ctx) = command_output_pipe(
3035                "rev-list", @_,
3036               );
3037        my @rv;
3038        while ( <$msg_fh> ) {
3039                chomp;
3040                push @rv, $_;
3041        }
3042        command_close_pipe($msg_fh, $ctx);
3043        @rv;
3044}
3045
3046sub check_cherry_pick {
3047        my $base = shift;
3048        my $tip = shift;
3049        my @ranges = @_;
3050        my %commits = map { $_ => 1 }
3051                _rev_list("--no-merges", $tip, "--not", $base);
3052        for my $range ( @ranges ) {
3053                delete @commits{_rev_list($range)};
3054        }
3055        return (keys %commits);
3056}
3057
3058BEGIN {
3059        memoize 'lookup_svn_merge';
3060        memoize 'check_cherry_pick';
3061}
3062
3063sub parents_exclude {
3064        my $parents = shift;
3065        my @commits = @_;
3066        return unless @commits;
3067
3068        my @excluded;
3069        my $excluded;
3070        do {
3071                my @cmd = ('rev-list', "-1", @commits, "--not", @$parents );
3072                $excluded = command_oneline(@cmd);
3073                if ( $excluded ) {
3074                        my @new;
3075                        my $found;
3076                        for my $commit ( @commits ) {
3077                                if ( $commit eq $excluded ) {
3078                                        push @excluded, $commit;
3079                                        $found++;
3080                                        last;
3081                                }
3082                                else {
3083                                        push @new, $commit;
3084                                }
3085                        }
3086                        die "saw commit '$excluded' in rev-list output, "
3087                                ."but we didn't ask for that commit (wanted: @commits --not @$parents)"
3088                                        unless $found;
3089                        @commits = @new;
3090                }
3091        }
3092                while ($excluded and @commits);
3093
3094        return @excluded;
3095}
3096
3097
3098# note: this function should only be called if the various dirprops
3099# have actually changed
3100sub find_extra_svn_parents {
3101        my ($self, $ed, $mergeinfo, $parents) = @_;
3102        # aha!  svk:merge property changed...
3103
3104        # We first search for merged tips which are not in our
3105        # history.  Then, we figure out which git revisions are in
3106        # that tip, but not this revision.  If all of those revisions
3107        # are now marked as merge, we can add the tip as a parent.
3108        my @merges = split "\n", $mergeinfo;
3109        my @merge_tips;
3110        my $url = $self->rewrite_root || $self->{url};
3111        my $uuid = $self->ra_uuid;
3112        my %ranges;
3113        for my $merge ( @merges ) {
3114                my ($tip_commit, @ranges) =
3115                        lookup_svn_merge( $uuid, $url, $merge );
3116                unless (!$tip_commit or
3117                                grep { $_ eq $tip_commit } @$parents ) {
3118                        push @merge_tips, $tip_commit;
3119                        $ranges{$tip_commit} = \@ranges;
3120                } else {
3121                        push @merge_tips, undef;
3122                }
3123        }
3124
3125        my %excluded = map { $_ => 1 }
3126                parents_exclude($parents, grep { defined } @merge_tips);
3127
3128        # check merge tips for new parents
3129        my @new_parents;
3130        for my $merge_tip ( @merge_tips ) {
3131                my $spec = shift @merges;
3132                next unless $merge_tip and $excluded{$merge_tip};
3133
3134                my $ranges = $ranges{$merge_tip};
3135
3136                # check out 'new' tips
3137                my $merge_base = command_oneline(
3138                        "merge-base",
3139                        @$parents, $merge_tip,
3140                       );
3141
3142                # double check that there are no missing non-merge commits
3143                my (@incomplete) = check_cherry_pick(
3144                        $merge_base, $merge_tip,
3145                        @$ranges,
3146                       );
3147
3148                if ( @incomplete ) {
3149                        warn "W:svn cherry-pick ignored ($spec) - missing "
3150                                .@incomplete." commit(s) (eg $incomplete[0])\n";
3151                } else {
3152                        warn
3153                                "Found merge parent (svn:mergeinfo prop): ",
3154                                        $merge_tip, "\n";
3155                        push @new_parents, $merge_tip;
3156                }
3157        }
3158
3159        # cater for merges which merge commits from multiple branches
3160        if ( @new_parents > 1 ) {
3161                for ( my $i = 0; $i <= $#new_parents; $i++ ) {
3162                        for ( my $j = 0; $j <= $#new_parents; $j++ ) {
3163                                next if $i == $j;
3164                                next unless $new_parents[$i];
3165                                next unless $new_parents[$j];
3166                                my $revs = command_oneline(
3167                                        "rev-list", "-1",
3168                                        "$new_parents[$i]..$new_parents[$j]",
3169                                       );
3170                                if ( !$revs ) {
3171                                        undef($new_parents[$i]);
3172                                }
3173                        }
3174                }
3175        }
3176        push @$parents, grep { defined } @new_parents;
3177}
3178
3179sub make_log_entry {
3180        my ($self, $rev, $parents, $ed) = @_;
3181        my $untracked = $self->get_untracked($ed);
3182
3183        my @parents = @$parents;
3184        my $ps = $ed->{path_strip} || "";
3185        for my $path ( grep { m/$ps/ } %{$ed->{dir_prop}} ) {
3186                my $props = $ed->{dir_prop}{$path};
3187                if ( $props->{"svk:merge"} ) {
3188                        $self->find_extra_svk_parents
3189                                ($ed, $props->{"svk:merge"}, \@parents);
3190                }
3191                if ( $props->{"svn:mergeinfo"} ) {
3192                        $self->find_extra_svn_parents
3193                                ($ed,
3194                                 $props->{"svn:mergeinfo"},
3195                                 \@parents);
3196                }
3197        }
3198
3199        open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
3200        print $un "r$rev\n" or croak $!;
3201        print $un $_, "\n" foreach @$untracked;
3202        my %log_entry = ( parents => \@parents, revision => $rev,
3203                          log => '');
3204
3205        my $headrev;
3206        my $logged = delete $self->{logged_rev_props};
3207        if (!$logged || $self->{-want_revprops}) {
3208                my $rp = $self->ra->rev_proplist($rev);
3209                foreach (sort keys %$rp) {
3210                        my $v = $rp->{$_};
3211                        if (/^svn:(author|date|log)$/) {
3212                                $log_entry{$1} = $v;
3213                        } elsif ($_ eq 'svm:headrev') {
3214                                $headrev = $v;
3215                        } else {
3216                                print $un "  rev_prop: ", uri_encode($_), ' ',
3217                                          uri_encode($v), "\n";
3218                        }
3219                }
3220        } else {
3221                map { $log_entry{$_} = $logged->{$_} } keys %$logged;
3222        }
3223        close $un or croak $!;
3224
3225        $log_entry{date} = parse_svn_date($log_entry{date});
3226        $log_entry{log} .= "\n";
3227        my $author = $log_entry{author} = check_author($log_entry{author});
3228        my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
3229                                                       : ($author, undef);
3230
3231        my ($commit_name, $commit_email) = ($name, $email);
3232        if ($_use_log_author) {
3233                my $name_field;
3234                if ($log_entry{log} =~ /From:\s+(.*\S)\s*\n/i) {
3235                        $name_field = $1;
3236                } elsif ($log_entry{log} =~ /Signed-off-by:\s+(.*\S)\s*\n/i) {
3237                        $name_field = $1;
3238                }
3239                if (!defined $name_field) {
3240                        if (!defined $email) {
3241                                $email = $name;
3242                        }
3243                } elsif ($name_field =~ /(.*?)\s+<(.*)>/) {
3244                        ($name, $email) = ($1, $2);
3245                } elsif ($name_field =~ /(.*)@/) {
3246                        ($name, $email) = ($1, $name_field);
3247                } else {
3248                        ($name, $email) = ($name_field, $name_field);
3249                }
3250        }
3251        if (defined $headrev && $self->use_svm_props) {
3252                if ($self->rewrite_root) {
3253                        die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
3254                            "options set!\n";
3255                }
3256                my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$}i;
3257                # we don't want "SVM: initializing mirror for junk" ...
3258                return undef if $r == 0;
3259                my $svm = $self->svm;
3260                if ($uuid ne $svm->{uuid}) {
3261                        die "UUID mismatch on SVM path:\n",
3262                            "expected: $svm->{uuid}\n",
3263                            "     got: $uuid\n";
3264                }
3265                my $full_url = $self->full_url;
3266                $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
3267                             die "Failed to replace '$svm->{replace}' with ",
3268                                 "'$svm->{source}' in $full_url\n";
3269                # throw away username for storing in records
3270                remove_username($full_url);
3271                $log_entry{metadata} = "$full_url\@$r $uuid";
3272                $log_entry{svm_revision} = $r;
3273                $email ||= "$author\@$uuid";
3274                $commit_email ||= "$author\@$uuid";
3275        } elsif ($self->use_svnsync_props) {
3276                my $full_url = $self->svnsync->{url};
3277                $full_url .= "/$self->{path}" if length $self->{path};
3278                remove_username($full_url);
3279                my $uuid = $self->svnsync->{uuid};
3280                $log_entry{metadata} = "$full_url\@$rev $uuid";
3281                $email ||= "$author\@$uuid";
3282                $commit_email ||= "$author\@$uuid";
3283        } else {
3284                my $url = $self->metadata_url;
3285                remove_username($url);
3286                $log_entry{metadata} = "$url\@$rev " .
3287                                       $self->ra->get_uuid;
3288                $email ||= "$author\@" . $self->ra->get_uuid;
3289                $commit_email ||= "$author\@" . $self->ra->get_uuid;
3290        }
3291        $log_entry{name} = $name;
3292        $log_entry{email} = $email;
3293        $log_entry{commit_name} = $commit_name;
3294        $log_entry{commit_email} = $commit_email;
3295        \%log_entry;
3296}
3297
3298sub fetch {
3299        my ($self, $min_rev, $max_rev, @parents) = @_;
3300        my ($last_rev, $last_commit) = $self->last_rev_commit;
3301        my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
3302        $self->ra->gs_fetch_loop_common($base, $head, [$self]);
3303}
3304
3305sub set_tree_cb {
3306        my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
3307        $self->{inject_parents} = { $rev => $tree };
3308        $self->fetch(undef, undef);
3309}
3310
3311sub set_tree {
3312        my ($self, $tree) = (shift, shift);
3313        my $log_entry = ::get_commit_entry($tree);
3314        unless ($self->{last_rev}) {
3315                ::fatal("Must have an existing revision to commit");
3316        }
3317        my %ed_opts = ( r => $self->{last_rev},
3318                        log => $log_entry->{log},
3319                        ra => $self->ra,
3320                        tree_a => $self->{last_commit},
3321                        tree_b => $tree,
3322                        editor_cb => sub {
3323                               $self->set_tree_cb($log_entry, $tree, @_) },
3324                        svn_path => $self->{path} );
3325        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
3326                print "No changes\nr$self->{last_rev} = $tree\n";
3327        }
3328}
3329
3330sub rebuild_from_rev_db {
3331        my ($self, $path) = @_;
3332        my $r = -1;
3333        open my $fh, '<', $path or croak "open: $!";
3334        binmode $fh or croak "binmode: $!";
3335        while (<$fh>) {
3336                length($_) == 41 or croak "inconsistent size in ($_) != 41";
3337                chomp($_);
3338                ++$r;
3339                next if $_ eq ('0' x 40);
3340                $self->rev_map_set($r, $_);
3341                print "r$r = $_\n";
3342        }
3343        close $fh or croak "close: $!";
3344        unlink $path or croak "unlink: $!";
3345}
3346
3347sub rebuild {
3348        my ($self) = @_;
3349        my $map_path = $self->map_path;
3350        my $partial = (-e $map_path && ! -z $map_path);
3351        return unless ::verify_ref($self->refname.'^0');
3352        if (!$partial && ($self->use_svm_props || $self->no_metadata)) {
3353                my $rev_db = $self->rev_db_path;
3354                $self->rebuild_from_rev_db($rev_db);
3355                if ($self->use_svm_props) {
3356                        my $svm_rev_db = $self->rev_db_path($self->svm_uuid);
3357                        $self->rebuild_from_rev_db($svm_rev_db);
3358                }
3359                $self->unlink_rev_db_symlink;
3360                return;
3361        }
3362        print "Rebuilding $map_path ...\n" if (!$partial);
3363        my ($base_rev, $head) = ($partial ? $self->rev_map_max_norebuild(1) :
3364                (undef, undef));
3365        my ($log, $ctx) =
3366            command_output_pipe(qw/rev-list --pretty=raw --no-color --reverse/,
3367                                ($head ? "$head.." : "") . $self->refname,
3368                                '--');
3369        my $metadata_url = $self->metadata_url;
3370        remove_username($metadata_url);
3371        my $svn_uuid = $self->ra_uuid;
3372        my $c;
3373        while (<$log>) {
3374                if ( m{^commit ($::sha1)$} ) {
3375                        $c = $1;
3376                        next;
3377                }
3378                next unless s{^\s*(git-svn-id:)}{$1};
3379                my ($url, $rev, $uuid) = ::extract_metadata($_);
3380                remove_username($url);
3381
3382                # ignore merges (from set-tree)
3383                next if (!defined $rev || !$uuid);
3384
3385                # if we merged or otherwise started elsewhere, this is
3386                # how we break out of it
3387                if (($uuid ne $svn_uuid) ||
3388                    ($metadata_url && $url && ($url ne $metadata_url))) {
3389                        next;
3390                }
3391                if ($partial && $head) {
3392                        print "Partial-rebuilding $map_path ...\n";
3393                        print "Currently at $base_rev = $head\n";
3394                        $head = undef;
3395                }
3396
3397                $self->rev_map_set($rev, $c);
3398                print "r$rev = $c\n";
3399        }
3400        command_close_pipe($log, $ctx);
3401        print "Done rebuilding $map_path\n" if (!$partial || !$head);
3402        my $rev_db_path = $self->rev_db_path;
3403        if (-f $self->rev_db_path) {
3404                unlink $self->rev_db_path or croak "unlink: $!";
3405        }
3406        $self->unlink_rev_db_symlink;
3407}
3408
3409# rev_map:
3410# Tie::File seems to be prone to offset errors if revisions get sparse,
3411# it's not that fast, either.  Tie::File is also not in Perl 5.6.  So
3412# one of my favorite modules is out :<  Next up would be one of the DBM
3413# modules, but I'm not sure which is most portable...
3414#
3415# This is the replacement for the rev_db format, which was too big
3416# and inefficient for large repositories with a lot of sparse history
3417# (mainly tags)
3418#
3419# The format is this:
3420#   - 24 bytes for every record,
3421#     * 4 bytes for the integer representing an SVN revision number
3422#     * 20 bytes representing the sha1 of a git commit
3423#   - No empty padding records like the old format
3424#     (except the last record, which can be overwritten)
3425#   - new records are written append-only since SVN revision numbers
3426#     increase monotonically
3427#   - lookups on SVN revision number are done via a binary search
3428#   - Piping the file to xxd -c24 is a good way of dumping it for
3429#     viewing or editing (piped back through xxd -r), should the need
3430#     ever arise.
3431#   - The last record can be padding revision with an all-zero sha1
3432#     This is used to optimize fetch performance when using multiple
3433#     "fetch" directives in .git/config
3434#
3435# These files are disposable unless noMetadata or useSvmProps is set
3436
3437sub _rev_map_set {
3438        my ($fh, $rev, $commit) = @_;
3439
3440        binmode $fh or croak "binmode: $!";
3441        my $size = (stat($fh))[7];
3442        ($size % 24) == 0 or croak "inconsistent size: $size";
3443
3444        my $wr_offset = 0;
3445        if ($size > 0) {
3446                sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3447                my $read = sysread($fh, my $buf, 24) or croak "read: $!";
3448                $read == 24 or croak "read only $read bytes (!= 24)";
3449                my ($last_rev, $last_commit) = unpack(rev_map_fmt, $buf);
3450                if ($last_commit eq ('0' x40)) {
3451                        if ($size >= 48) {
3452                                sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3453                                $read = sysread($fh, $buf, 24) or
3454                                    croak "read: $!";
3455                                $read == 24 or
3456                                    croak "read only $read bytes (!= 24)";
3457                                ($last_rev, $last_commit) =
3458                                    unpack(rev_map_fmt, $buf);
3459                                if ($last_commit eq ('0' x40)) {
3460                                        croak "inconsistent .rev_map\n";
3461                                }
3462                        }
3463                        if ($last_rev >= $rev) {
3464                                croak "last_rev is higher!: $last_rev >= $rev";
3465                        }
3466                        $wr_offset = -24;
3467                }
3468        }
3469        sysseek($fh, $wr_offset, SEEK_END) or croak "seek: $!";
3470        syswrite($fh, pack(rev_map_fmt, $rev, $commit), 24) == 24 or
3471          croak "write: $!";
3472}
3473
3474sub _rev_map_reset {
3475        my ($fh, $rev, $commit) = @_;
3476        my $c = _rev_map_get($fh, $rev);
3477        $c eq $commit or die "_rev_map_reset(@_) commit $c does not match!\n";
3478        my $offset = sysseek($fh, 0, SEEK_CUR) or croak "seek: $!";
3479        truncate $fh, $offset or croak "truncate: $!";
3480}
3481
3482sub mkfile {
3483        my ($path) = @_;
3484        unless (-e $path) {
3485                my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
3486                mkpath([$dir]) unless -d $dir;
3487                open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
3488                close $fh or die "Couldn't close (create) $path: $!\n";
3489        }
3490}
3491
3492sub rev_map_set {
3493        my ($self, $rev, $commit, $update_ref, $uuid) = @_;
3494        length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
3495        my $db = $self->map_path($uuid);
3496        my $db_lock = "$db.lock";
3497        my $sig;
3498        $update_ref ||= 0;
3499        if ($update_ref) {
3500                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3501                            $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
3502        }
3503        mkfile($db);
3504
3505        $LOCKFILES{$db_lock} = 1;
3506        my $sync;
3507        # both of these options make our .rev_db file very, very important
3508        # and we can't afford to lose it because rebuild() won't work
3509        if ($self->use_svm_props || $self->no_metadata) {
3510                $sync = 1;
3511                copy($db, $db_lock) or die "rev_map_set(@_): ",
3512                                           "Failed to copy: ",
3513                                           "$db => $db_lock ($!)\n";
3514        } else {
3515                rename $db, $db_lock or die "rev_map_set(@_): ",
3516                                            "Failed to rename: ",
3517                                            "$db => $db_lock ($!)\n";
3518        }
3519
3520        sysopen(my $fh, $db_lock, O_RDWR | O_CREAT)
3521             or croak "Couldn't open $db_lock: $!\n";
3522        $update_ref eq 'reset' ? _rev_map_reset($fh, $rev, $commit) :
3523                                 _rev_map_set($fh, $rev, $commit);
3524        if ($sync) {
3525                $fh->flush or die "Couldn't flush $db_lock: $!\n";
3526                $fh->sync or die "Couldn't sync $db_lock: $!\n";
3527        }
3528        close $fh or croak $!;
3529        if ($update_ref) {
3530                $_head = $self;
3531                my $note = "";
3532                $note = " ($update_ref)" if ($update_ref !~ /^\d*$/);
3533                command_noisy('update-ref', '-m', "r$rev$note",
3534                              $self->refname, $commit);
3535        }
3536        rename $db_lock, $db or die "rev_map_set(@_): ", "Failed to rename: ",
3537                                    "$db_lock => $db ($!)\n";
3538        delete $LOCKFILES{$db_lock};
3539        if ($update_ref) {
3540                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3541                            $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
3542                kill $sig, $$ if defined $sig;
3543        }
3544}
3545
3546# If want_commit, this will return an array of (rev, commit) where
3547# commit _must_ be a valid commit in the archive.
3548# Otherwise, it'll return the max revision (whether or not the
3549# commit is valid or just a 0x40 placeholder).
3550sub rev_map_max {
3551        my ($self, $want_commit) = @_;
3552        $self->rebuild;
3553        my ($r, $c) = $self->rev_map_max_norebuild($want_commit);
3554        $want_commit ? ($r, $c) : $r;
3555}
3556
3557sub rev_map_max_norebuild {
3558        my ($self, $want_commit) = @_;
3559        my $map_path = $self->map_path;
3560        stat $map_path or return $want_commit ? (0, undef) : 0;
3561        sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3562        binmode $fh or croak "binmode: $!";
3563        my $size = (stat($fh))[7];
3564        ($size % 24) == 0 or croak "inconsistent size: $size";
3565
3566        if ($size == 0) {
3567                close $fh or croak "close: $!";
3568                return $want_commit ? (0, undef) : 0;
3569        }
3570
3571        sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3572        sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3573        my ($r, $c) = unpack(rev_map_fmt, $buf);
3574        if ($want_commit && $c eq ('0' x40)) {
3575                if ($size < 48) {
3576                        return $want_commit ? (0, undef) : 0;
3577                }
3578                sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3579                sysread($fh, $buf, 24) == 24 or croak "read: $!";
3580                ($r, $c) = unpack(rev_map_fmt, $buf);
3581                if ($c eq ('0'x40)) {
3582                        croak "Penultimate record is all-zeroes in $map_path";
3583                }
3584        }
3585        close $fh or croak "close: $!";
3586        $want_commit ? ($r, $c) : $r;
3587}
3588
3589sub rev_map_get {
3590        my ($self, $rev, $uuid) = @_;
3591        my $map_path = $self->map_path($uuid);
3592        return undef unless -e $map_path;
3593
3594        sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3595        my $c = _rev_map_get($fh, $rev);
3596        close($fh) or croak "close: $!";
3597        $c
3598}
3599
3600sub _rev_map_get {
3601        my ($fh, $rev) = @_;
3602
3603        binmode $fh or croak "binmode: $!";
3604        my $size = (stat($fh))[7];
3605        ($size % 24) == 0 or croak "inconsistent size: $size";
3606
3607        if ($size == 0) {
3608                return undef;
3609        }
3610
3611        my ($l, $u) = (0, $size - 24);
3612        my ($r, $c, $buf);
3613
3614        while ($l <= $u) {
3615                my $i = int(($l/24 + $u/24) / 2) * 24;
3616                sysseek($fh, $i, SEEK_SET) or croak "seek: $!";
3617                sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3618                my ($r, $c) = unpack(rev_map_fmt, $buf);
3619
3620                if ($r < $rev) {
3621                        $l = $i + 24;
3622                } elsif ($r > $rev) {
3623                        $u = $i - 24;
3624                } else { # $r == $rev
3625                        return $c eq ('0' x 40) ? undef : $c;
3626                }
3627        }
3628        undef;
3629}
3630
3631# Finds the first svn revision that exists on (if $eq_ok is true) or
3632# before $rev for the current branch.  It will not search any lower
3633# than $min_rev.  Returns the git commit hash and svn revision number
3634# if found, else (undef, undef).
3635sub find_rev_before {
3636        my ($self, $rev, $eq_ok, $min_rev) = @_;
3637        --$rev unless $eq_ok;
3638        $min_rev ||= 1;
3639        my $max_rev = $self->rev_map_max;
3640        $rev = $max_rev if ($rev > $max_rev);
3641        while ($rev >= $min_rev) {
3642                if (my $c = $self->rev_map_get($rev)) {
3643                        return ($rev, $c);
3644                }
3645                --$rev;
3646        }
3647        return (undef, undef);
3648}
3649
3650# Finds the first svn revision that exists on (if $eq_ok is true) or
3651# after $rev for the current branch.  It will not search any higher
3652# than $max_rev.  Returns the git commit hash and svn revision number
3653# if found, else (undef, undef).
3654sub find_rev_after {
3655        my ($self, $rev, $eq_ok, $max_rev) = @_;
3656        ++$rev unless $eq_ok;
3657        $max_rev ||= $self->rev_map_max;
3658        while ($rev <= $max_rev) {
3659                if (my $c = $self->rev_map_get($rev)) {
3660                        return ($rev, $c);
3661                }
3662                ++$rev;
3663        }
3664        return (undef, undef);
3665}
3666
3667sub _new {
3668        my ($class, $repo_id, $ref_id, $path) = @_;
3669        unless (defined $repo_id && length $repo_id) {
3670                $repo_id = $Git::SVN::default_repo_id;
3671        }
3672        unless (defined $ref_id && length $ref_id) {
3673                $_prefix = '' unless defined($_prefix);
3674                $_[2] = $ref_id =
3675                             "refs/remotes/$_prefix$Git::SVN::default_ref_id";
3676        }
3677        $_[1] = $repo_id;
3678        my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
3679
3680        # Older repos imported by us used $GIT_DIR/svn/foo instead of
3681        # $GIT_DIR/svn/refs/remotes/foo when tracking refs/remotes/foo
3682        if ($ref_id =~ m{^refs/remotes/(.*)}) {
3683                my $old_dir = "$ENV{GIT_DIR}/svn/$1";
3684                if (-d $old_dir && ! -d $dir) {
3685                        $dir = $old_dir;
3686                }
3687        }
3688
3689        $_[3] = $path = '' unless (defined $path);
3690        mkpath([$dir]);
3691        bless {
3692                ref_id => $ref_id, dir => $dir, index => "$dir/index",
3693                path => $path, config => "$ENV{GIT_DIR}/svn/config",
3694                map_root => "$dir/.rev_map", repo_id => $repo_id }, $class;
3695}
3696
3697# for read-only access of old .rev_db formats
3698sub unlink_rev_db_symlink {
3699        my ($self) = @_;
3700        my $link = $self->rev_db_path;
3701        $link =~ s/\.[\w-]+$// or croak "missing UUID at the end of $link";
3702        if (-l $link) {
3703                unlink $link or croak "unlink: $link failed!";
3704        }
3705}
3706
3707sub rev_db_path {
3708        my ($self, $uuid) = @_;
3709        my $db_path = $self->map_path($uuid);
3710        $db_path =~ s{/\.rev_map\.}{/\.rev_db\.}
3711            or croak "map_path: $db_path does not contain '/.rev_map.' !";
3712        $db_path;
3713}
3714
3715# the new replacement for .rev_db
3716sub map_path {
3717        my ($self, $uuid) = @_;
3718        $uuid ||= $self->ra_uuid;
3719        "$self->{map_root}.$uuid";
3720}
3721
3722sub uri_encode {
3723        my ($f) = @_;
3724        $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
3725        $f
3726}
3727
3728sub uri_decode {
3729        my ($f) = @_;
3730        $f =~ s#%([0-9a-fA-F]{2})#chr(hex($1))#eg;
3731        $f
3732}
3733
3734sub remove_username {
3735        $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
3736}
3737
3738package Git::SVN::Prompt;
3739use strict;
3740use warnings;
3741require SVN::Core;
3742use vars qw/$_no_auth_cache $_username/;
3743
3744sub simple {
3745        my ($cred, $realm, $default_username, $may_save, $pool) = @_;
3746        $may_save = undef if $_no_auth_cache;
3747        $default_username = $_username if defined $_username;
3748        if (defined $default_username && length $default_username) {
3749                if (defined $realm && length $realm) {
3750                        print STDERR "Authentication realm: $realm\n";
3751                        STDERR->flush;
3752                }
3753                $cred->username($default_username);
3754        } else {
3755                username($cred, $realm, $may_save, $pool);
3756        }
3757        $cred->password(_read_password("Password for '" .
3758                                       $cred->username . "': ", $realm));
3759        $cred->may_save($may_save);
3760        $SVN::_Core::SVN_NO_ERROR;
3761}
3762
3763sub ssl_server_trust {
3764        my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
3765        $may_save = undef if $_no_auth_cache;
3766        print STDERR "Error validating server certificate for '$realm':\n";
3767        {
3768                no warnings 'once';
3769                # All variables SVN::Auth::SSL::* are used only once,
3770                # so we're shutting up Perl warnings about this.
3771                if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
3772                        print STDERR " - The certificate is not issued ",
3773                            "by a trusted authority. Use the\n",
3774                            "   fingerprint to validate ",
3775                            "the certificate manually!\n";
3776                }
3777                if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
3778                        print STDERR " - The certificate hostname ",
3779                            "does not match.\n";
3780                }
3781                if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
3782                        print STDERR " - The certificate is not yet valid.\n";
3783                }
3784                if ($failures & $SVN::Auth::SSL::EXPIRED) {
3785                        print STDERR " - The certificate has expired.\n";
3786                }
3787                if ($failures & $SVN::Auth::SSL::OTHER) {
3788                        print STDERR " - The certificate has ",
3789                            "an unknown error.\n";
3790                }
3791        } # no warnings 'once'
3792        printf STDERR
3793                "Certificate information:\n".
3794                " - Hostname: %s\n".
3795                " - Valid: from %s until %s\n".
3796                " - Issuer: %s\n".
3797                " - Fingerprint: %s\n",
3798                map $cert_info->$_, qw(hostname valid_from valid_until
3799                                       issuer_dname fingerprint);
3800        my $choice;
3801prompt:
3802        print STDERR $may_save ?
3803              "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
3804              "(R)eject or accept (t)emporarily? ";
3805        STDERR->flush;
3806        $choice = lc(substr(<STDIN> || 'R', 0, 1));
3807        if ($choice =~ /^t$/i) {
3808                $cred->may_save(undef);
3809        } elsif ($choice =~ /^r$/i) {
3810                return -1;
3811        } elsif ($may_save && $choice =~ /^p$/i) {
3812                $cred->may_save($may_save);
3813        } else {
3814                goto prompt;
3815        }
3816        $cred->accepted_failures($failures);
3817        $SVN::_Core::SVN_NO_ERROR;
3818}
3819
3820sub ssl_client_cert {
3821        my ($cred, $realm, $may_save, $pool) = @_;
3822        $may_save = undef if $_no_auth_cache;
3823        print STDERR "Client certificate filename: ";
3824        STDERR->flush;
3825        chomp(my $filename = <STDIN>);
3826        $cred->cert_file($filename);
3827        $cred->may_save($may_save);
3828        $SVN::_Core::SVN_NO_ERROR;
3829}
3830
3831sub ssl_client_cert_pw {
3832        my ($cred, $realm, $may_save, $pool) = @_;
3833        $may_save = undef if $_no_auth_cache;
3834        $cred->password(_read_password("Password: ", $realm));
3835        $cred->may_save($may_save);
3836        $SVN::_Core::SVN_NO_ERROR;
3837}
3838
3839sub username {
3840        my ($cred, $realm, $may_save, $pool) = @_;
3841        $may_save = undef if $_no_auth_cache;
3842        if (defined $realm && length $realm) {
3843                print STDERR "Authentication realm: $realm\n";
3844        }
3845        my $username;
3846        if (defined $_username) {
3847                $username = $_username;
3848        } else {
3849                print STDERR "Username: ";
3850                STDERR->flush;
3851                chomp($username = <STDIN>);
3852        }
3853        $cred->username($username);
3854        $cred->may_save($may_save);
3855        $SVN::_Core::SVN_NO_ERROR;
3856}
3857
3858sub _read_password {
3859        my ($prompt, $realm) = @_;
3860        print STDERR $prompt;
3861        STDERR->flush;
3862        require Term::ReadKey;
3863        Term::ReadKey::ReadMode('noecho');
3864        my $password = '';
3865        while (defined(my $key = Term::ReadKey::ReadKey(0))) {
3866                last if $key =~ /[\012\015]/; # \n\r
3867                $password .= $key;
3868        }
3869        Term::ReadKey::ReadMode('restore');
3870        print STDERR "\n";
3871        STDERR->flush;
3872        $password;
3873}
3874
3875package SVN::Git::Fetcher;
3876use vars qw/@ISA/;
3877use strict;
3878use warnings;
3879use Carp qw/croak/;
3880use File::Temp qw/tempfile/;
3881use IO::File qw//;
3882use vars qw/$_ignore_regex/;
3883
3884# file baton members: path, mode_a, mode_b, pool, fh, blob, base
3885sub new {
3886        my ($class, $git_svn, $switch_path) = @_;
3887        my $self = SVN::Delta::Editor->new;
3888        bless $self, $class;
3889        if (exists $git_svn->{last_commit}) {
3890                $self->{c} = $git_svn->{last_commit};
3891                $self->{empty_symlinks} =
3892                                  _mark_empty_symlinks($git_svn, $switch_path);
3893        }
3894        $self->{ignore_regex} = eval { command_oneline('config', '--get',
3895                             "svn-remote.$git_svn->{repo_id}.ignore-paths") };
3896        $self->{empty} = {};
3897        $self->{dir_prop} = {};
3898        $self->{file_prop} = {};
3899        $self->{absent_dir} = {};
3900        $self->{absent_file} = {};
3901        $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
3902        $self;
3903}
3904
3905# this uses the Ra object, so it must be called before do_{switch,update},
3906# not inside them (when the Git::SVN::Fetcher object is passed) to
3907# do_{switch,update}
3908sub _mark_empty_symlinks {
3909        my ($git_svn, $switch_path) = @_;
3910        my $bool = Git::config_bool('svn.brokenSymlinkWorkaround');
3911        return {} if (!defined($bool)) || (defined($bool) && ! $bool);
3912
3913        my %ret;
3914        my ($rev, $cmt) = $git_svn->last_rev_commit;
3915        return {} unless ($rev && $cmt);
3916
3917        # allow the warning to be printed for each revision we fetch to
3918        # ensure the user sees it.  The user can also disable the workaround
3919        # on the repository even while git svn is running and the next
3920        # revision fetched will skip this expensive function.
3921        my $printed_warning;
3922        chomp(my $empty_blob = `git hash-object -t blob --stdin < /dev/null`);
3923        my ($ls, $ctx) = command_output_pipe(qw/ls-tree -r -z/, $cmt);
3924        local $/ = "\0";
3925        my $pfx = defined($switch_path) ? $switch_path : $git_svn->{path};
3926        $pfx .= '/' if length($pfx);
3927        while (<$ls>) {
3928                chomp;
3929                s/\A100644 blob $empty_blob\t//o or next;
3930                unless ($printed_warning) {
3931                        print STDERR "Scanning for empty symlinks, ",
3932                                     "this may take a while if you have ",
3933                                     "many empty files\n",
3934                                     "You may disable this with `",
3935                                     "git config svn.brokenSymlinkWorkaround ",
3936                                     "false'.\n",
3937                                     "This may be done in a different ",
3938                                     "terminal without restarting ",
3939                                     "git svn\n";
3940                        $printed_warning = 1;
3941                }
3942                my $path = $_;
3943                my (undef, $props) =
3944                               $git_svn->ra->get_file($pfx.$path, $rev, undef);
3945                if ($props->{'svn:special'}) {
3946                        $ret{$path} = 1;
3947                }
3948        }
3949        command_close_pipe($ls, $ctx);
3950        \%ret;
3951}
3952
3953# returns true if a given path is inside a ".git" directory
3954sub in_dot_git {
3955        $_[0] =~ m{(?:^|/)\.git(?:/|$)};
3956}
3957
3958# return value: 0 -- don't ignore, 1 -- ignore
3959sub is_path_ignored {
3960        my ($self, $path) = @_;
3961        return 1 if in_dot_git($path);
3962        return 1 if defined($self->{ignore_regex}) &&
3963                    $path =~ m!$self->{ignore_regex}!;
3964        return 0 unless defined($_ignore_regex);
3965        return 1 if $path =~ m!$_ignore_regex!o;
3966        return 0;
3967}
3968
3969sub set_path_strip {
3970        my ($self, $path) = @_;
3971        $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
3972}
3973
3974sub open_root {
3975        { path => '' };
3976}
3977
3978sub open_directory {
3979        my ($self, $path, $pb, $rev) = @_;
3980        { path => $path };
3981}
3982
3983sub git_path {
3984        my ($self, $path) = @_;
3985        if ($self->{path_strip}) {
3986                $path =~ s!$self->{path_strip}!! or
3987                  die "Failed to strip path '$path' ($self->{path_strip})\n";
3988        }
3989        $path;
3990}
3991
3992sub delete_entry {
3993        my ($self, $path, $rev, $pb) = @_;
3994        return undef if $self->is_path_ignored($path);
3995
3996        my $gpath = $self->git_path($path);
3997        return undef if ($gpath eq '');
3998
3999        # remove entire directories.
4000        my ($tree) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
4001                         =~ /\A040000 tree ([a-f\d]{40})\t\Q$gpath\E\0/);
4002        if ($tree) {
4003                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
4004                                                     -r --name-only -z/,
4005                                                     $tree);
4006                local $/ = "\0";
4007                while (<$ls>) {
4008                        chomp;
4009                        my $rmpath = "$gpath/$_";
4010                        $self->{gii}->remove($rmpath);
4011                        print "\tD\t$rmpath\n" unless $::_q;
4012                }
4013                print "\tD\t$gpath/\n" unless $::_q;
4014                command_close_pipe($ls, $ctx);
4015        } else {
4016                $self->{gii}->remove($gpath);
4017                print "\tD\t$gpath\n" unless $::_q;
4018        }
4019        $self->{empty}->{$path} = 0;
4020        undef;
4021}
4022
4023sub open_file {
4024        my ($self, $path, $pb, $rev) = @_;
4025        my ($mode, $blob);
4026
4027        goto out if $self->is_path_ignored($path);
4028
4029        my $gpath = $self->git_path($path);
4030        ($mode, $blob) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
4031                             =~ /\A(\d{6}) blob ([a-f\d]{40})\t\Q$gpath\E\0/);
4032        unless (defined $mode && defined $blob) {
4033                die "$path was not found in commit $self->{c} (r$rev)\n";
4034        }
4035        if ($mode eq '100644' && $self->{empty_symlinks}->{$path}) {
4036                $mode = '120000';
4037        }
4038out:
4039        { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
4040          pool => SVN::Pool->new, action => 'M' };
4041}
4042
4043sub add_file {
4044        my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
4045        my $mode;
4046
4047        if (!$self->is_path_ignored($path)) {
4048                my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
4049                delete $self->{empty}->{$dir};
4050                $mode = '100644';
4051        }
4052        { path => $path, mode_a => $mode, mode_b => $mode,
4053          pool => SVN::Pool->new, action => 'A' };
4054}
4055
4056sub add_directory {
4057        my ($self, $path, $cp_path, $cp_rev) = @_;
4058        goto out if $self->is_path_ignored($path);
4059        my $gpath = $self->git_path($path);
4060        if ($gpath eq '') {
4061                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
4062                                                     -r --name-only -z/,
4063                                                     $self->{c});
4064                local $/ = "\0";
4065                while (<$ls>) {
4066                        chomp;
4067                        $self->{gii}->remove($_);
4068                        print "\tD\t$_\n" unless $::_q;
4069                }
4070                command_close_pipe($ls, $ctx);
4071                $self->{empty}->{$path} = 0;
4072        }
4073        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
4074        delete $self->{empty}->{$dir};
4075        $self->{empty}->{$path} = 1;
4076out:
4077        { path => $path };
4078}
4079
4080sub change_dir_prop {
4081        my ($self, $db, $prop, $value) = @_;
4082        return undef if $self->is_path_ignored($db->{path});
4083        $self->{dir_prop}->{$db->{path}} ||= {};
4084        $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
4085        undef;
4086}
4087
4088sub absent_directory {
4089        my ($self, $path, $pb) = @_;
4090        return undef if $self->is_path_ignored($path);
4091        $self->{absent_dir}->{$pb->{path}} ||= [];
4092        push @{$self->{absent_dir}->{$pb->{path}}}, $path;
4093        undef;
4094}
4095
4096sub absent_file {
4097        my ($self, $path, $pb) = @_;
4098        return undef if $self->is_path_ignored($path);
4099        $self->{absent_file}->{$pb->{path}} ||= [];
4100        push @{$self->{absent_file}->{$pb->{path}}}, $path;
4101        undef;
4102}
4103
4104sub change_file_prop {
4105        my ($self, $fb, $prop, $value) = @_;
4106        return undef if $self->is_path_ignored($fb->{path});
4107        if ($prop eq 'svn:executable') {
4108                if ($fb->{mode_b} != 120000) {
4109                        $fb->{mode_b} = defined $value ? 100755 : 100644;
4110                }
4111        } elsif ($prop eq 'svn:special') {
4112                $fb->{mode_b} = defined $value ? 120000 : 100644;
4113        } else {
4114                $self->{file_prop}->{$fb->{path}} ||= {};
4115                $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
4116        }
4117        undef;
4118}
4119
4120sub apply_textdelta {
4121        my ($self, $fb, $exp) = @_;
4122        return undef if $self->is_path_ignored($fb->{path});
4123        my $fh = $::_repository->temp_acquire('svn_delta');
4124        # $fh gets auto-closed() by SVN::TxDelta::apply(),
4125        # (but $base does not,) so dup() it for reading in close_file
4126        open my $dup, '<&', $fh or croak $!;
4127        my $base = $::_repository->temp_acquire('git_blob');
4128
4129        if ($fb->{blob}) {
4130                my ($base_is_link, $size);
4131
4132                if ($fb->{mode_a} eq '120000' &&
4133                    ! $self->{empty_symlinks}->{$fb->{path}}) {
4134                        print $base 'link ' or die "print $!\n";
4135                        $base_is_link = 1;
4136                }
4137        retry:
4138                $size = $::_repository->cat_blob($fb->{blob}, $base);
4139                die "Failed to read object $fb->{blob}" if ($size < 0);
4140
4141                if (defined $exp) {
4142                        seek $base, 0, 0 or croak $!;
4143                        my $got = ::md5sum($base);
4144                        if ($got ne $exp) {
4145                                my $err = "Checksum mismatch: ".
4146                                       "$fb->{path} $fb->{blob}\n" .
4147                                       "expected: $exp\n" .
4148                                       "     got: $got\n";
4149                                if ($base_is_link) {
4150                                        warn $err,
4151                                             "Retrying... (possibly ",
4152                                             "a bad symlink from SVN)\n";
4153                                        $::_repository->temp_reset($base);
4154                                        $base_is_link = 0;
4155                                        goto retry;
4156                                }
4157                                die $err;
4158                        }
4159                }
4160        }
4161        seek $base, 0, 0 or croak $!;
4162        $fb->{fh} = $fh;
4163        $fb->{base} = $base;
4164        [ SVN::TxDelta::apply($base, $dup, undef, $fb->{path}, $fb->{pool}) ];
4165}
4166
4167sub close_file {
4168        my ($self, $fb, $exp) = @_;
4169        return undef if $self->is_path_ignored($fb->{path});
4170
4171        my $hash;
4172        my $path = $self->git_path($fb->{path});
4173        if (my $fh = $fb->{fh}) {
4174                if (defined $exp) {
4175                        seek($fh, 0, 0) or croak $!;
4176                        my $got = ::md5sum($fh);
4177                        if ($got ne $exp) {
4178                                die "Checksum mismatch: $path\n",
4179                                    "expected: $exp\n    got: $got\n";
4180                        }
4181                }
4182                if ($fb->{mode_b} == 120000) {
4183                        sysseek($fh, 0, 0) or croak $!;
4184                        my $rd = sysread($fh, my $buf, 5);
4185
4186                        if (!defined $rd) {
4187                                croak "sysread: $!\n";
4188                        } elsif ($rd == 0) {
4189                                warn "$path has mode 120000",
4190                                     " but it points to nothing\n",
4191                                     "converting to an empty file with mode",
4192                                     " 100644\n";
4193                                $fb->{mode_b} = '100644';
4194                        } elsif ($buf ne 'link ') {
4195                                warn "$path has mode 120000",
4196                                     " but is not a link\n";
4197                        } else {
4198                                my $tmp_fh = $::_repository->temp_acquire(
4199                                        'svn_hash');
4200                                my $res;
4201                                while ($res = sysread($fh, my $str, 1024)) {
4202                                        my $out = syswrite($tmp_fh, $str, $res);
4203                                        defined($out) && $out == $res
4204                                                or croak("write ",
4205                                                        Git::temp_path($tmp_fh),
4206                                                        ": $!\n");
4207                                }
4208                                defined $res or croak $!;
4209
4210                                ($fh, $tmp_fh) = ($tmp_fh, $fh);
4211                                Git::temp_release($tmp_fh, 1);
4212                        }
4213                }
4214
4215                $hash = $::_repository->hash_and_insert_object(
4216                                Git::temp_path($fh));
4217                $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
4218
4219                Git::temp_release($fb->{base}, 1);
4220                Git::temp_release($fh, 1);
4221        } else {
4222                $hash = $fb->{blob} or die "no blob information\n";
4223        }
4224        $fb->{pool}->clear;
4225        $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
4226        print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
4227        undef;
4228}
4229
4230sub abort_edit {
4231        my $self = shift;
4232        $self->{nr} = $self->{gii}->{nr};
4233        delete $self->{gii};
4234        $self->SUPER::abort_edit(@_);
4235}
4236
4237sub close_edit {
4238        my $self = shift;
4239        $self->{git_commit_ok} = 1;
4240        $self->{nr} = $self->{gii}->{nr};
4241        delete $self->{gii};
4242        $self->SUPER::close_edit(@_);
4243}
4244
4245package SVN::Git::Editor;
4246use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
4247use strict;
4248use warnings;
4249use Carp qw/croak/;
4250use IO::File;
4251
4252sub new {
4253        my ($class, $opts) = @_;
4254        foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
4255                die "$_ required!\n" unless (defined $opts->{$_});
4256        }
4257
4258        my $pool = SVN::Pool->new;
4259        my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
4260        my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
4261                                     $opts->{r}, $mods);
4262
4263        # $opts->{ra} functions should not be used after this:
4264        my @ce  = $opts->{ra}->get_commit_editor($opts->{log},
4265                                                $opts->{editor_cb}, $pool);
4266        my $self = SVN::Delta::Editor->new(@ce, $pool);
4267        bless $self, $class;
4268        foreach (qw/svn_path r tree_a tree_b/) {
4269                $self->{$_} = $opts->{$_};
4270        }
4271        $self->{url} = $opts->{ra}->{url};
4272        $self->{mods} = $mods;
4273        $self->{types} = $types;
4274        $self->{pool} = $pool;
4275        $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
4276        $self->{rm} = { };
4277        $self->{path_prefix} = length $self->{svn_path} ?
4278                               "$self->{svn_path}/" : '';
4279        $self->{config} = $opts->{config};
4280        return $self;
4281}
4282
4283sub generate_diff {
4284        my ($tree_a, $tree_b) = @_;
4285        my @diff_tree = qw(diff-tree -z -r);
4286        if ($_cp_similarity) {
4287                push @diff_tree, "-C$_cp_similarity";
4288        } else {
4289                push @diff_tree, '-C';
4290        }
4291        push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
4292        push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
4293        push @diff_tree, $tree_a, $tree_b;
4294        my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
4295        local $/ = "\0";
4296        my $state = 'meta';
4297        my @mods;
4298        while (<$diff_fh>) {
4299                chomp $_; # this gets rid of the trailing "\0"
4300                if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
4301                                        ($::sha1)\s($::sha1)\s
4302                                        ([MTCRAD])\d*$/xo) {
4303                        push @mods, {   mode_a => $1, mode_b => $2,
4304                                        sha1_a => $3, sha1_b => $4,
4305                                        chg => $5 };
4306                        if ($5 =~ /^(?:C|R)$/) {
4307                                $state = 'file_a';
4308                        } else {
4309                                $state = 'file_b';
4310                        }
4311                } elsif ($state eq 'file_a') {
4312                        my $x = $mods[$#mods] or croak "Empty array\n";
4313                        if ($x->{chg} !~ /^(?:C|R)$/) {
4314                                croak "Error parsing $_, $x->{chg}\n";
4315                        }
4316                        $x->{file_a} = $_;
4317                        $state = 'file_b';
4318                } elsif ($state eq 'file_b') {
4319                        my $x = $mods[$#mods] or croak "Empty array\n";
4320                        if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
4321                                croak "Error parsing $_, $x->{chg}\n";
4322                        }
4323                        if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
4324                                croak "Error parsing $_, $x->{chg}\n";
4325                        }
4326                        $x->{file_b} = $_;
4327                        $state = 'meta';
4328                } else {
4329                        croak "Error parsing $_\n";
4330                }
4331        }
4332        command_close_pipe($diff_fh, $ctx);
4333        \@mods;
4334}
4335
4336sub check_diff_paths {
4337        my ($ra, $pfx, $rev, $mods) = @_;
4338        my %types;
4339        $pfx .= '/' if length $pfx;
4340
4341        sub type_diff_paths {
4342                my ($ra, $types, $path, $rev) = @_;
4343                my @p = split m#/+#, $path;
4344                my $c = shift @p;
4345                unless (defined $types->{$c}) {
4346                        $types->{$c} = $ra->check_path($c, $rev);
4347                }
4348                while (@p) {
4349                        $c .= '/' . shift @p;
4350                        next if defined $types->{$c};
4351                        $types->{$c} = $ra->check_path($c, $rev);
4352                }
4353        }
4354
4355        foreach my $m (@$mods) {
4356                foreach my $f (qw/file_a file_b/) {
4357                        next unless defined $m->{$f};
4358                        my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
4359                        if (length $pfx.$dir && ! defined $types{$dir}) {
4360                                type_diff_paths($ra, \%types, $pfx.$dir, $rev);
4361                        }
4362                }
4363        }
4364        \%types;
4365}
4366
4367sub split_path {
4368        return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
4369}
4370
4371sub repo_path {
4372        my ($self, $path) = @_;
4373        $self->{path_prefix}.(defined $path ? $path : '');
4374}
4375
4376sub url_path {
4377        my ($self, $path) = @_;
4378        if ($self->{url} =~ m#^https?://#) {
4379                $path =~ s!([^~a-zA-Z0-9_./-])!uc sprintf("%%%02x",ord($1))!eg;
4380        }
4381        $self->{url} . '/' . $self->repo_path($path);
4382}
4383
4384sub rmdirs {
4385        my ($self) = @_;
4386        my $rm = $self->{rm};
4387        delete $rm->{''}; # we never delete the url we're tracking
4388        return unless %$rm;
4389
4390        foreach (keys %$rm) {
4391                my @d = split m#/#, $_;
4392                my $c = shift @d;
4393                $rm->{$c} = 1;
4394                while (@d) {
4395                        $c .= '/' . shift @d;
4396                        $rm->{$c} = 1;
4397                }
4398        }
4399        delete $rm->{$self->{svn_path}};
4400        delete $rm->{''}; # we never delete the url we're tracking
4401        return unless %$rm;
4402
4403        my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
4404                                             $self->{tree_b});
4405        local $/ = "\0";
4406        while (<$fh>) {
4407                chomp;
4408                my @dn = split m#/#, $_;
4409                while (pop @dn) {
4410                        delete $rm->{join '/', @dn};
4411                }
4412                unless (%$rm) {
4413                        close $fh;
4414                        return;
4415                }
4416        }
4417        command_close_pipe($fh, $ctx);
4418
4419        my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
4420        foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
4421                $self->close_directory($bat->{$d}, $p);
4422                my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
4423                print "\tD+\t$d/\n" unless $::_q;
4424                $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
4425                delete $bat->{$d};
4426        }
4427}
4428
4429sub open_or_add_dir {
4430        my ($self, $full_path, $baton) = @_;
4431        my $t = $self->{types}->{$full_path};
4432        if (!defined $t) {
4433                die "$full_path not known in r$self->{r} or we have a bug!\n";
4434        }
4435        {
4436                no warnings 'once';
4437                # SVN::Node::none and SVN::Node::file are used only once,
4438                # so we're shutting up Perl's warnings about them.
4439                if ($t == $SVN::Node::none) {
4440                        return $self->add_directory($full_path, $baton,
4441                            undef, -1, $self->{pool});
4442                } elsif ($t == $SVN::Node::dir) {
4443                        return $self->open_directory($full_path, $baton,
4444                            $self->{r}, $self->{pool});
4445                } # no warnings 'once'
4446                print STDERR "$full_path already exists in repository at ",
4447                    "r$self->{r} and it is not a directory (",
4448                    ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
4449        } # no warnings 'once'
4450        exit 1;
4451}
4452
4453sub ensure_path {
4454        my ($self, $path) = @_;
4455        my $bat = $self->{bat};
4456        my $repo_path = $self->repo_path($path);
4457        return $bat->{''} unless (length $repo_path);
4458        my @p = split m#/+#, $repo_path;
4459        my $c = shift @p;
4460        $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
4461        while (@p) {
4462                my $c0 = $c;
4463                $c .= '/' . shift @p;
4464                $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
4465        }
4466        return $bat->{$c};
4467}
4468
4469# Subroutine to convert a globbing pattern to a regular expression.
4470# From perl cookbook.
4471sub glob2pat {
4472        my $globstr = shift;
4473        my %patmap = ('*' => '.*', '?' => '.', '[' => '[', ']' => ']');
4474        $globstr =~ s{(.)} { $patmap{$1} || "\Q$1" }ge;
4475        return '^' . $globstr . '$';
4476}
4477
4478sub check_autoprop {
4479        my ($self, $pattern, $properties, $file, $fbat) = @_;
4480        # Convert the globbing pattern to a regular expression.
4481        my $regex = glob2pat($pattern);
4482        # Check if the pattern matches the file name.
4483        if($file =~ m/($regex)/) {
4484                # Parse the list of properties to set.
4485                my @props = split(/;/, $properties);
4486                foreach my $prop (@props) {
4487                        # Parse 'name=value' syntax and set the property.
4488                        if ($prop =~ /([^=]+)=(.*)/) {
4489                                my ($n,$v) = ($1,$2);
4490                                for ($n, $v) {
4491                                        s/^\s+//; s/\s+$//;
4492                                }
4493                                $self->change_file_prop($fbat, $n, $v);
4494                        }
4495                }
4496        }
4497}
4498
4499sub apply_autoprops {
4500        my ($self, $file, $fbat) = @_;
4501        my $conf_t = ${$self->{config}}{'config'};
4502        no warnings 'once';
4503        # Check [miscellany]/enable-auto-props in svn configuration.
4504        if (SVN::_Core::svn_config_get_bool(
4505                $conf_t,
4506                $SVN::_Core::SVN_CONFIG_SECTION_MISCELLANY,
4507                $SVN::_Core::SVN_CONFIG_OPTION_ENABLE_AUTO_PROPS,
4508                0)) {
4509                # Auto-props are enabled.  Enumerate them to look for matches.
4510                my $callback = sub {
4511                        $self->check_autoprop($_[0], $_[1], $file, $fbat);
4512                };
4513                SVN::_Core::svn_config_enumerate(
4514                        $conf_t,
4515                        $SVN::_Core::SVN_CONFIG_SECTION_AUTO_PROPS,
4516                        $callback);
4517        }
4518}
4519
4520sub A {
4521        my ($self, $m) = @_;
4522        my ($dir, $file) = split_path($m->{file_b});
4523        my $pbat = $self->ensure_path($dir);
4524        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4525                                        undef, -1);
4526        print "\tA\t$m->{file_b}\n" unless $::_q;
4527        $self->apply_autoprops($file, $fbat);
4528        $self->chg_file($fbat, $m);
4529        $self->close_file($fbat,undef,$self->{pool});
4530}
4531
4532sub C {
4533        my ($self, $m) = @_;
4534        my ($dir, $file) = split_path($m->{file_b});
4535        my $pbat = $self->ensure_path($dir);
4536        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4537                                $self->url_path($m->{file_a}), $self->{r});
4538        print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
4539        $self->chg_file($fbat, $m);
4540        $self->close_file($fbat,undef,$self->{pool});
4541}
4542
4543sub delete_entry {
4544        my ($self, $path, $pbat) = @_;
4545        my $rpath = $self->repo_path($path);
4546        my ($dir, $file) = split_path($rpath);
4547        $self->{rm}->{$dir} = 1;
4548        $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
4549}
4550
4551sub R {
4552        my ($self, $m) = @_;
4553        my ($dir, $file) = split_path($m->{file_b});
4554        my $pbat = $self->ensure_path($dir);
4555        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4556                                $self->url_path($m->{file_a}), $self->{r});
4557        print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
4558        $self->apply_autoprops($file, $fbat);
4559        $self->chg_file($fbat, $m);
4560        $self->close_file($fbat,undef,$self->{pool});
4561
4562        ($dir, $file) = split_path($m->{file_a});
4563        $pbat = $self->ensure_path($dir);
4564        $self->delete_entry($m->{file_a}, $pbat);
4565}
4566
4567sub M {
4568        my ($self, $m) = @_;
4569        my ($dir, $file) = split_path($m->{file_b});
4570        my $pbat = $self->ensure_path($dir);
4571        my $fbat = $self->open_file($self->repo_path($m->{file_b}),
4572                                $pbat,$self->{r},$self->{pool});
4573        print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
4574        $self->chg_file($fbat, $m);
4575        $self->close_file($fbat,undef,$self->{pool});
4576}
4577
4578sub T { shift->M(@_) }
4579
4580sub change_file_prop {
4581        my ($self, $fbat, $pname, $pval) = @_;
4582        $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
4583}
4584
4585sub _chg_file_get_blob ($$$$) {
4586        my ($self, $fbat, $m, $which) = @_;
4587        my $fh = $::_repository->temp_acquire("git_blob_$which");
4588        if ($m->{"mode_$which"} =~ /^120/) {
4589                print $fh 'link ' or croak $!;
4590                $self->change_file_prop($fbat,'svn:special','*');
4591        } elsif ($m->{mode_a} =~ /^120/ && $m->{"mode_$which"} !~ /^120/) {
4592                $self->change_file_prop($fbat,'svn:special',undef);
4593        }
4594        my $blob = $m->{"sha1_$which"};
4595        return ($fh,) if ($blob =~ /^0{40}$/);
4596        my $size = $::_repository->cat_blob($blob, $fh);
4597        croak "Failed to read object $blob" if ($size < 0);
4598        $fh->flush == 0 or croak $!;
4599        seek $fh, 0, 0 or croak $!;
4600
4601        my $exp = ::md5sum($fh);
4602        seek $fh, 0, 0 or croak $!;
4603        return ($fh, $exp);
4604}
4605
4606sub chg_file {
4607        my ($self, $fbat, $m) = @_;
4608        if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
4609                $self->change_file_prop($fbat,'svn:executable','*');
4610        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
4611                $self->change_file_prop($fbat,'svn:executable',undef);
4612        }
4613        my ($fh_a, $exp_a) = _chg_file_get_blob $self, $fbat, $m, 'a';
4614        my ($fh_b, $exp_b) = _chg_file_get_blob $self, $fbat, $m, 'b';
4615        my $pool = SVN::Pool->new;
4616        my $atd = $self->apply_textdelta($fbat, $exp_a, $pool);
4617        if (-s $fh_a) {
4618                my $txstream = SVN::TxDelta::new ($fh_a, $fh_b, $pool);
4619                my $res = SVN::TxDelta::send_txstream($txstream, @$atd, $pool);
4620                if (defined $res) {
4621                        die "Unexpected result from send_txstream: $res\n",
4622                            "(SVN::Core::VERSION: $SVN::Core::VERSION)\n";
4623                }
4624        } else {
4625                my $got = SVN::TxDelta::send_stream($fh_b, @$atd, $pool);
4626                die "Checksum mismatch\nexpected: $exp_b\ngot: $got\n"
4627                    if ($got ne $exp_b);
4628        }
4629        Git::temp_release($fh_b, 1);
4630        Git::temp_release($fh_a, 1);
4631        $pool->clear;
4632}
4633
4634sub D {
4635        my ($self, $m) = @_;
4636        my ($dir, $file) = split_path($m->{file_b});
4637        my $pbat = $self->ensure_path($dir);
4638        print "\tD\t$m->{file_b}\n" unless $::_q;
4639        $self->delete_entry($m->{file_b}, $pbat);
4640}
4641
4642sub close_edit {
4643        my ($self) = @_;
4644        my ($p,$bat) = ($self->{pool}, $self->{bat});
4645        foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
4646                next if $_ eq '';
4647                $self->close_directory($bat->{$_}, $p);
4648        }
4649        $self->close_directory($bat->{''}, $p);
4650        $self->SUPER::close_edit($p);
4651        $p->clear;
4652}
4653
4654sub abort_edit {
4655        my ($self) = @_;
4656        $self->SUPER::abort_edit($self->{pool});
4657}
4658
4659sub DESTROY {
4660        my $self = shift;
4661        $self->SUPER::DESTROY(@_);
4662        $self->{pool}->clear;
4663}
4664
4665# this drives the editor
4666sub apply_diff {
4667        my ($self) = @_;
4668        my $mods = $self->{mods};
4669        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
4670        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
4671                my $f = $m->{chg};
4672                if (defined $o{$f}) {
4673                        $self->$f($m);
4674                } else {
4675                        fatal("Invalid change type: $f");
4676                }
4677        }
4678        $self->rmdirs if $_rmdir;
4679        if (@$mods == 0) {
4680                $self->abort_edit;
4681        } else {
4682                $self->close_edit;
4683        }
4684        return scalar @$mods;
4685}
4686
4687package Git::SVN::Ra;
4688use vars qw/@ISA $config_dir $_log_window_size/;
4689use strict;
4690use warnings;
4691my ($ra_invalid, $can_do_switch, %ignored_err, $RA);
4692
4693BEGIN {
4694        # enforce temporary pool usage for some simple functions
4695        no strict 'refs';
4696        for my $f (qw/rev_proplist get_latest_revnum get_uuid get_repos_root
4697                      get_file/) {
4698                my $SUPER = "SUPER::$f";
4699                *$f = sub {
4700                        my $self = shift;
4701                        my $pool = SVN::Pool->new;
4702                        my @ret = $self->$SUPER(@_,$pool);
4703                        $pool->clear;
4704                        wantarray ? @ret : $ret[0];
4705                };
4706        }
4707}
4708
4709sub _auth_providers () {
4710        [
4711          SVN::Client::get_simple_provider(),
4712          SVN::Client::get_ssl_server_trust_file_provider(),
4713          SVN::Client::get_simple_prompt_provider(
4714            \&Git::SVN::Prompt::simple, 2),
4715          SVN::Client::get_ssl_client_cert_file_provider(),
4716          SVN::Client::get_ssl_client_cert_prompt_provider(
4717            \&Git::SVN::Prompt::ssl_client_cert, 2),
4718          SVN::Client::get_ssl_client_cert_pw_file_provider(),
4719          SVN::Client::get_ssl_client_cert_pw_prompt_provider(
4720            \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
4721          SVN::Client::get_username_provider(),
4722          SVN::Client::get_ssl_server_trust_prompt_provider(
4723            \&Git::SVN::Prompt::ssl_server_trust),
4724          SVN::Client::get_username_prompt_provider(
4725            \&Git::SVN::Prompt::username, 2)
4726        ]
4727}
4728
4729sub escape_uri_only {
4730        my ($uri) = @_;
4731        my @tmp;
4732        foreach (split m{/}, $uri) {
4733                s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
4734                push @tmp, $_;
4735        }
4736        join('/', @tmp);
4737}
4738
4739sub escape_url {
4740        my ($url) = @_;
4741        if ($url =~ m#^(https?)://([^/]+)(.*)$#) {
4742                my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
4743                $url = "$scheme://$domain$uri";
4744        }
4745        $url;
4746}
4747
4748sub new {
4749        my ($class, $url) = @_;
4750        $url =~ s!/+$!!;
4751        return $RA if ($RA && $RA->{url} eq $url);
4752
4753        SVN::_Core::svn_config_ensure($config_dir, undef);
4754        my ($baton, $callbacks) = SVN::Core::auth_open_helper(_auth_providers);
4755        my $config = SVN::Core::config_get_config($config_dir);
4756        $RA = undef;
4757        my $dont_store_passwords = 1;
4758        my $conf_t = ${$config}{'config'};
4759        {
4760                no warnings 'once';
4761                # The usage of $SVN::_Core::SVN_CONFIG_* variables
4762                # produces warnings that variables are used only once.
4763                # I had not found the better way to shut them up, so
4764                # the warnings of type 'once' are disabled in this block.
4765                if (SVN::_Core::svn_config_get_bool($conf_t,
4766                    $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4767                    $SVN::_Core::SVN_CONFIG_OPTION_STORE_PASSWORDS,
4768                    1) == 0) {
4769                        SVN::_Core::svn_auth_set_parameter($baton,
4770                            $SVN::_Core::SVN_AUTH_PARAM_DONT_STORE_PASSWORDS,
4771                            bless (\$dont_store_passwords, "_p_void"));
4772                }
4773                if (SVN::_Core::svn_config_get_bool($conf_t,
4774                    $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4775                    $SVN::_Core::SVN_CONFIG_OPTION_STORE_AUTH_CREDS,
4776                    1) == 0) {
4777                        $Git::SVN::Prompt::_no_auth_cache = 1;
4778                }
4779        } # no warnings 'once'
4780        my $self = SVN::Ra->new(url => escape_url($url), auth => $baton,
4781                              config => $config,
4782                              pool => SVN::Pool->new,
4783                              auth_provider_callbacks => $callbacks);
4784        $self->{url} = $url;
4785        $self->{svn_path} = $url;
4786        $self->{repos_root} = $self->get_repos_root;
4787        $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
4788        $self->{cache} = { check_path => { r => 0, data => {} },
4789                           get_dir => { r => 0, data => {} } };
4790        $RA = bless $self, $class;
4791}
4792
4793sub check_path {
4794        my ($self, $path, $r) = @_;
4795        my $cache = $self->{cache}->{check_path};
4796        if ($r == $cache->{r} && exists $cache->{data}->{$path}) {
4797                return $cache->{data}->{$path};
4798        }
4799        my $pool = SVN::Pool->new;
4800        my $t = $self->SUPER::check_path($path, $r, $pool);
4801        $pool->clear;
4802        if ($r != $cache->{r}) {
4803                %{$cache->{data}} = ();
4804                $cache->{r} = $r;
4805        }
4806        $cache->{data}->{$path} = $t;
4807}
4808
4809sub get_dir {
4810        my ($self, $dir, $r) = @_;
4811        my $cache = $self->{cache}->{get_dir};
4812        if ($r == $cache->{r}) {
4813                if (my $x = $cache->{data}->{$dir}) {
4814                        return wantarray ? @$x : $x->[0];
4815                }
4816        }
4817        my $pool = SVN::Pool->new;
4818        my ($d, undef, $props) = $self->SUPER::get_dir($dir, $r, $pool);
4819        my %dirents = map { $_ => { kind => $d->{$_}->kind } } keys %$d;
4820        $pool->clear;
4821        if ($r != $cache->{r}) {
4822                %{$cache->{data}} = ();
4823                $cache->{r} = $r;
4824        }
4825        $cache->{data}->{$dir} = [ \%dirents, $r, $props ];
4826        wantarray ? (\%dirents, $r, $props) : \%dirents;
4827}
4828
4829sub DESTROY {
4830        # do not call the real DESTROY since we store ourselves in $RA
4831}
4832
4833# get_log(paths, start, end, limit,
4834#         discover_changed_paths, strict_node_history, receiver)
4835sub get_log {
4836        my ($self, @args) = @_;
4837        my $pool = SVN::Pool->new;
4838
4839        # svn_log_changed_path_t objects passed to get_log are likely to be
4840        # overwritten even if only the refs are copied to an external variable,
4841        # so we should dup the structures in their entirety.  Using an
4842        # externally passed pool (instead of our temporary and quickly cleared
4843        # pool in Git::SVN::Ra) does not help matters at all...
4844        my $receiver = pop @args;
4845        my $prefix = "/".$self->{svn_path};
4846        $prefix =~ s#/+($)##;
4847        my $prefix_regex = qr#^\Q$prefix\E#;
4848        push(@args, sub {
4849                my ($paths) = $_[0];
4850                return &$receiver(@_) unless $paths;
4851                $_[0] = ();
4852                foreach my $p (keys %$paths) {
4853                        my $i = $paths->{$p};
4854                        # Make path relative to our url, not repos_root
4855                        $p =~ s/$prefix_regex//;
4856                        my %s = map { $_ => $i->$_; }
4857                                qw/copyfrom_path copyfrom_rev action/;
4858                        if ($s{'copyfrom_path'}) {
4859                                $s{'copyfrom_path'} =~ s/$prefix_regex//;
4860                        }
4861                        $_[0]{$p} = \%s;
4862                }
4863                &$receiver(@_);
4864        });
4865
4866
4867        # the limit parameter was not supported in SVN 1.1.x, so we
4868        # drop it.  Therefore, the receiver callback passed to it
4869        # is made aware of this limitation by being wrapped if
4870        # the limit passed to is being wrapped.
4871        if ($SVN::Core::VERSION le '1.2.0') {
4872                my $limit = splice(@args, 3, 1);
4873                if ($limit > 0) {
4874                        my $receiver = pop @args;
4875                        push(@args, sub { &$receiver(@_) if (--$limit >= 0) });
4876                }
4877        }
4878        my $ret = $self->SUPER::get_log(@args, $pool);
4879        $pool->clear;
4880        $ret;
4881}
4882
4883sub trees_match {
4884        my ($self, $url1, $rev1, $url2, $rev2) = @_;
4885        my $ctx = SVN::Client->new(auth => _auth_providers);
4886        my $out = IO::File->new_tmpfile;
4887
4888        # older SVN (1.1.x) doesn't take $pool as the last parameter for
4889        # $ctx->diff(), so we'll create a default one
4890        my $pool = SVN::Pool->new_default_sub;
4891
4892        $ra_invalid = 1; # this will open a new SVN::Ra connection to $url1
4893        $ctx->diff([], $url1, $rev1, $url2, $rev2, 1, 1, 0, $out, $out);
4894        $out->flush;
4895        my $ret = (($out->stat)[7] == 0);
4896        close $out or croak $!;
4897
4898        $ret;
4899}
4900
4901sub get_commit_editor {
4902        my ($self, $log, $cb, $pool) = @_;
4903        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
4904        $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
4905}
4906
4907sub gs_do_update {
4908        my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
4909        my $new = ($rev_a == $rev_b);
4910        my $path = $gs->{path};
4911
4912        if ($new && -e $gs->{index}) {
4913                unlink $gs->{index} or die
4914                  "Couldn't unlink index: $gs->{index}: $!\n";
4915        }
4916        my $pool = SVN::Pool->new;
4917        $editor->set_path_strip($path);
4918        my (@pc) = split m#/#, $path;
4919        my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
4920                                        1, $editor, $pool);
4921        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
4922
4923        # Since we can't rely on svn_ra_reparent being available, we'll
4924        # just have to do some magic with set_path to make it so
4925        # we only want a partial path.
4926        my $sp = '';
4927        my $final = join('/', @pc);
4928        while (@pc) {
4929                $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
4930                $sp .= '/' if length $sp;
4931                $sp .= shift @pc;
4932        }
4933        die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
4934
4935        $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
4936
4937        $reporter->finish_report($pool);
4938        $pool->clear;
4939        $editor->{git_commit_ok};
4940}
4941
4942# this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
4943# svn_ra_reparent didn't work before 1.4)
4944sub gs_do_switch {
4945        my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
4946        my $path = $gs->{path};
4947        my $pool = SVN::Pool->new;
4948
4949        my $full_url = $self->{url};
4950        my $old_url = $full_url;
4951        $full_url .= '/' . $path if length $path;
4952        my ($ra, $reparented);
4953
4954        if ($old_url =~ m#^svn(\+ssh)?://# ||
4955            ($full_url =~ m#^https?://# &&
4956             escape_url($full_url) ne $full_url)) {
4957                $_[0] = undef;
4958                $self = undef;
4959                $RA = undef;
4960                $ra = Git::SVN::Ra->new($full_url);
4961                $ra_invalid = 1;
4962        } elsif ($old_url ne $full_url) {
4963                SVN::_Ra::svn_ra_reparent($self->{session}, $full_url, $pool);
4964                $self->{url} = $full_url;
4965                $reparented = 1;
4966        }
4967
4968        $ra ||= $self;
4969        $url_b = escape_url($url_b);
4970        my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
4971        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
4972        $reporter->set_path('', $rev_a, 0, @lock, $pool);
4973        $reporter->finish_report($pool);
4974
4975        if ($reparented) {
4976                SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
4977                $self->{url} = $old_url;
4978        }
4979
4980        $pool->clear;
4981        $editor->{git_commit_ok};
4982}
4983
4984sub longest_common_path {
4985        my ($gsv, $globs) = @_;
4986        my %common;
4987        my $common_max = scalar @$gsv;
4988
4989        foreach my $gs (@$gsv) {
4990                my @tmp = split m#/#, $gs->{path};
4991                my $p = '';
4992                foreach (@tmp) {
4993                        $p .= length($p) ? "/$_" : $_;
4994                        $common{$p} ||= 0;
4995                        $common{$p}++;
4996                }
4997        }
4998        $globs ||= [];
4999        $common_max += scalar @$globs;
5000        foreach my $glob (@$globs) {
5001                my @tmp = split m#/#, $glob->{path}->{left};
5002                my $p = '';
5003                foreach (@tmp) {
5004                        $p .= length($p) ? "/$_" : $_;
5005                        $common{$p} ||= 0;
5006                        $common{$p}++;
5007                }
5008        }
5009
5010        my $longest_path = '';
5011        foreach (sort {length $b <=> length $a} keys %common) {
5012                if ($common{$_} == $common_max) {
5013                        $longest_path = $_;
5014                        last;
5015                }
5016        }
5017        $longest_path;
5018}
5019
5020sub gs_fetch_loop_common {
5021        my ($self, $base, $head, $gsv, $globs) = @_;
5022        return if ($base > $head);
5023        my $inc = $_log_window_size;
5024        my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
5025        my $longest_path = longest_common_path($gsv, $globs);
5026        my $ra_url = $self->{url};
5027        my $find_trailing_edge;
5028        while (1) {
5029                my %revs;
5030                my $err;
5031                my $err_handler = $SVN::Error::handler;
5032                $SVN::Error::handler = sub {
5033                        ($err) = @_;
5034                        skip_unknown_revs($err);
5035                };
5036                sub _cb {
5037                        my ($paths, $r, $author, $date, $log) = @_;
5038                        [ $paths,
5039                          { author => $author, date => $date, log => $log } ];
5040                }
5041                $self->get_log([$longest_path], $min, $max, 0, 1, 1,
5042                               sub { $revs{$_[1]} = _cb(@_) });
5043                if ($err) {
5044                        print "Checked through r$max\r";
5045                } else {
5046                        $find_trailing_edge = 1;
5047                }
5048                if ($err and $find_trailing_edge) {
5049                        print STDERR "Path '$longest_path' ",
5050                                     "was probably deleted:\n",
5051                                     $err->expanded_message,
5052                                     "\nWill attempt to follow ",
5053                                     "revisions r$min .. r$max ",
5054                                     "committed before the deletion\n";
5055                        my $hi = $max;
5056                        while (--$hi >= $min) {
5057                                my $ok;
5058                                $self->get_log([$longest_path], $min, $hi,
5059                                               0, 1, 1, sub {
5060                                               $ok = $_[1];
5061                                               $revs{$_[1]} = _cb(@_) });
5062                                if ($ok) {
5063                                        print STDERR "r$min .. r$ok OK\n";
5064                                        last;
5065                                }
5066                        }
5067                        $find_trailing_edge = 0;
5068                }
5069                $SVN::Error::handler = $err_handler;
5070
5071                my %exists = map { $_->{path} => $_ } @$gsv;
5072                foreach my $r (sort {$a <=> $b} keys %revs) {
5073                        my ($paths, $logged) = @{$revs{$r}};
5074
5075                        foreach my $gs ($self->match_globs(\%exists, $paths,
5076                                                           $globs, $r)) {
5077                                if ($gs->rev_map_max >= $r) {
5078                                        next;
5079                                }
5080                                next unless $gs->match_paths($paths, $r);
5081                                $gs->{logged_rev_props} = $logged;
5082                                if (my $last_commit = $gs->last_commit) {
5083                                        $gs->assert_index_clean($last_commit);
5084                                }
5085                                my $log_entry = $gs->do_fetch($paths, $r);
5086                                if ($log_entry) {
5087                                        $gs->do_git_commit($log_entry);
5088                                }
5089                                $INDEX_FILES{$gs->{index}} = 1;
5090                        }
5091                        foreach my $g (@$globs) {
5092                                my $k = "svn-remote.$g->{remote}." .
5093                                        "$g->{t}-maxRev";
5094                                Git::SVN::tmp_config($k, $r);
5095                        }
5096                        if ($ra_invalid) {
5097                                $_[0] = undef;
5098                                $self = undef;
5099                                $RA = undef;
5100                                $self = Git::SVN::Ra->new($ra_url);
5101                                $ra_invalid = undef;
5102                        }
5103                }
5104                # pre-fill the .rev_db since it'll eventually get filled in
5105                # with '0' x40 if something new gets committed
5106                foreach my $gs (@$gsv) {
5107                        next if $gs->rev_map_max >= $max;
5108                        next if defined $gs->rev_map_get($max);
5109                        $gs->rev_map_set($max, 0 x40);
5110                }
5111                foreach my $g (@$globs) {
5112                        my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
5113                        Git::SVN::tmp_config($k, $max);
5114                }
5115                last if $max >= $head;
5116                $min = $max + 1;
5117                $max += $inc;
5118                $max = $head if ($max > $head);
5119        }
5120        Git::SVN::gc();
5121}
5122
5123sub get_dir_globbed {
5124        my ($self, $left, $depth, $r) = @_;
5125
5126        my @x = eval { $self->get_dir($left, $r) };
5127        return unless scalar @x == 3;
5128        my $dirents = $x[0];
5129        my @finalents;
5130        foreach my $de (keys %$dirents) {
5131                next if $dirents->{$de}->{kind} != $SVN::Node::dir;
5132                if ($depth > 1) {
5133                        my @args = ("$left/$de", $depth - 1, $r);
5134                        foreach my $dir ($self->get_dir_globbed(@args)) {
5135                                push @finalents, "$de/$dir";
5136                        }
5137                } else {
5138                        push @finalents, $de;
5139                }
5140        }
5141        @finalents;
5142}
5143
5144sub match_globs {
5145        my ($self, $exists, $paths, $globs, $r) = @_;
5146
5147        sub get_dir_check {
5148                my ($self, $exists, $g, $r) = @_;
5149
5150                my @dirs = $self->get_dir_globbed($g->{path}->{left},
5151                                                  $g->{path}->{depth},
5152                                                  $r);
5153
5154                foreach my $de (@dirs) {
5155                        my $p = $g->{path}->full_path($de);
5156                        next if $exists->{$p};
5157                        next if (length $g->{path}->{right} &&
5158                                 ($self->check_path($p, $r) !=
5159                                  $SVN::Node::dir));
5160                        $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
5161                                         $g->{ref}->full_path($de), 1);
5162                }
5163        }
5164        foreach my $g (@$globs) {
5165                if (my $path = $paths->{"/$g->{path}->{left}"}) {
5166                        if ($path->{action} =~ /^[AR]$/) {
5167                                get_dir_check($self, $exists, $g, $r);
5168                        }
5169                }
5170                foreach (keys %$paths) {
5171                        if (/$g->{path}->{left_regex}/ &&
5172                            !/$g->{path}->{regex}/) {
5173                                next if $paths->{$_}->{action} !~ /^[AR]$/;
5174                                get_dir_check($self, $exists, $g, $r);
5175                        }
5176                        next unless /$g->{path}->{regex}/;
5177                        my $p = $1;
5178                        my $pathname = $g->{path}->full_path($p);
5179                        next if $exists->{$pathname};
5180                        next if ($self->check_path($pathname, $r) !=
5181                                 $SVN::Node::dir);
5182                        $exists->{$pathname} = Git::SVN->init(
5183                                              $self->{url}, $pathname, undef,
5184                                              $g->{ref}->full_path($p), 1);
5185                }
5186                my $c = '';
5187                foreach (split m#/#, $g->{path}->{left}) {
5188                        $c .= "/$_";
5189                        next unless ($paths->{$c} &&
5190                                     ($paths->{$c}->{action} =~ /^[AR]$/));
5191                        get_dir_check($self, $exists, $g, $r);
5192                }
5193        }
5194        values %$exists;
5195}
5196
5197sub minimize_url {
5198        my ($self) = @_;
5199        return $self->{url} if ($self->{url} eq $self->{repos_root});
5200        my $url = $self->{repos_root};
5201        my @components = split(m!/!, $self->{svn_path});
5202        my $c = '';
5203        do {
5204                $url .= "/$c" if length $c;
5205                eval {
5206                        my $ra = (ref $self)->new($url);
5207                        my $latest = $ra->get_latest_revnum;
5208                        $ra->get_log("", $latest, 0, 1, 0, 1, sub {});
5209                };
5210        } while ($@ && ($c = shift @components));
5211        $url;
5212}
5213
5214sub can_do_switch {
5215        my $self = shift;
5216        unless (defined $can_do_switch) {
5217                my $pool = SVN::Pool->new;
5218                my $rep = eval {
5219                        $self->do_switch(1, '', 0, $self->{url},
5220                                         SVN::Delta::Editor->new, $pool);
5221                };
5222                if ($@) {
5223                        $can_do_switch = 0;
5224                } else {
5225                        $rep->abort_report($pool);
5226                        $can_do_switch = 1;
5227                }
5228                $pool->clear;
5229        }
5230        $can_do_switch;
5231}
5232
5233sub skip_unknown_revs {
5234        my ($err) = @_;
5235        my $errno = $err->apr_err();
5236        # Maybe the branch we're tracking didn't
5237        # exist when the repo started, so it's
5238        # not an error if it doesn't, just continue
5239        #
5240        # Wonderfully consistent library, eh?
5241        # 160013 - svn:// and file://
5242        # 175002 - http(s)://
5243        # 175007 - http(s):// (this repo required authorization, too...)
5244        #   More codes may be discovered later...
5245        if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
5246                my $err_key = $err->expanded_message;
5247                # revision numbers change every time, filter them out
5248                $err_key =~ s/\d+/\0/g;
5249                $err_key = "$errno\0$err_key";
5250                unless ($ignored_err{$err_key}) {
5251                        warn "W: Ignoring error from SVN, path probably ",
5252                             "does not exist: ($errno): ",
5253                             $err->expanded_message,"\n";
5254                        warn "W: Do not be alarmed at the above message ",
5255                             "git-svn is just searching aggressively for ",
5256                             "old history.\n",
5257                             "This may take a while on large repositories\n";
5258                        $ignored_err{$err_key} = 1;
5259                }
5260                return;
5261        }
5262        die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
5263}
5264
5265package Git::SVN::Log;
5266use strict;
5267use warnings;
5268use POSIX qw/strftime/;
5269use Time::Local;
5270use constant commit_log_separator => ('-' x 72) . "\n";
5271use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
5272            %rusers $show_commit $incremental/;
5273my $l_fmt;
5274
5275sub cmt_showable {
5276        my ($c) = @_;
5277        return 1 if defined $c->{r};
5278
5279        # big commit message got truncated by the 16k pretty buffer in rev-list
5280        if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
5281                                $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
5282                @{$c->{l}} = ();
5283                my @log = command(qw/cat-file commit/, $c->{c});
5284
5285                # shift off the headers
5286                shift @log while ($log[0] ne '');
5287                shift @log;
5288
5289                # TODO: make $c->{l} not have a trailing newline in the future
5290                @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
5291
5292                (undef, $c->{r}, undef) = ::extract_metadata(
5293                                (grep(/^git-svn-id: /, @log))[-1]);
5294        }
5295        return defined $c->{r};
5296}
5297
5298sub log_use_color {
5299        return $color || Git->repository->get_colorbool('color.diff');
5300}
5301
5302sub git_svn_log_cmd {
5303        my ($r_min, $r_max, @args) = @_;
5304        my $head = 'HEAD';
5305        my (@files, @log_opts);
5306        foreach my $x (@args) {
5307                if ($x eq '--' || @files) {
5308                        push @files, $x;
5309                } else {
5310                        if (::verify_ref("$x^0")) {
5311                                $head = $x;
5312                        } else {
5313                                push @log_opts, $x;
5314                        }
5315                }
5316        }
5317
5318        my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
5319        $gs ||= Git::SVN->_new;
5320        my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
5321                   $gs->refname);
5322        push @cmd, '-r' unless $non_recursive;
5323        push @cmd, qw/--raw --name-status/ if $verbose;
5324        push @cmd, '--color' if log_use_color();
5325        push @cmd, @log_opts;
5326        if (defined $r_max && $r_max == $r_min) {
5327                push @cmd, '--max-count=1';
5328                if (my $c = $gs->rev_map_get($r_max)) {
5329                        push @cmd, $c;
5330                }
5331        } elsif (defined $r_max) {
5332                if ($r_max < $r_min) {
5333                        ($r_min, $r_max) = ($r_max, $r_min);
5334                }
5335                my (undef, $c_max) = $gs->find_rev_before($r_max, 1, $r_min);
5336                my (undef, $c_min) = $gs->find_rev_after($r_min, 1, $r_max);
5337                # If there are no commits in the range, both $c_max and $c_min
5338                # will be undefined.  If there is at least 1 commit in the
5339                # range, both will be defined.
5340                return () if !defined $c_min || !defined $c_max;
5341                if ($c_min eq $c_max) {
5342                        push @cmd, '--max-count=1', $c_min;
5343                } else {
5344                        push @cmd, '--boundary', "$c_min..$c_max";
5345                }
5346        }
5347        return (@cmd, @files);
5348}
5349
5350# adapted from pager.c
5351sub config_pager {
5352        chomp(my $pager = command_oneline(qw(var GIT_PAGER)));
5353        if ($pager eq 'cat') {
5354                $pager = undef;
5355        }
5356        $ENV{GIT_PAGER_IN_USE} = defined($pager);
5357}
5358
5359sub run_pager {
5360        return unless -t *STDOUT && defined $pager;
5361        pipe my ($rfd, $wfd) or return;
5362        defined(my $pid = fork) or ::fatal "Can't fork: $!";
5363        if (!$pid) {
5364                open STDOUT, '>&', $wfd or
5365                                     ::fatal "Can't redirect to stdout: $!";
5366                return;
5367        }
5368        open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!";
5369        $ENV{LESS} ||= 'FRSX';
5370        exec $pager or ::fatal "Can't run pager: $! ($pager)";
5371}
5372
5373sub format_svn_date {
5374        # some systmes don't handle or mishandle %z, so be creative.
5375        my $t = shift || time;
5376        my $gm = timelocal(gmtime($t));
5377        my $sign = qw( + + - )[ $t <=> $gm ];
5378        my $gmoff = sprintf("%s%02d%02d", $sign, (gmtime(abs($t - $gm)))[2,1]);
5379        return strftime("%Y-%m-%d %H:%M:%S $gmoff (%a, %d %b %Y)", localtime($t));
5380}
5381
5382sub parse_git_date {
5383        my ($t, $tz) = @_;
5384        # Date::Parse isn't in the standard Perl distro :(
5385        if ($tz =~ s/^\+//) {
5386                $t += tz_to_s_offset($tz);
5387        } elsif ($tz =~ s/^\-//) {
5388                $t -= tz_to_s_offset($tz);
5389        }
5390        return $t;
5391}
5392
5393sub set_local_timezone {
5394        if (defined $TZ) {
5395                $ENV{TZ} = $TZ;
5396        } else {
5397                delete $ENV{TZ};
5398        }
5399}
5400
5401sub tz_to_s_offset {
5402        my ($tz) = @_;
5403        $tz =~ s/(\d\d)$//;
5404        return ($1 * 60) + ($tz * 3600);
5405}
5406
5407sub get_author_info {
5408        my ($dest, $author, $t, $tz) = @_;
5409        $author =~ s/(?:^\s*|\s*$)//g;
5410        $dest->{a_raw} = $author;
5411        my $au;
5412        if ($::_authors) {
5413                $au = $rusers{$author} || undef;
5414        }
5415        if (!$au) {
5416                ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
5417        }
5418        $dest->{t} = $t;
5419        $dest->{tz} = $tz;
5420        $dest->{a} = $au;
5421        $dest->{t_utc} = parse_git_date($t, $tz);
5422}
5423
5424sub process_commit {
5425        my ($c, $r_min, $r_max, $defer) = @_;
5426        if (defined $r_min && defined $r_max) {
5427                if ($r_min == $c->{r} && $r_min == $r_max) {
5428                        show_commit($c);
5429                        return 0;
5430                }
5431                return 1 if $r_min == $r_max;
5432                if ($r_min < $r_max) {
5433                        # we need to reverse the print order
5434                        return 0 if (defined $limit && --$limit < 0);
5435                        push @$defer, $c;
5436                        return 1;
5437                }
5438                if ($r_min != $r_max) {
5439                        return 1 if ($r_min < $c->{r});
5440                        return 1 if ($r_max > $c->{r});
5441                }
5442        }
5443        return 0 if (defined $limit && --$limit < 0);
5444        show_commit($c);
5445        return 1;
5446}
5447
5448sub show_commit {
5449        my $c = shift;
5450        if ($oneline) {
5451                my $x = "\n";
5452                if (my $l = $c->{l}) {
5453                        while ($l->[0] =~ /^\s*$/) { shift @$l }
5454                        $x = $l->[0];
5455                }
5456                $l_fmt ||= 'A' . length($c->{r});
5457                print 'r',pack($l_fmt, $c->{r}),' | ';
5458                print "$c->{c} | " if $show_commit;
5459                print $x;
5460        } else {
5461                show_commit_normal($c);
5462        }
5463}
5464
5465sub show_commit_changed_paths {
5466        my ($c) = @_;
5467        return unless $c->{changed};
5468        print "Changed paths:\n", @{$c->{changed}};
5469}
5470
5471sub show_commit_normal {
5472        my ($c) = @_;
5473        print commit_log_separator, "r$c->{r} | ";
5474        print "$c->{c} | " if $show_commit;
5475        print "$c->{a} | ", format_svn_date($c->{t_utc}), ' | ';
5476        my $nr_line = 0;
5477
5478        if (my $l = $c->{l}) {
5479                while ($l->[$#$l] eq "\n" && $#$l > 0
5480                                          && $l->[($#$l - 1)] eq "\n") {
5481                        pop @$l;
5482                }
5483                $nr_line = scalar @$l;
5484                if (!$nr_line) {
5485                        print "1 line\n\n\n";
5486                } else {
5487                        if ($nr_line == 1) {
5488                                $nr_line = '1 line';
5489                        } else {
5490                                $nr_line .= ' lines';
5491                        }
5492                        print $nr_line, "\n";
5493                        show_commit_changed_paths($c);
5494                        print "\n";
5495                        print $_ foreach @$l;
5496                }
5497        } else {
5498                print "1 line\n";
5499                show_commit_changed_paths($c);
5500                print "\n";
5501
5502        }
5503        foreach my $x (qw/raw stat diff/) {
5504                if ($c->{$x}) {
5505                        print "\n";
5506                        print $_ foreach @{$c->{$x}}
5507                }
5508        }
5509}
5510
5511sub cmd_show_log {
5512        my (@args) = @_;
5513        my ($r_min, $r_max);
5514        my $r_last = -1; # prevent dupes
5515        set_local_timezone();
5516        if (defined $::_revision) {
5517                if ($::_revision =~ /^(\d+):(\d+)$/) {
5518                        ($r_min, $r_max) = ($1, $2);
5519                } elsif ($::_revision =~ /^\d+$/) {
5520                        $r_min = $r_max = $::_revision;
5521                } else {
5522                        ::fatal "-r$::_revision is not supported, use ",
5523                                "standard 'git log' arguments instead";
5524                }
5525        }
5526
5527        config_pager();
5528        @args = git_svn_log_cmd($r_min, $r_max, @args);
5529        if (!@args) {
5530                print commit_log_separator unless $incremental || $oneline;
5531                return;
5532        }
5533        my $log = command_output_pipe(@args);
5534        run_pager();
5535        my (@k, $c, $d, $stat);
5536        my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
5537        while (<$log>) {
5538                if (/^${esc_color}commit -?($::sha1_short)/o) {
5539                        my $cmt = $1;
5540                        if ($c && cmt_showable($c) && $c->{r} != $r_last) {
5541                                $r_last = $c->{r};
5542                                process_commit($c, $r_min, $r_max, \@k) or
5543                                                                goto out;
5544                        }
5545                        $d = undef;
5546                        $c = { c => $cmt };
5547                } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
5548                        get_author_info($c, $1, $2, $3);
5549                } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
5550                        # ignore
5551                } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
5552                        push @{$c->{raw}}, $_;
5553                } elsif (/^${esc_color}[ACRMDT]\t/) {
5554                        # we could add $SVN->{svn_path} here, but that requires
5555                        # remote access at the moment (repo_path_split)...
5556                        s#^(${esc_color})([ACRMDT])\t#$1   $2 #o;
5557                        push @{$c->{changed}}, $_;
5558                } elsif (/^${esc_color}diff /o) {
5559                        $d = 1;
5560                        push @{$c->{diff}}, $_;
5561                } elsif ($d) {
5562                        push @{$c->{diff}}, $_;
5563                } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
5564                          $esc_color*[\+\-]*$esc_color$/x) {
5565                        $stat = 1;
5566                        push @{$c->{stat}}, $_;
5567                } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
5568                        push @{$c->{stat}}, $_;
5569                        $stat = undef;
5570                } elsif (/^${esc_color}    (git-svn-id:.+)$/o) {
5571                        ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
5572                } elsif (s/^${esc_color}    //o) {
5573                        push @{$c->{l}}, $_;
5574                }
5575        }
5576        if ($c && defined $c->{r} && $c->{r} != $r_last) {
5577                $r_last = $c->{r};
5578                process_commit($c, $r_min, $r_max, \@k);
5579        }
5580        if (@k) {
5581                ($r_min, $r_max) = ($r_max, $r_min);
5582                process_commit($_, $r_min, $r_max) foreach reverse @k;
5583        }
5584out:
5585        close $log;
5586        print commit_log_separator unless $incremental || $oneline;
5587}
5588
5589sub cmd_blame {
5590        my $path = pop;
5591
5592        config_pager();
5593        run_pager();
5594
5595        my ($fh, $ctx, $rev);
5596
5597        if ($_git_format) {
5598                ($fh, $ctx) = command_output_pipe('blame', @_, $path);
5599                while (my $line = <$fh>) {
5600                        if ($line =~ /^\^?([[:xdigit:]]+)\s/) {
5601                                # Uncommitted edits show up as a rev ID of
5602                                # all zeros, which we can't look up with
5603                                # cmt_metadata
5604                                if ($1 !~ /^0+$/) {
5605                                        (undef, $rev, undef) =
5606                                                ::cmt_metadata($1);
5607                                        $rev = '0' if (!$rev);
5608                                } else {
5609                                        $rev = '0';
5610                                }
5611                                $rev = sprintf('%-10s', $rev);
5612                                $line =~ s/^\^?[[:xdigit:]]+(\s)/$rev$1/;
5613                        }
5614                        print $line;
5615                }
5616        } else {
5617                ($fh, $ctx) = command_output_pipe('blame', '-p', @_, 'HEAD',
5618                                                  '--', $path);
5619                my ($sha1);
5620                my %authors;
5621                my @buffer;
5622                my %dsha; #distinct sha keys
5623
5624                while (my $line = <$fh>) {
5625                        push @buffer, $line;
5626                        if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5627                                $dsha{$1} = 1;
5628                        }
5629                }
5630
5631                my $s2r = ::cmt_sha2rev_batch([keys %dsha]);
5632
5633                foreach my $line (@buffer) {
5634                        if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5635                                $rev = $s2r->{$1};
5636                                $rev = '0' if (!$rev)
5637                        }
5638                        elsif ($line =~ /^author (.*)/) {
5639                                $authors{$rev} = $1;
5640                                $authors{$rev} =~ s/\s/_/g;
5641                        }
5642                        elsif ($line =~ /^\t(.*)$/) {
5643                                printf("%6s %10s %s\n", $rev, $authors{$rev}, $1);
5644                        }
5645                }
5646        }
5647        command_close_pipe($fh, $ctx);
5648}
5649
5650package Git::SVN::Migration;
5651# these version numbers do NOT correspond to actual version numbers
5652# of git nor git-svn.  They are just relative.
5653#
5654# v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
5655#
5656# v1 layout: .git/$id/info/url, refs/remotes/$id
5657#
5658# v2 layout: .git/svn/$id/info/url, refs/remotes/$id
5659#
5660# v3 layout: .git/svn/$id, refs/remotes/$id
5661#            - info/url may remain for backwards compatibility
5662#            - this is what we migrate up to this layout automatically,
5663#            - this will be used by git svn init on single branches
5664# v3.1 layout (auto migrated):
5665#            - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
5666#              for backwards compatibility
5667#
5668# v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
5669#            - this is only created for newly multi-init-ed
5670#              repositories.  Similar in spirit to the
5671#              --use-separate-remotes option in git-clone (now default)
5672#            - we do not automatically migrate to this (following
5673#              the example set by core git)
5674#
5675# v5 layout: .rev_db.$UUID => .rev_map.$UUID
5676#            - newer, more-efficient format that uses 24-bytes per record
5677#              with no filler space.
5678#            - use xxd -c24 < .rev_map.$UUID to view and debug
5679#            - This is a one-way migration, repositories updated to the
5680#              new format will not be able to use old git-svn without
5681#              rebuilding the .rev_db.  Rebuilding the rev_db is not
5682#              possible if noMetadata or useSvmProps are set; but should
5683#              be no problem for users that use the (sensible) defaults.
5684use strict;
5685use warnings;
5686use Carp qw/croak/;
5687use File::Path qw/mkpath/;
5688use File::Basename qw/dirname basename/;
5689use vars qw/$_minimize/;
5690
5691sub migrate_from_v0 {
5692        my $git_dir = $ENV{GIT_DIR};
5693        return undef unless -d $git_dir;
5694        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5695        my $migrated = 0;
5696        while (<$fh>) {
5697                chomp;
5698                my ($id, $orig_ref) = ($_, $_);
5699                next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
5700                next unless -f "$git_dir/$id/info/url";
5701                my $new_ref = "refs/remotes/$id";
5702                if (::verify_ref("$new_ref^0")) {
5703                        print STDERR "W: $orig_ref is probably an old ",
5704                                     "branch used by an ancient version of ",
5705                                     "git-svn.\n",
5706                                     "However, $new_ref also exists.\n",
5707                                     "We will not be able ",
5708                                     "to use this branch until this ",
5709                                     "ambiguity is resolved.\n";
5710                        next;
5711                }
5712                print STDERR "Migrating from v0 layout...\n" if !$migrated;
5713                print STDERR "Renaming ref: $orig_ref => $new_ref\n";
5714                command_noisy('update-ref', $new_ref, $orig_ref);
5715                command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
5716                $migrated++;
5717        }
5718        command_close_pipe($fh, $ctx);
5719        print STDERR "Done migrating from v0 layout...\n" if $migrated;
5720        $migrated;
5721}
5722
5723sub migrate_from_v1 {
5724        my $git_dir = $ENV{GIT_DIR};
5725        my $migrated = 0;
5726        return $migrated unless -d $git_dir;
5727        my $svn_dir = "$git_dir/svn";
5728
5729        # just in case somebody used 'svn' as their $id at some point...
5730        return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
5731
5732        print STDERR "Migrating from a git-svn v1 layout...\n";
5733        mkpath([$svn_dir]);
5734        print STDERR "Data from a previous version of git-svn exists, but\n\t",
5735                     "$svn_dir\n\t(required for this version ",
5736                     "($::VERSION) of git-svn) does not exist.\n";
5737        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5738        while (<$fh>) {
5739                my $x = $_;
5740                next unless $x =~ s#^refs/remotes/##;
5741                chomp $x;
5742                next unless -f "$git_dir/$x/info/url";
5743                my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
5744                next unless $u;
5745                my $dn = dirname("$git_dir/svn/$x");
5746                mkpath([$dn]) unless -d $dn;
5747                if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
5748                        mkpath(["$git_dir/svn/svn"]);
5749                        print STDERR " - $git_dir/$x/info => ",
5750                                        "$git_dir/svn/$x/info\n";
5751                        rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
5752                               croak "$!: $x";
5753                        # don't worry too much about these, they probably
5754                        # don't exist with repos this old (save for index,
5755                        # and we can easily regenerate that)
5756                        foreach my $f (qw/unhandled.log index .rev_db/) {
5757                                rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
5758                        }
5759                } else {
5760                        print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
5761                        rename "$git_dir/$x", "$git_dir/svn/$x" or
5762                               croak "$!: $x";
5763                }
5764                $migrated++;
5765        }
5766        command_close_pipe($fh, $ctx);
5767        print STDERR "Done migrating from a git-svn v1 layout\n";
5768        $migrated;
5769}
5770
5771sub read_old_urls {
5772        my ($l_map, $pfx, $path) = @_;
5773        my @dir;
5774        foreach (<$path/*>) {
5775                if (-r "$_/info/url") {
5776                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
5777                        my $ref_id = $pfx . basename $_;
5778                        my $url = ::file_to_s("$_/info/url");
5779                        $l_map->{$ref_id} = $url;
5780                } elsif (-d $_) {
5781                        push @dir, $_;
5782                }
5783        }
5784        foreach (@dir) {
5785                my $x = $_;
5786                $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
5787                read_old_urls($l_map, $x, $_);
5788        }
5789}
5790
5791sub migrate_from_v2 {
5792        my @cfg = command(qw/config -l/);
5793        return if grep /^svn-remote\..+\.url=/, @cfg;
5794        my %l_map;
5795        read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
5796        my $migrated = 0;
5797
5798        foreach my $ref_id (sort keys %l_map) {
5799                eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
5800                if ($@) {
5801                        Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
5802                }
5803                $migrated++;
5804        }
5805        $migrated;
5806}
5807
5808sub minimize_connections {
5809        my $r = Git::SVN::read_all_remotes();
5810        my $new_urls = {};
5811        my $root_repos = {};
5812        foreach my $repo_id (keys %$r) {
5813                my $url = $r->{$repo_id}->{url} or next;
5814                my $fetch = $r->{$repo_id}->{fetch} or next;
5815                my $ra = Git::SVN::Ra->new($url);
5816
5817                # skip existing cases where we already connect to the root
5818                if (($ra->{url} eq $ra->{repos_root}) ||
5819                    ($ra->{repos_root} eq $repo_id)) {
5820                        $root_repos->{$ra->{url}} = $repo_id;
5821                        next;
5822                }
5823
5824                my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
5825                my $root_path = $ra->{url};
5826                $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
5827                foreach my $path (keys %$fetch) {
5828                        my $ref_id = $fetch->{$path};
5829                        my $gs = Git::SVN->new($ref_id, $repo_id, $path);
5830
5831                        # make sure we can read when connecting to
5832                        # a higher level of a repository
5833                        my ($last_rev, undef) = $gs->last_rev_commit;
5834                        if (!defined $last_rev) {
5835                                $last_rev = eval {
5836                                        $root_ra->get_latest_revnum;
5837                                };
5838                                next if $@;
5839                        }
5840                        my $new = $root_path;
5841                        $new .= length $path ? "/$path" : '';
5842                        eval {
5843                                $root_ra->get_log([$new], $last_rev, $last_rev,
5844                                                  0, 0, 1, sub { });
5845                        };
5846                        next if $@;
5847                        $new_urls->{$ra->{repos_root}}->{$new} =
5848                                { ref_id => $ref_id,
5849                                  old_repo_id => $repo_id,
5850                                  old_path => $path };
5851                }
5852        }
5853
5854        my @emptied;
5855        foreach my $url (keys %$new_urls) {
5856                # see if we can re-use an existing [svn-remote "repo_id"]
5857                # instead of creating a(n ugly) new section:
5858                my $repo_id = $root_repos->{$url} || $url;
5859
5860                my $fetch = $new_urls->{$url};
5861                foreach my $path (keys %$fetch) {
5862                        my $x = $fetch->{$path};
5863                        Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
5864                        my $pfx = "svn-remote.$x->{old_repo_id}";
5865
5866                        my $old_fetch = quotemeta("$x->{old_path}:".
5867                                                  "$x->{ref_id}");
5868                        command_noisy(qw/config --unset/,
5869                                      "$pfx.fetch", '^'. $old_fetch . '$');
5870                        delete $r->{$x->{old_repo_id}}->
5871                               {fetch}->{$x->{old_path}};
5872                        if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
5873                                command_noisy(qw/config --unset/,
5874                                              "$pfx.url");
5875                                push @emptied, $x->{old_repo_id}
5876                        }
5877                }
5878        }
5879        if (@emptied) {
5880                my $file = $ENV{GIT_CONFIG} || "$ENV{GIT_DIR}/config";
5881                print STDERR <<EOF;
5882The following [svn-remote] sections in your config file ($file) are empty
5883and can be safely removed:
5884EOF
5885                print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
5886        }
5887}
5888
5889sub migration_check {
5890        migrate_from_v0();
5891        migrate_from_v1();
5892        migrate_from_v2();
5893        minimize_connections() if $_minimize;
5894}
5895
5896package Git::IndexInfo;
5897use strict;
5898use warnings;
5899use Git qw/command_input_pipe command_close_pipe/;
5900
5901sub new {
5902        my ($class) = @_;
5903        my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
5904        bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
5905}
5906
5907sub remove {
5908        my ($self, $path) = @_;
5909        if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
5910                return ++$self->{nr};
5911        }
5912        undef;
5913}
5914
5915sub update {
5916        my ($self, $mode, $hash, $path) = @_;
5917        if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
5918                return ++$self->{nr};
5919        }
5920        undef;
5921}
5922
5923sub DESTROY {
5924        my ($self) = @_;
5925        command_close_pipe($self->{gui}, $self->{ctx});
5926}
5927
5928package Git::SVN::GlobSpec;
5929use strict;
5930use warnings;
5931
5932sub new {
5933        my ($class, $glob) = @_;
5934        my $re = $glob;
5935        $re =~ s!/+$!!g; # no need for trailing slashes
5936        $re =~ m!^([^*]*)(\*(?:/\*)*)(.*)$!;
5937        my $temp = $re;
5938        my ($left, $right) = ($1, $3);
5939        $re = $2;
5940        my $depth = $re =~ tr/*/*/;
5941        if ($depth != $temp =~ tr/*/*/) {
5942                die "Only one set of wildcard directories " .
5943                        "(e.g. '*' or '*/*/*') is supported: '$glob'\n";
5944        }
5945        if ($depth == 0) {
5946                die "One '*' is needed for glob: '$glob'\n";
5947        }
5948        $re =~ s!\*!\[^/\]*!g;
5949        $re = quotemeta($left) . "($re)" . quotemeta($right);
5950        if (length $left && !($left =~ s!/+$!!g)) {
5951                die "Missing trailing '/' on left side of: '$glob' ($left)\n";
5952        }
5953        if (length $right && !($right =~ s!^/+!!g)) {
5954                die "Missing leading '/' on right side of: '$glob' ($right)\n";
5955        }
5956        my $left_re = qr/^\/\Q$left\E(\/|$)/;
5957        bless { left => $left, right => $right, left_regex => $left_re,
5958                regex => qr/$re/, glob => $glob, depth => $depth }, $class;
5959}
5960
5961sub full_path {
5962        my ($self, $path) = @_;
5963        return (length $self->{left} ? "$self->{left}/" : '') .
5964               $path . (length $self->{right} ? "/$self->{right}" : '');
5965}
5966
5967__END__
5968
5969Data structures:
5970
5971
5972$remotes = { # returned by read_all_remotes()
5973        'svn' => {
5974                # svn-remote.svn.url=https://svn.musicpd.org
5975                url => 'https://svn.musicpd.org',
5976                # svn-remote.svn.fetch=mpd/trunk:trunk
5977                fetch => {
5978                        'mpd/trunk' => 'trunk',
5979                },
5980                # svn-remote.svn.tags=mpd/tags/*:tags/*
5981                tags => {
5982                        path => {
5983                                left => 'mpd/tags',
5984                                right => '',
5985                                regex => qr!mpd/tags/([^/]+)$!,
5986                                glob => 'tags/*',
5987                        },
5988                        ref => {
5989                                left => 'tags',
5990                                right => '',
5991                                regex => qr!tags/([^/]+)$!,
5992                                glob => 'tags/*',
5993                        },
5994                }
5995        }
5996};
5997
5998$log_entry hashref as returned by libsvn_log_entry()
5999{
6000        log => 'whitespace-formatted log entry
6001',                                              # trailing newline is preserved
6002        revision => '8',                        # integer
6003        date => '2004-02-24T17:01:44.108345Z',  # commit date
6004        author => 'committer name'
6005};
6006
6007
6008# this is generated by generate_diff();
6009@mods = array of diff-index line hashes, each element represents one line
6010        of diff-index output
6011
6012diff-index line ($m hash)
6013{
6014        mode_a => first column of diff-index output, no leading ':',
6015        mode_b => second column of diff-index output,
6016        sha1_b => sha1sum of the final blob,
6017        chg => change type [MCRADT],
6018        file_a => original file name of a file (iff chg is 'C' or 'R')
6019        file_b => new/current file name of a file (any chg)
6020}
6021;
6022
6023# retval of read_url_paths{,_all}();
6024$l_map = {
6025        # repository root url
6026        'https://svn.musicpd.org' => {
6027                # repository path               # GIT_SVN_ID
6028                'mpd/trunk'             =>      'trunk',
6029                'mpd/tags/0.11.5'       =>      'tags/0.11.5',
6030        },
6031}
6032
6033Notes:
6034        I don't trust the each() function on unless I created %hash myself
6035        because the internal iterator may not have started at base.