git-svn.perlon commit Merge branch 'maint' (0bb3a0b)
   1#!/usr/bin/env perl
   2# Copyright (C) 2006, Eric Wong <normalperson@yhbt.net>
   3# License: GPL v2 or later
   4use warnings;
   5use strict;
   6use vars qw/    $AUTHOR $VERSION
   7                $sha1 $sha1_short $_revision $_repository
   8                $_q $_authors %users/;
   9$AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
  10$VERSION = '@@GIT_VERSION@@';
  11
  12# From which subdir have we been invoked?
  13my $cmd_dir_prefix = eval {
  14        command_oneline([qw/rev-parse --show-prefix/], STDERR => 0)
  15} || '';
  16
  17my $git_dir_user_set = 1 if defined $ENV{GIT_DIR};
  18$ENV{GIT_DIR} ||= '.git';
  19$Git::SVN::default_repo_id = 'svn';
  20$Git::SVN::default_ref_id = $ENV{GIT_SVN_ID} || 'git-svn';
  21$Git::SVN::Ra::_log_window_size = 100;
  22
  23$Git::SVN::Log::TZ = $ENV{TZ};
  24$ENV{TZ} = 'UTC';
  25$| = 1; # unbuffer STDOUT
  26
  27sub fatal (@) { print STDERR "@_\n"; exit 1 }
  28require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
  29require SVN::Ra;
  30require SVN::Delta;
  31if ($SVN::Core::VERSION lt '1.1.0') {
  32        fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)";
  33}
  34push @Git::SVN::Ra::ISA, 'SVN::Ra';
  35push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
  36push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
  37use Carp qw/croak/;
  38use Digest::MD5;
  39use IO::File qw//;
  40use File::Basename qw/dirname basename/;
  41use File::Path qw/mkpath/;
  42use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev/;
  43use IPC::Open3;
  44use Git;
  45
  46BEGIN {
  47        # import functions from Git into our packages, en masse
  48        no strict 'refs';
  49        foreach (qw/command command_oneline command_noisy command_output_pipe
  50                    command_input_pipe command_close_pipe/) {
  51                for my $package ( qw(SVN::Git::Editor SVN::Git::Fetcher
  52                        Git::SVN::Migration Git::SVN::Log Git::SVN),
  53                        __PACKAGE__) {
  54                        *{"${package}::$_"} = \&{"Git::$_"};
  55                }
  56        }
  57}
  58
  59my ($SVN);
  60
  61$sha1 = qr/[a-f\d]{40}/;
  62$sha1_short = qr/[a-f\d]{4,40}/;
  63my ($_stdin, $_help, $_edit,
  64        $_message, $_file,
  65        $_template, $_shared,
  66        $_version, $_fetch_all, $_no_rebase,
  67        $_merge, $_strategy, $_dry_run, $_local,
  68        $_prefix, $_no_checkout, $_url, $_verbose,
  69        $_git_format, $_commit_url);
  70$Git::SVN::_follow_parent = 1;
  71my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
  72                    'config-dir=s' => \$Git::SVN::Ra::config_dir,
  73                    'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache );
  74my %fc_opts = ( 'follow-parent|follow!' => \$Git::SVN::_follow_parent,
  75                'authors-file|A=s' => \$_authors,
  76                'repack:i' => \$Git::SVN::_repack,
  77                'noMetadata' => \$Git::SVN::_no_metadata,
  78                'useSvmProps' => \$Git::SVN::_use_svm_props,
  79                'useSvnsyncProps' => \$Git::SVN::_use_svnsync_props,
  80                'log-window-size=i' => \$Git::SVN::Ra::_log_window_size,
  81                'no-checkout' => \$_no_checkout,
  82                'quiet|q' => \$_q,
  83                'repack-flags|repack-args|repack-opts=s' =>
  84                   \$Git::SVN::_repack_flags,
  85                'use-log-author' => \$Git::SVN::_use_log_author,
  86                'add-author-from' => \$Git::SVN::_add_author_from,
  87                %remote_opts );
  88
  89my ($_trunk, $_tags, $_branches, $_stdlayout);
  90my %icv;
  91my %init_opts = ( 'template=s' => \$_template, 'shared:s' => \$_shared,
  92                  'trunk|T=s' => \$_trunk, 'tags|t=s' => \$_tags,
  93                  'branches|b=s' => \$_branches, 'prefix=s' => \$_prefix,
  94                  'stdlayout|s' => \$_stdlayout,
  95                  'minimize-url|m' => \$Git::SVN::_minimize_url,
  96                  'no-metadata' => sub { $icv{noMetadata} = 1 },
  97                  'use-svm-props' => sub { $icv{useSvmProps} = 1 },
  98                  'use-svnsync-props' => sub { $icv{useSvnsyncProps} = 1 },
  99                  'rewrite-root=s' => sub { $icv{rewriteRoot} = $_[1] },
 100                  %remote_opts );
 101my %cmt_opts = ( 'edit|e' => \$_edit,
 102                'rmdir' => \$SVN::Git::Editor::_rmdir,
 103                'find-copies-harder' => \$SVN::Git::Editor::_find_copies_harder,
 104                'l=i' => \$SVN::Git::Editor::_rename_limit,
 105                'copy-similarity|C=i'=> \$SVN::Git::Editor::_cp_similarity
 106);
 107
 108my %cmd = (
 109        fetch => [ \&cmd_fetch, "Download new revisions from SVN",
 110                        { 'revision|r=s' => \$_revision,
 111                          'fetch-all|all' => \$_fetch_all,
 112                           %fc_opts } ],
 113        clone => [ \&cmd_clone, "Initialize and fetch revisions",
 114                        { 'revision|r=s' => \$_revision,
 115                           %fc_opts, %init_opts } ],
 116        init => [ \&cmd_init, "Initialize a repo for tracking" .
 117                          " (requires URL argument)",
 118                          \%init_opts ],
 119        'multi-init' => [ \&cmd_multi_init,
 120                          "Deprecated alias for ".
 121                          "'$0 init -T<trunk> -b<branches> -t<tags>'",
 122                          \%init_opts ],
 123        dcommit => [ \&cmd_dcommit,
 124                     'Commit several diffs to merge with upstream',
 125                        { 'merge|m|M' => \$_merge,
 126                          'strategy|s=s' => \$_strategy,
 127                          'verbose|v' => \$_verbose,
 128                          'dry-run|n' => \$_dry_run,
 129                          'fetch-all|all' => \$_fetch_all,
 130                          'commit-url=s' => \$_commit_url,
 131                          'revision|r=i' => \$_revision,
 132                          'no-rebase' => \$_no_rebase,
 133                        %cmt_opts, %fc_opts } ],
 134        'set-tree' => [ \&cmd_set_tree,
 135                        "Set an SVN repository to a git tree-ish",
 136                        { 'stdin|' => \$_stdin, %cmt_opts, %fc_opts, } ],
 137        'create-ignore' => [ \&cmd_create_ignore,
 138                             'Create a .gitignore per svn:ignore',
 139                             { 'revision|r=i' => \$_revision
 140                             } ],
 141        'propget' => [ \&cmd_propget,
 142                       'Print the value of a property on a file or directory',
 143                       { 'revision|r=i' => \$_revision } ],
 144        'proplist' => [ \&cmd_proplist,
 145                       'List all properties of a file or directory',
 146                       { 'revision|r=i' => \$_revision } ],
 147        'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
 148                        { 'revision|r=i' => \$_revision
 149                        } ],
 150        'show-externals' => [ \&cmd_show_externals, "Show svn:externals listings",
 151                        { 'revision|r=i' => \$_revision
 152                        } ],
 153        'multi-fetch' => [ \&cmd_multi_fetch,
 154                           "Deprecated alias for $0 fetch --all",
 155                           { 'revision|r=s' => \$_revision, %fc_opts } ],
 156        'migrate' => [ sub { },
 157                       # no-op, we automatically run this anyways,
 158                       'Migrate configuration/metadata/layout from
 159                        previous versions of git-svn',
 160                       { 'minimize' => \$Git::SVN::Migration::_minimize,
 161                         %remote_opts } ],
 162        'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
 163                        { 'limit=i' => \$Git::SVN::Log::limit,
 164                          'revision|r=s' => \$_revision,
 165                          'verbose|v' => \$Git::SVN::Log::verbose,
 166                          'incremental' => \$Git::SVN::Log::incremental,
 167                          'oneline' => \$Git::SVN::Log::oneline,
 168                          'show-commit' => \$Git::SVN::Log::show_commit,
 169                          'non-recursive' => \$Git::SVN::Log::non_recursive,
 170                          'authors-file|A=s' => \$_authors,
 171                          'color' => \$Git::SVN::Log::color,
 172                          'pager=s' => \$Git::SVN::Log::pager
 173                        } ],
 174        'find-rev' => [ \&cmd_find_rev,
 175                        "Translate between SVN revision numbers and tree-ish",
 176                        {} ],
 177        'rebase' => [ \&cmd_rebase, "Fetch and rebase your working directory",
 178                        { 'merge|m|M' => \$_merge,
 179                          'verbose|v' => \$_verbose,
 180                          'strategy|s=s' => \$_strategy,
 181                          'local|l' => \$_local,
 182                          'fetch-all|all' => \$_fetch_all,
 183                          'dry-run|n' => \$_dry_run,
 184                          %fc_opts } ],
 185        'commit-diff' => [ \&cmd_commit_diff,
 186                           'Commit a diff between two trees',
 187                        { 'message|m=s' => \$_message,
 188                          'file|F=s' => \$_file,
 189                          'revision|r=s' => \$_revision,
 190                        %cmt_opts } ],
 191        'info' => [ \&cmd_info,
 192                    "Show info about the latest SVN revision
 193                     on the current branch",
 194                    { 'url' => \$_url, } ],
 195        'blame' => [ \&Git::SVN::Log::cmd_blame,
 196                    "Show what revision and author last modified each line of a file",
 197                    { 'git-format' => \$_git_format } ],
 198);
 199
 200my $cmd;
 201for (my $i = 0; $i < @ARGV; $i++) {
 202        if (defined $cmd{$ARGV[$i]}) {
 203                $cmd = $ARGV[$i];
 204                splice @ARGV, $i, 1;
 205                last;
 206        }
 207};
 208
 209# make sure we're always running at the top-level working directory
 210unless ($cmd && $cmd =~ /(?:clone|init|multi-init)$/) {
 211        unless (-d $ENV{GIT_DIR}) {
 212                if ($git_dir_user_set) {
 213                        die "GIT_DIR=$ENV{GIT_DIR} explicitly set, ",
 214                            "but it is not a directory\n";
 215                }
 216                my $git_dir = delete $ENV{GIT_DIR};
 217                chomp(my $cdup = command_oneline(qw/rev-parse --show-cdup/));
 218                unless (length $cdup) {
 219                        die "Already at toplevel, but $git_dir ",
 220                            "not found '$cdup'\n";
 221                }
 222                chdir $cdup or die "Unable to chdir up to '$cdup'\n";
 223                unless (-d $git_dir) {
 224                        die "$git_dir still not found after going to ",
 225                            "'$cdup'\n";
 226                }
 227                $ENV{GIT_DIR} = $git_dir;
 228        }
 229        $_repository = Git->repository(Repository => $ENV{GIT_DIR});
 230}
 231
 232my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
 233
 234read_repo_config(\%opts);
 235if ($cmd && ($cmd eq 'log' || $cmd eq 'blame')) {
 236        Getopt::Long::Configure('pass_through');
 237}
 238my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
 239                    'minimize-connections' => \$Git::SVN::Migration::_minimize,
 240                    'id|i=s' => \$Git::SVN::default_ref_id,
 241                    'svn-remote|remote|R=s' => sub {
 242                       $Git::SVN::no_reuse_existing = 1;
 243                       $Git::SVN::default_repo_id = $_[1] });
 244exit 1 if (!$rv && $cmd && $cmd ne 'log');
 245
 246usage(0) if $_help;
 247version() if $_version;
 248usage(1) unless defined $cmd;
 249load_authors() if $_authors;
 250
 251unless ($cmd =~ /^(?:clone|init|multi-init|commit-diff)$/) {
 252        Git::SVN::Migration::migration_check();
 253}
 254Git::SVN::init_vars();
 255eval {
 256        Git::SVN::verify_remotes_sanity();
 257        $cmd{$cmd}->[0]->(@ARGV);
 258};
 259fatal $@ if $@;
 260post_fetch_checkout();
 261exit 0;
 262
 263####################### primary functions ######################
 264sub usage {
 265        my $exit = shift || 0;
 266        my $fd = $exit ? \*STDERR : \*STDOUT;
 267        print $fd <<"";
 268git-svn - bidirectional operations between a single Subversion tree and git
 269Usage: git svn <command> [options] [arguments]\n
 270
 271        print $fd "Available commands:\n" unless $cmd;
 272
 273        foreach (sort keys %cmd) {
 274                next if $cmd && $cmd ne $_;
 275                next if /^multi-/; # don't show deprecated commands
 276                print $fd '  ',pack('A17',$_),$cmd{$_}->[1],"\n";
 277                foreach (sort keys %{$cmd{$_}->[2]}) {
 278                        # mixed-case options are for .git/config only
 279                        next if /[A-Z]/ && /^[a-z]+$/i;
 280                        # prints out arguments as they should be passed:
 281                        my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
 282                        print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
 283                                                        "--$_" : "-$_" }
 284                                                split /\|/,$_)," $x\n";
 285                }
 286        }
 287        print $fd <<"";
 288\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
 289arbitrary identifier if you're tracking multiple SVN branches/repositories in
 290one git repository and want to keep them separate.  See git-svn(1) for more
 291information.
 292
 293        exit $exit;
 294}
 295
 296sub version {
 297        print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
 298        exit 0;
 299}
 300
 301sub do_git_init_db {
 302        unless (-d $ENV{GIT_DIR}) {
 303                my @init_db = ('init');
 304                push @init_db, "--template=$_template" if defined $_template;
 305                if (defined $_shared) {
 306                        if ($_shared =~ /[a-z]/) {
 307                                push @init_db, "--shared=$_shared";
 308                        } else {
 309                                push @init_db, "--shared";
 310                        }
 311                }
 312                command_noisy(@init_db);
 313                $_repository = Git->repository(Repository => ".git");
 314        }
 315        my $set;
 316        my $pfx = "svn-remote.$Git::SVN::default_repo_id";
 317        foreach my $i (keys %icv) {
 318                die "'$set' and '$i' cannot both be set\n" if $set;
 319                next unless defined $icv{$i};
 320                command_noisy('config', "$pfx.$i", $icv{$i});
 321                $set = $i;
 322        }
 323}
 324
 325sub init_subdir {
 326        my $repo_path = shift or return;
 327        mkpath([$repo_path]) unless -d $repo_path;
 328        chdir $repo_path or die "Couldn't chdir to $repo_path: $!\n";
 329        $ENV{GIT_DIR} = '.git';
 330        $_repository = Git->repository(Repository => $ENV{GIT_DIR});
 331}
 332
 333sub cmd_clone {
 334        my ($url, $path) = @_;
 335        if (!defined $path &&
 336            (defined $_trunk || defined $_branches || defined $_tags ||
 337             defined $_stdlayout) &&
 338            $url !~ m#^[a-z\+]+://#) {
 339                $path = $url;
 340        }
 341        $path = basename($url) if !defined $path || !length $path;
 342        cmd_init($url, $path);
 343        Git::SVN::fetch_all($Git::SVN::default_repo_id);
 344}
 345
 346sub cmd_init {
 347        if (defined $_stdlayout) {
 348                $_trunk = 'trunk' if (!defined $_trunk);
 349                $_tags = 'tags' if (!defined $_tags);
 350                $_branches = 'branches' if (!defined $_branches);
 351        }
 352        if (defined $_trunk || defined $_branches || defined $_tags) {
 353                return cmd_multi_init(@_);
 354        }
 355        my $url = shift or die "SVN repository location required ",
 356                               "as a command-line argument\n";
 357        init_subdir(@_);
 358        do_git_init_db();
 359
 360        Git::SVN->init($url);
 361}
 362
 363sub cmd_fetch {
 364        if (grep /^\d+=./, @_) {
 365                die "'<rev>=<commit>' fetch arguments are ",
 366                    "no longer supported.\n";
 367        }
 368        my ($remote) = @_;
 369        if (@_ > 1) {
 370                die "Usage: $0 fetch [--all] [svn-remote]\n";
 371        }
 372        $remote ||= $Git::SVN::default_repo_id;
 373        if ($_fetch_all) {
 374                cmd_multi_fetch();
 375        } else {
 376                Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
 377        }
 378}
 379
 380sub cmd_set_tree {
 381        my (@commits) = @_;
 382        if ($_stdin || !@commits) {
 383                print "Reading from stdin...\n";
 384                @commits = ();
 385                while (<STDIN>) {
 386                        if (/\b($sha1_short)\b/o) {
 387                                unshift @commits, $1;
 388                        }
 389                }
 390        }
 391        my @revs;
 392        foreach my $c (@commits) {
 393                my @tmp = command('rev-parse',$c);
 394                if (scalar @tmp == 1) {
 395                        push @revs, $tmp[0];
 396                } elsif (scalar @tmp > 1) {
 397                        push @revs, reverse(command('rev-list',@tmp));
 398                } else {
 399                        fatal "Failed to rev-parse $c";
 400                }
 401        }
 402        my $gs = Git::SVN->new;
 403        my ($r_last, $cmt_last) = $gs->last_rev_commit;
 404        $gs->fetch;
 405        if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
 406                fatal "There are new revisions that were fetched ",
 407                      "and need to be merged (or acknowledged) ",
 408                      "before committing.\nlast rev: $r_last\n",
 409                      " current: $gs->{last_rev}";
 410        }
 411        $gs->set_tree($_) foreach @revs;
 412        print "Done committing ",scalar @revs," revisions to SVN\n";
 413        unlink $gs->{index};
 414}
 415
 416sub cmd_dcommit {
 417        my $head = shift;
 418        git_cmd_try { command_oneline(qw/diff-index --quiet HEAD/) }
 419                'Cannot dcommit with a dirty index.  Commit your changes first, '
 420                . "or stash them with `git stash'.\n";
 421        $head ||= 'HEAD';
 422        my @refs;
 423        my ($url, $rev, $uuid, $gs) = working_head_info($head, \@refs);
 424        $url = $_commit_url if defined $_commit_url;
 425        my $last_rev = $_revision if defined $_revision;
 426        if ($url) {
 427                print "Committing to $url ...\n";
 428        }
 429        unless ($gs) {
 430                die "Unable to determine upstream SVN information from ",
 431                    "$head history.\nPerhaps the repository is empty.";
 432        }
 433        my ($linear_refs, $parents) = linearize_history($gs, \@refs);
 434        if ($_no_rebase && scalar(@$linear_refs) > 1) {
 435                warn "Attempting to commit more than one change while ",
 436                     "--no-rebase is enabled.\n",
 437                     "If these changes depend on each other, re-running ",
 438                     "without --no-rebase may be required."
 439        }
 440        while (1) {
 441                my $d = shift @$linear_refs or last;
 442                unless (defined $last_rev) {
 443                        (undef, $last_rev, undef) = cmt_metadata("$d~1");
 444                        unless (defined $last_rev) {
 445                                fatal "Unable to extract revision information ",
 446                                      "from commit $d~1";
 447                        }
 448                }
 449                if ($_dry_run) {
 450                        print "diff-tree $d~1 $d\n";
 451                } else {
 452                        my $cmt_rev;
 453                        my %ed_opts = ( r => $last_rev,
 454                                        log => get_commit_entry($d)->{log},
 455                                        ra => Git::SVN::Ra->new($url),
 456                                        config => SVN::Core::config_get_config(
 457                                                $Git::SVN::Ra::config_dir
 458                                        ),
 459                                        tree_a => "$d~1",
 460                                        tree_b => $d,
 461                                        editor_cb => sub {
 462                                               print "Committed r$_[0]\n";
 463                                               $cmt_rev = $_[0];
 464                                        },
 465                                        svn_path => '');
 466                        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
 467                                print "No changes\n$d~1 == $d\n";
 468                        } elsif ($parents->{$d} && @{$parents->{$d}}) {
 469                                $gs->{inject_parents_dcommit}->{$cmt_rev} =
 470                                                               $parents->{$d};
 471                        }
 472                        $_fetch_all ? $gs->fetch_all : $gs->fetch;
 473                        $last_rev = $cmt_rev;
 474                        next if $_no_rebase;
 475
 476                        # we always want to rebase against the current HEAD,
 477                        # not any head that was passed to us
 478                        my @diff = command('diff-tree', $d,
 479                                           $gs->refname, '--');
 480                        my @finish;
 481                        if (@diff) {
 482                                @finish = rebase_cmd();
 483                                print STDERR "W: $d and ", $gs->refname,
 484                                             " differ, using @finish:\n",
 485                                             join("\n", @diff), "\n";
 486                        } else {
 487                                print "No changes between current HEAD and ",
 488                                      $gs->refname,
 489                                      "\nResetting to the latest ",
 490                                      $gs->refname, "\n";
 491                                @finish = qw/reset --mixed/;
 492                        }
 493                        command_noisy(@finish, $gs->refname);
 494                        if (@diff) {
 495                                @refs = ();
 496                                my ($url_, $rev_, $uuid_, $gs_) =
 497                                              working_head_info($head, \@refs);
 498                                my ($linear_refs_, $parents_) =
 499                                              linearize_history($gs_, \@refs);
 500                                if (scalar(@$linear_refs) !=
 501                                    scalar(@$linear_refs_)) {
 502                                        fatal "# of revisions changed ",
 503                                          "\nbefore:\n",
 504                                          join("\n", @$linear_refs),
 505                                          "\n\nafter:\n",
 506                                          join("\n", @$linear_refs_), "\n",
 507                                          'If you are attempting to commit ',
 508                                          "merges, try running:\n\t",
 509                                          'git rebase --interactive',
 510                                          '--preserve-merges ',
 511                                          $gs->refname,
 512                                          "\nBefore dcommitting";
 513                                }
 514                                if ($url_ ne $url) {
 515                                        fatal "URL mismatch after rebase: ",
 516                                              "$url_ != $url";
 517                                }
 518                                if ($uuid_ ne $uuid) {
 519                                        fatal "uuid mismatch after rebase: ",
 520                                              "$uuid_ != $uuid";
 521                                }
 522                                # remap parents
 523                                my (%p, @l, $i);
 524                                for ($i = 0; $i < scalar @$linear_refs; $i++) {
 525                                        my $new = $linear_refs_->[$i] or next;
 526                                        $p{$new} =
 527                                                $parents->{$linear_refs->[$i]};
 528                                        push @l, $new;
 529                                }
 530                                $parents = \%p;
 531                                $linear_refs = \@l;
 532                        }
 533                }
 534        }
 535        unlink $gs->{index};
 536}
 537
 538sub cmd_find_rev {
 539        my $revision_or_hash = shift or die "SVN or git revision required ",
 540                                            "as a command-line argument\n";
 541        my $result;
 542        if ($revision_or_hash =~ /^r\d+$/) {
 543                my $head = shift;
 544                $head ||= 'HEAD';
 545                my @refs;
 546                my (undef, undef, $uuid, $gs) = working_head_info($head, \@refs);
 547                unless ($gs) {
 548                        die "Unable to determine upstream SVN information from ",
 549                            "$head history\n";
 550                }
 551                my $desired_revision = substr($revision_or_hash, 1);
 552                $result = $gs->rev_map_get($desired_revision, $uuid);
 553        } else {
 554                my (undef, $rev, undef) = cmt_metadata($revision_or_hash);
 555                $result = $rev;
 556        }
 557        print "$result\n" if $result;
 558}
 559
 560sub cmd_rebase {
 561        command_noisy(qw/update-index --refresh/);
 562        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 563        unless ($gs) {
 564                die "Unable to determine upstream SVN information from ",
 565                    "working tree history\n";
 566        }
 567        if ($_dry_run) {
 568                print "Remote Branch: " . $gs->refname . "\n";
 569                print "SVN URL: " . $url . "\n";
 570                return;
 571        }
 572        if (command(qw/diff-index HEAD --/)) {
 573                print STDERR "Cannot rebase with uncommited changes:\n";
 574                command_noisy('status');
 575                exit 1;
 576        }
 577        unless ($_local) {
 578                # rebase will checkout for us, so no need to do it explicitly
 579                $_no_checkout = 'true';
 580                $_fetch_all ? $gs->fetch_all : $gs->fetch;
 581        }
 582        command_noisy(rebase_cmd(), $gs->refname);
 583}
 584
 585sub cmd_show_ignore {
 586        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 587        $gs ||= Git::SVN->new;
 588        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 589        $gs->prop_walk($gs->{path}, $r, sub {
 590                my ($gs, $path, $props) = @_;
 591                print STDOUT "\n# $path\n";
 592                my $s = $props->{'svn:ignore'} or return;
 593                $s =~ s/[\r\n]+/\n/g;
 594                chomp $s;
 595                $s =~ s#^#$path#gm;
 596                print STDOUT "$s\n";
 597        });
 598}
 599
 600sub cmd_show_externals {
 601        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 602        $gs ||= Git::SVN->new;
 603        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 604        $gs->prop_walk($gs->{path}, $r, sub {
 605                my ($gs, $path, $props) = @_;
 606                print STDOUT "\n# $path\n";
 607                my $s = $props->{'svn:externals'} or return;
 608                $s =~ s/[\r\n]+/\n/g;
 609                chomp $s;
 610                $s =~ s#^#$path#gm;
 611                print STDOUT "$s\n";
 612        });
 613}
 614
 615sub cmd_create_ignore {
 616        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 617        $gs ||= Git::SVN->new;
 618        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 619        $gs->prop_walk($gs->{path}, $r, sub {
 620                my ($gs, $path, $props) = @_;
 621                # $path is of the form /path/to/dir/
 622                my $ignore = '.' . $path . '.gitignore';
 623                my $s = $props->{'svn:ignore'} or return;
 624                open(GITIGNORE, '>', $ignore)
 625                  or fatal("Failed to open `$ignore' for writing: $!");
 626                $s =~ s/[\r\n]+/\n/g;
 627                chomp $s;
 628                # Prefix all patterns so that the ignore doesn't apply
 629                # to sub-directories.
 630                $s =~ s#^#/#gm;
 631                print GITIGNORE "$s\n";
 632                close(GITIGNORE)
 633                  or fatal("Failed to close `$ignore': $!");
 634                command_noisy('add', '-f', $ignore);
 635        });
 636}
 637
 638sub canonicalize_path {
 639        my ($path) = @_;
 640        my $dot_slash_added = 0;
 641        if (substr($path, 0, 1) ne "/") {
 642                $path = "./" . $path;
 643                $dot_slash_added = 1;
 644        }
 645        # File::Spec->canonpath doesn't collapse x/../y into y (for a
 646        # good reason), so let's do this manually.
 647        $path =~ s#/+#/#g;
 648        $path =~ s#/\.(?:/|$)#/#g;
 649        $path =~ s#/[^/]+/\.\.##g;
 650        $path =~ s#/$##g;
 651        $path =~ s#^\./## if $dot_slash_added;
 652        $path =~ s#^/##;
 653        $path =~ s#^\.$##;
 654        return $path;
 655}
 656
 657# get_svnprops(PATH)
 658# ------------------
 659# Helper for cmd_propget and cmd_proplist below.
 660sub get_svnprops {
 661        my $path = shift;
 662        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 663        $gs ||= Git::SVN->new;
 664
 665        # prefix THE PATH by the sub-directory from which the user
 666        # invoked us.
 667        $path = $cmd_dir_prefix . $path;
 668        fatal("No such file or directory: $path") unless -e $path;
 669        my $is_dir = -d $path ? 1 : 0;
 670        $path = $gs->{path} . '/' . $path;
 671
 672        # canonicalize the path (otherwise libsvn will abort or fail to
 673        # find the file)
 674        $path = canonicalize_path($path);
 675
 676        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 677        my $props;
 678        if ($is_dir) {
 679                (undef, undef, $props) = $gs->ra->get_dir($path, $r);
 680        }
 681        else {
 682                (undef, $props) = $gs->ra->get_file($path, $r, undef);
 683        }
 684        return $props;
 685}
 686
 687# cmd_propget (PROP, PATH)
 688# ------------------------
 689# Print the SVN property PROP for PATH.
 690sub cmd_propget {
 691        my ($prop, $path) = @_;
 692        $path = '.' if not defined $path;
 693        usage(1) if not defined $prop;
 694        my $props = get_svnprops($path);
 695        if (not defined $props->{$prop}) {
 696                fatal("`$path' does not have a `$prop' SVN property.");
 697        }
 698        print $props->{$prop} . "\n";
 699}
 700
 701# cmd_proplist (PATH)
 702# -------------------
 703# Print the list of SVN properties for PATH.
 704sub cmd_proplist {
 705        my $path = shift;
 706        $path = '.' if not defined $path;
 707        my $props = get_svnprops($path);
 708        print "Properties on '$path':\n";
 709        foreach (sort keys %{$props}) {
 710                print "  $_\n";
 711        }
 712}
 713
 714sub cmd_multi_init {
 715        my $url = shift;
 716        unless (defined $_trunk || defined $_branches || defined $_tags) {
 717                usage(1);
 718        }
 719
 720        # there are currently some bugs that prevent multi-init/multi-fetch
 721        # setups from working well without this.
 722        $Git::SVN::_minimize_url = 1;
 723
 724        $_prefix = '' unless defined $_prefix;
 725        if (defined $url) {
 726                $url =~ s#/+$##;
 727                init_subdir(@_);
 728        }
 729        do_git_init_db();
 730        if (defined $_trunk) {
 731                my $trunk_ref = $_prefix . 'trunk';
 732                # try both old-style and new-style lookups:
 733                my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
 734                unless ($gs_trunk) {
 735                        my ($trunk_url, $trunk_path) =
 736                                              complete_svn_url($url, $_trunk);
 737                        $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
 738                                                   undef, $trunk_ref);
 739                }
 740        }
 741        return unless defined $_branches || defined $_tags;
 742        my $ra = $url ? Git::SVN::Ra->new($url) : undef;
 743        complete_url_ls_init($ra, $_branches, '--branches/-b', $_prefix);
 744        complete_url_ls_init($ra, $_tags, '--tags/-t', $_prefix . 'tags/');
 745}
 746
 747sub cmd_multi_fetch {
 748        my $remotes = Git::SVN::read_all_remotes();
 749        foreach my $repo_id (sort keys %$remotes) {
 750                if ($remotes->{$repo_id}->{url}) {
 751                        Git::SVN::fetch_all($repo_id, $remotes);
 752                }
 753        }
 754}
 755
 756# this command is special because it requires no metadata
 757sub cmd_commit_diff {
 758        my ($ta, $tb, $url) = @_;
 759        my $usage = "Usage: $0 commit-diff -r<revision> ".
 760                    "<tree-ish> <tree-ish> [<URL>]";
 761        fatal($usage) if (!defined $ta || !defined $tb);
 762        my $svn_path = '';
 763        if (!defined $url) {
 764                my $gs = eval { Git::SVN->new };
 765                if (!$gs) {
 766                        fatal("Needed URL or usable git-svn --id in ",
 767                              "the command-line\n", $usage);
 768                }
 769                $url = $gs->{url};
 770                $svn_path = $gs->{path};
 771        }
 772        unless (defined $_revision) {
 773                fatal("-r|--revision is a required argument\n", $usage);
 774        }
 775        if (defined $_message && defined $_file) {
 776                fatal("Both --message/-m and --file/-F specified ",
 777                      "for the commit message.\n",
 778                      "I have no idea what you mean");
 779        }
 780        if (defined $_file) {
 781                $_message = file_to_s($_file);
 782        } else {
 783                $_message ||= get_commit_entry($tb)->{log};
 784        }
 785        my $ra ||= Git::SVN::Ra->new($url);
 786        my $r = $_revision;
 787        if ($r eq 'HEAD') {
 788                $r = $ra->get_latest_revnum;
 789        } elsif ($r !~ /^\d+$/) {
 790                die "revision argument: $r not understood by git-svn\n";
 791        }
 792        my %ed_opts = ( r => $r,
 793                        log => $_message,
 794                        ra => $ra,
 795                        tree_a => $ta,
 796                        tree_b => $tb,
 797                        editor_cb => sub { print "Committed r$_[0]\n" },
 798                        svn_path => $svn_path );
 799        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
 800                print "No changes\n$ta == $tb\n";
 801        }
 802}
 803
 804sub cmd_info {
 805        my $path = canonicalize_path(defined($_[0]) ? $_[0] : ".");
 806        if (exists $_[1]) {
 807                die "Too many arguments specified\n";
 808        }
 809
 810        my ($file_type, $diff_status) = find_file_type_and_diff_status($path);
 811
 812        if (!$file_type && !$diff_status) {
 813                print STDERR "$path:  (Not a versioned resource)\n\n";
 814                return;
 815        }
 816
 817        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 818        unless ($gs) {
 819                die "Unable to determine upstream SVN information from ",
 820                    "working tree history\n";
 821        }
 822
 823        # canonicalize_path() will return "" to make libsvn 1.5.x happy,
 824        $path = "." if $path eq "";
 825
 826        my $full_url = $url . ($path eq "." ? "" : "/$path");
 827
 828        if ($_url) {
 829                print $full_url, "\n";
 830                return;
 831        }
 832
 833        my $result = "Path: $path\n";
 834        $result .= "Name: " . basename($path) . "\n" if $file_type ne "dir";
 835        $result .= "URL: " . $full_url . "\n";
 836
 837        eval {
 838                my $repos_root = $gs->repos_root;
 839                Git::SVN::remove_username($repos_root);
 840                $result .= "Repository Root: $repos_root\n";
 841        };
 842        if ($@) {
 843                $result .= "Repository Root: (offline)\n";
 844        }
 845        $result .= "Repository UUID: $uuid\n" unless $diff_status eq "A";
 846        $result .= "Revision: " . ($diff_status eq "A" ? 0 : $rev) . "\n";
 847
 848        $result .= "Node Kind: " .
 849                   ($file_type eq "dir" ? "directory" : "file") . "\n";
 850
 851        my $schedule = $diff_status eq "A"
 852                       ? "add"
 853                       : ($diff_status eq "D" ? "delete" : "normal");
 854        $result .= "Schedule: $schedule\n";
 855
 856        if ($diff_status eq "A") {
 857                print $result, "\n";
 858                return;
 859        }
 860
 861        my ($lc_author, $lc_rev, $lc_date_utc);
 862        my @args = Git::SVN::Log::git_svn_log_cmd($rev, $rev, "--", $path);
 863        my $log = command_output_pipe(@args);
 864        my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
 865        while (<$log>) {
 866                if (/^${esc_color}author (.+) <[^>]+> (\d+) ([\-\+]?\d+)$/o) {
 867                        $lc_author = $1;
 868                        $lc_date_utc = Git::SVN::Log::parse_git_date($2, $3);
 869                } elsif (/^${esc_color}    (git-svn-id:.+)$/o) {
 870                        (undef, $lc_rev, undef) = ::extract_metadata($1);
 871                }
 872        }
 873        close $log;
 874
 875        Git::SVN::Log::set_local_timezone();
 876
 877        $result .= "Last Changed Author: $lc_author\n";
 878        $result .= "Last Changed Rev: $lc_rev\n";
 879        $result .= "Last Changed Date: " .
 880                   Git::SVN::Log::format_svn_date($lc_date_utc) . "\n";
 881
 882        if ($file_type ne "dir") {
 883                my $text_last_updated_date =
 884                    ($diff_status eq "D" ? $lc_date_utc : (stat $path)[9]);
 885                $result .=
 886                    "Text Last Updated: " .
 887                    Git::SVN::Log::format_svn_date($text_last_updated_date) .
 888                    "\n";
 889                my $checksum;
 890                if ($diff_status eq "D") {
 891                        my ($fh, $ctx) =
 892                            command_output_pipe(qw(cat-file blob), "HEAD:$path");
 893                        if ($file_type eq "link") {
 894                                my $file_name = <$fh>;
 895                                $checksum = md5sum("link $file_name");
 896                        } else {
 897                                $checksum = md5sum($fh);
 898                        }
 899                        command_close_pipe($fh, $ctx);
 900                } elsif ($file_type eq "link") {
 901                        my $file_name =
 902                            command(qw(cat-file blob), "HEAD:$path");
 903                        $checksum =
 904                            md5sum("link " . $file_name);
 905                } else {
 906                        open FILE, "<", $path or die $!;
 907                        $checksum = md5sum(\*FILE);
 908                        close FILE or die $!;
 909                }
 910                $result .= "Checksum: " . $checksum . "\n";
 911        }
 912
 913        print $result, "\n";
 914}
 915
 916########################### utility functions #########################
 917
 918sub rebase_cmd {
 919        my @cmd = qw/rebase/;
 920        push @cmd, '-v' if $_verbose;
 921        push @cmd, qw/--merge/ if $_merge;
 922        push @cmd, "--strategy=$_strategy" if $_strategy;
 923        @cmd;
 924}
 925
 926sub post_fetch_checkout {
 927        return if $_no_checkout;
 928        my $gs = $Git::SVN::_head or return;
 929        return if verify_ref('refs/heads/master^0');
 930
 931        my $valid_head = verify_ref('HEAD^0');
 932        command_noisy(qw(update-ref refs/heads/master), $gs->refname);
 933        return if ($valid_head || !verify_ref('HEAD^0'));
 934
 935        return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
 936        my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
 937        return if -f $index;
 938
 939        return if command_oneline(qw/rev-parse --is-inside-work-tree/) eq 'false';
 940        return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
 941        command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
 942        print STDERR "Checked out HEAD:\n  ",
 943                     $gs->full_url, " r", $gs->last_rev, "\n";
 944}
 945
 946sub complete_svn_url {
 947        my ($url, $path) = @_;
 948        $path =~ s#/+$##;
 949        if ($path !~ m#^[a-z\+]+://#) {
 950                if (!defined $url || $url !~ m#^[a-z\+]+://#) {
 951                        fatal("E: '$path' is not a complete URL ",
 952                              "and a separate URL is not specified");
 953                }
 954                return ($url, $path);
 955        }
 956        return ($path, '');
 957}
 958
 959sub complete_url_ls_init {
 960        my ($ra, $repo_path, $switch, $pfx) = @_;
 961        unless ($repo_path) {
 962                print STDERR "W: $switch not specified\n";
 963                return;
 964        }
 965        $repo_path =~ s#/+$##;
 966        if ($repo_path =~ m#^[a-z\+]+://#) {
 967                $ra = Git::SVN::Ra->new($repo_path);
 968                $repo_path = '';
 969        } else {
 970                $repo_path =~ s#^/+##;
 971                unless ($ra) {
 972                        fatal("E: '$repo_path' is not a complete URL ",
 973                              "and a separate URL is not specified");
 974                }
 975        }
 976        my $url = $ra->{url};
 977        my $gs = Git::SVN->init($url, undef, undef, undef, 1);
 978        my $k = "svn-remote.$gs->{repo_id}.url";
 979        my $orig_url = eval { command_oneline(qw/config --get/, $k) };
 980        if ($orig_url && ($orig_url ne $gs->{url})) {
 981                die "$k already set: $orig_url\n",
 982                    "wanted to set to: $gs->{url}\n";
 983        }
 984        command_oneline('config', $k, $gs->{url}) unless $orig_url;
 985        my $remote_path = "$ra->{svn_path}/$repo_path";
 986        $remote_path =~ s#/+#/#g;
 987        $remote_path =~ s#^/##g;
 988        $remote_path .= "/*" if $remote_path !~ /\*/;
 989        my ($n) = ($switch =~ /^--(\w+)/);
 990        if (length $pfx && $pfx !~ m#/$#) {
 991                die "--prefix='$pfx' must have a trailing slash '/'\n";
 992        }
 993        command_noisy('config',
 994                      "svn-remote.$gs->{repo_id}.$n",
 995                      "$remote_path:refs/remotes/$pfx*" .
 996                        ('/*' x (($remote_path =~ tr/*/*/) - 1)) );
 997}
 998
 999sub verify_ref {
1000        my ($ref) = @_;
1001        eval { command_oneline([ 'rev-parse', '--verify', $ref ],
1002                               { STDERR => 0 }); };
1003}
1004
1005sub get_tree_from_treeish {
1006        my ($treeish) = @_;
1007        # $treeish can be a symbolic ref, too:
1008        my $type = command_oneline(qw/cat-file -t/, $treeish);
1009        my $expected;
1010        while ($type eq 'tag') {
1011                ($treeish, $type) = command(qw/cat-file tag/, $treeish);
1012        }
1013        if ($type eq 'commit') {
1014                $expected = (grep /^tree /, command(qw/cat-file commit/,
1015                                                    $treeish))[0];
1016                ($expected) = ($expected =~ /^tree ($sha1)$/o);
1017                die "Unable to get tree from $treeish\n" unless $expected;
1018        } elsif ($type eq 'tree') {
1019                $expected = $treeish;
1020        } else {
1021                die "$treeish is a $type, expected tree, tag or commit\n";
1022        }
1023        return $expected;
1024}
1025
1026sub get_commit_entry {
1027        my ($treeish) = shift;
1028        my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
1029        my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
1030        my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
1031        open my $log_fh, '>', $commit_editmsg or croak $!;
1032
1033        my $type = command_oneline(qw/cat-file -t/, $treeish);
1034        if ($type eq 'commit' || $type eq 'tag') {
1035                my ($msg_fh, $ctx) = command_output_pipe('cat-file',
1036                                                         $type, $treeish);
1037                my $in_msg = 0;
1038                my $author;
1039                my $saw_from = 0;
1040                my $msgbuf = "";
1041                while (<$msg_fh>) {
1042                        if (!$in_msg) {
1043                                $in_msg = 1 if (/^\s*$/);
1044                                $author = $1 if (/^author (.*>)/);
1045                        } elsif (/^git-svn-id: /) {
1046                                # skip this for now, we regenerate the
1047                                # correct one on re-fetch anyways
1048                                # TODO: set *:merge properties or like...
1049                        } else {
1050                                if (/^From:/ || /^Signed-off-by:/) {
1051                                        $saw_from = 1;
1052                                }
1053                                $msgbuf .= $_;
1054                        }
1055                }
1056                $msgbuf =~ s/\s+$//s;
1057                if ($Git::SVN::_add_author_from && defined($author)
1058                    && !$saw_from) {
1059                        $msgbuf .= "\n\nFrom: $author";
1060                }
1061                print $log_fh $msgbuf or croak $!;
1062                command_close_pipe($msg_fh, $ctx);
1063        }
1064        close $log_fh or croak $!;
1065
1066        if ($_edit || ($type eq 'tree')) {
1067                my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
1068                # TODO: strip out spaces, comments, like git-commit.sh
1069                system($editor, $commit_editmsg);
1070        }
1071        rename $commit_editmsg, $commit_msg or croak $!;
1072        open $log_fh, '<', $commit_msg or croak $!;
1073        { local $/; chomp($log_entry{log} = <$log_fh>); }
1074        close $log_fh or croak $!;
1075        unlink $commit_msg;
1076        \%log_entry;
1077}
1078
1079sub s_to_file {
1080        my ($str, $file, $mode) = @_;
1081        open my $fd,'>',$file or croak $!;
1082        print $fd $str,"\n" or croak $!;
1083        close $fd or croak $!;
1084        chmod ($mode &~ umask, $file) if (defined $mode);
1085}
1086
1087sub file_to_s {
1088        my $file = shift;
1089        open my $fd,'<',$file or croak "$!: file: $file\n";
1090        local $/;
1091        my $ret = <$fd>;
1092        close $fd or croak $!;
1093        $ret =~ s/\s*$//s;
1094        return $ret;
1095}
1096
1097# '<svn username> = real-name <email address>' mapping based on git-svnimport:
1098sub load_authors {
1099        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1100        my $log = $cmd eq 'log';
1101        while (<$authors>) {
1102                chomp;
1103                next unless /^(.+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
1104                my ($user, $name, $email) = ($1, $2, $3);
1105                if ($log) {
1106                        $Git::SVN::Log::rusers{"$name <$email>"} = $user;
1107                } else {
1108                        $users{$user} = [$name, $email];
1109                }
1110        }
1111        close $authors or croak $!;
1112}
1113
1114# convert GetOpt::Long specs for use by git-config
1115sub read_repo_config {
1116        return unless -d $ENV{GIT_DIR};
1117        my $opts = shift;
1118        my @config_only;
1119        foreach my $o (keys %$opts) {
1120                # if we have mixedCase and a long option-only, then
1121                # it's a config-only variable that we don't need for
1122                # the command-line.
1123                push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
1124                my $v = $opts->{$o};
1125                my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
1126                $key =~ s/-//g;
1127                my $arg = 'git-config';
1128                $arg .= ' --int' if ($o =~ /[:=]i$/);
1129                $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
1130                if (ref $v eq 'ARRAY') {
1131                        chomp(my @tmp = `$arg --get-all svn.$key`);
1132                        @$v = @tmp if @tmp;
1133                } else {
1134                        chomp(my $tmp = `$arg --get svn.$key`);
1135                        if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
1136                                $$v = $tmp;
1137                        }
1138                }
1139        }
1140        delete @$opts{@config_only} if @config_only;
1141}
1142
1143sub extract_metadata {
1144        my $id = shift or return (undef, undef, undef);
1145        my ($url, $rev, $uuid) = ($id =~ /^\s*git-svn-id:\s+(.*)\@(\d+)
1146                                                        \s([a-f\d\-]+)$/x);
1147        if (!defined $rev || !$uuid || !$url) {
1148                # some of the original repositories I made had
1149                # identifiers like this:
1150                ($rev, $uuid) = ($id =~/^\s*git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
1151        }
1152        return ($url, $rev, $uuid);
1153}
1154
1155sub cmt_metadata {
1156        return extract_metadata((grep(/^git-svn-id: /,
1157                command(qw/cat-file commit/, shift)))[-1]);
1158}
1159
1160sub working_head_info {
1161        my ($head, $refs) = @_;
1162        my @args = ('log', '--no-color', '--first-parent', '--pretty=medium');
1163        my ($fh, $ctx) = command_output_pipe(@args, $head);
1164        my $hash;
1165        my %max;
1166        while (<$fh>) {
1167                if ( m{^commit ($::sha1)$} ) {
1168                        unshift @$refs, $hash if $hash and $refs;
1169                        $hash = $1;
1170                        next;
1171                }
1172                next unless s{^\s*(git-svn-id:)}{$1};
1173                my ($url, $rev, $uuid) = extract_metadata($_);
1174                if (defined $url && defined $rev) {
1175                        next if $max{$url} and $max{$url} < $rev;
1176                        if (my $gs = Git::SVN->find_by_url($url)) {
1177                                my $c = $gs->rev_map_get($rev, $uuid);
1178                                if ($c && $c eq $hash) {
1179                                        close $fh; # break the pipe
1180                                        return ($url, $rev, $uuid, $gs);
1181                                } else {
1182                                        $max{$url} ||= $gs->rev_map_max;
1183                                }
1184                        }
1185                }
1186        }
1187        command_close_pipe($fh, $ctx);
1188        (undef, undef, undef, undef);
1189}
1190
1191sub read_commit_parents {
1192        my ($parents, $c) = @_;
1193        chomp(my $p = command_oneline(qw/rev-list --parents -1/, $c));
1194        $p =~ s/^($c)\s*// or die "rev-list --parents -1 $c failed!\n";
1195        @{$parents->{$c}} = split(/ /, $p);
1196}
1197
1198sub linearize_history {
1199        my ($gs, $refs) = @_;
1200        my %parents;
1201        foreach my $c (@$refs) {
1202                read_commit_parents(\%parents, $c);
1203        }
1204
1205        my @linear_refs;
1206        my %skip = ();
1207        my $last_svn_commit = $gs->last_commit;
1208        foreach my $c (reverse @$refs) {
1209                next if $c eq $last_svn_commit;
1210                last if $skip{$c};
1211
1212                unshift @linear_refs, $c;
1213                $skip{$c} = 1;
1214
1215                # we only want the first parent to diff against for linear
1216                # history, we save the rest to inject when we finalize the
1217                # svn commit
1218                my $fp_a = verify_ref("$c~1");
1219                my $fp_b = shift @{$parents{$c}} if $parents{$c};
1220                if (!$fp_a || !$fp_b) {
1221                        die "Commit $c\n",
1222                            "has no parent commit, and therefore ",
1223                            "nothing to diff against.\n",
1224                            "You should be working from a repository ",
1225                            "originally created by git-svn\n";
1226                }
1227                if ($fp_a ne $fp_b) {
1228                        die "$c~1 = $fp_a, however parsing commit $c ",
1229                            "revealed that:\n$c~1 = $fp_b\nBUG!\n";
1230                }
1231
1232                foreach my $p (@{$parents{$c}}) {
1233                        $skip{$p} = 1;
1234                }
1235        }
1236        (\@linear_refs, \%parents);
1237}
1238
1239sub find_file_type_and_diff_status {
1240        my ($path) = @_;
1241        return ('dir', '') if $path eq '';
1242
1243        my $diff_output =
1244            command_oneline(qw(diff --cached --name-status --), $path) || "";
1245        my $diff_status = (split(' ', $diff_output))[0] || "";
1246
1247        my $ls_tree = command_oneline(qw(ls-tree HEAD), $path) || "";
1248
1249        return (undef, undef) if !$diff_status && !$ls_tree;
1250
1251        if ($diff_status eq "A") {
1252                return ("link", $diff_status) if -l $path;
1253                return ("dir", $diff_status) if -d $path;
1254                return ("file", $diff_status);
1255        }
1256
1257        my $mode = (split(' ', $ls_tree))[0] || "";
1258
1259        return ("link", $diff_status) if $mode eq "120000";
1260        return ("dir", $diff_status) if $mode eq "040000";
1261        return ("file", $diff_status);
1262}
1263
1264sub md5sum {
1265        my $arg = shift;
1266        my $ref = ref $arg;
1267        my $md5 = Digest::MD5->new();
1268        if ($ref eq 'GLOB' || $ref eq 'IO::File') {
1269                $md5->addfile($arg) or croak $!;
1270        } elsif ($ref eq 'SCALAR') {
1271                $md5->add($$arg) or croak $!;
1272        } elsif (!$ref) {
1273                $md5->add($arg) or croak $!;
1274        } else {
1275                ::fatal "Can't provide MD5 hash for unknown ref type: '", $ref, "'";
1276        }
1277        return $md5->hexdigest();
1278}
1279
1280package Git::SVN;
1281use strict;
1282use warnings;
1283use Fcntl qw/:DEFAULT :seek/;
1284use constant rev_map_fmt => 'NH40';
1285use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
1286            $_repack $_repack_flags $_use_svm_props $_head
1287            $_use_svnsync_props $no_reuse_existing $_minimize_url
1288            $_use_log_author $_add_author_from/;
1289use Carp qw/croak/;
1290use File::Path qw/mkpath/;
1291use File::Copy qw/copy/;
1292use IPC::Open3;
1293
1294my ($_gc_nr, $_gc_period);
1295
1296# properties that we do not log:
1297my %SKIP_PROP;
1298BEGIN {
1299        %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
1300                                        svn:special svn:executable
1301                                        svn:entry:committed-rev
1302                                        svn:entry:last-author
1303                                        svn:entry:uuid
1304                                        svn:entry:committed-date/;
1305
1306        # some options are read globally, but can be overridden locally
1307        # per [svn-remote "..."] section.  Command-line options will *NOT*
1308        # override options set in an [svn-remote "..."] section
1309        no strict 'refs';
1310        for my $option (qw/follow_parent no_metadata use_svm_props
1311                           use_svnsync_props/) {
1312                my $key = $option;
1313                $key =~ tr/_//d;
1314                my $prop = "-$option";
1315                *$option = sub {
1316                        my ($self) = @_;
1317                        return $self->{$prop} if exists $self->{$prop};
1318                        my $k = "svn-remote.$self->{repo_id}.$key";
1319                        eval { command_oneline(qw/config --get/, $k) };
1320                        if ($@) {
1321                                $self->{$prop} = ${"Git::SVN::_$option"};
1322                        } else {
1323                                my $v = command_oneline(qw/config --bool/,$k);
1324                                $self->{$prop} = $v eq 'false' ? 0 : 1;
1325                        }
1326                        return $self->{$prop};
1327                }
1328        }
1329}
1330
1331my (%LOCKFILES, %INDEX_FILES);
1332END {
1333        unlink keys %LOCKFILES if %LOCKFILES;
1334        unlink keys %INDEX_FILES if %INDEX_FILES;
1335}
1336
1337sub resolve_local_globs {
1338        my ($url, $fetch, $glob_spec) = @_;
1339        return unless defined $glob_spec;
1340        my $ref = $glob_spec->{ref};
1341        my $path = $glob_spec->{path};
1342        foreach (command(qw#for-each-ref --format=%(refname) refs/remotes#)) {
1343                next unless m#^refs/remotes/$ref->{regex}$#;
1344                my $p = $1;
1345                my $pathname = desanitize_refname($path->full_path($p));
1346                my $refname = desanitize_refname($ref->full_path($p));
1347                if (my $existing = $fetch->{$pathname}) {
1348                        if ($existing ne $refname) {
1349                                die "Refspec conflict:\n",
1350                                    "existing: refs/remotes/$existing\n",
1351                                    " globbed: refs/remotes/$refname\n";
1352                        }
1353                        my $u = (::cmt_metadata("refs/remotes/$refname"))[0];
1354                        $u =~ s!^\Q$url\E(/|$)!! or die
1355                          "refs/remotes/$refname: '$url' not found in '$u'\n";
1356                        if ($pathname ne $u) {
1357                                warn "W: Refspec glob conflict ",
1358                                     "(ref: refs/remotes/$refname):\n",
1359                                     "expected path: $pathname\n",
1360                                     "    real path: $u\n",
1361                                     "Continuing ahead with $u\n";
1362                                next;
1363                        }
1364                } else {
1365                        $fetch->{$pathname} = $refname;
1366                }
1367        }
1368}
1369
1370sub parse_revision_argument {
1371        my ($base, $head) = @_;
1372        if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
1373                return ($base, $head);
1374        }
1375        return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
1376        return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
1377        return ($head, $head) if ($::_revision eq 'HEAD');
1378        return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
1379        return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
1380        die "revision argument: $::_revision not understood by git-svn\n";
1381}
1382
1383sub fetch_all {
1384        my ($repo_id, $remotes) = @_;
1385        if (ref $repo_id) {
1386                my $gs = $repo_id;
1387                $repo_id = undef;
1388                $repo_id = $gs->{repo_id};
1389        }
1390        $remotes ||= read_all_remotes();
1391        my $remote = $remotes->{$repo_id} or
1392                     die "[svn-remote \"$repo_id\"] unknown\n";
1393        my $fetch = $remote->{fetch};
1394        my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
1395        my (@gs, @globs);
1396        my $ra = Git::SVN::Ra->new($url);
1397        my $uuid = $ra->get_uuid;
1398        my $head = $ra->get_latest_revnum;
1399        my $base = defined $fetch ? $head : 0;
1400
1401        # read the max revs for wildcard expansion (branches/*, tags/*)
1402        foreach my $t (qw/branches tags/) {
1403                defined $remote->{$t} or next;
1404                push @globs, $remote->{$t};
1405                my $max_rev = eval { tmp_config(qw/--int --get/,
1406                                         "svn-remote.$repo_id.${t}-maxRev") };
1407                if (defined $max_rev && ($max_rev < $base)) {
1408                        $base = $max_rev;
1409                } elsif (!defined $max_rev) {
1410                        $base = 0;
1411                }
1412        }
1413
1414        if ($fetch) {
1415                foreach my $p (sort keys %$fetch) {
1416                        my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
1417                        my $lr = $gs->rev_map_max;
1418                        if (defined $lr) {
1419                                $base = $lr if ($lr < $base);
1420                        }
1421                        push @gs, $gs;
1422                }
1423        }
1424
1425        ($base, $head) = parse_revision_argument($base, $head);
1426        $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
1427}
1428
1429sub read_all_remotes {
1430        my $r = {};
1431        my $use_svm_props = eval { command_oneline(qw/config --bool
1432            svn.useSvmProps/) };
1433        $use_svm_props = $use_svm_props eq 'true' if $use_svm_props;
1434        foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
1435                if (m!^(.+)\.fetch=\s*(.*)\s*:\s*(.+)\s*$!) {
1436                        my ($remote, $local_ref, $_remote_ref) = ($1, $2, $3);
1437                        die("svn-remote.$remote: remote ref '$_remote_ref' "
1438                            . "must start with 'refs/remotes/'\n")
1439                                unless $_remote_ref =~ m{^refs/remotes/(.+)};
1440                        my $remote_ref = $1;
1441                        $local_ref =~ s{^/}{};
1442                        $r->{$remote}->{fetch}->{$local_ref} = $remote_ref;
1443                        $r->{$remote}->{svm} = {} if $use_svm_props;
1444                } elsif (m!^(.+)\.usesvmprops=\s*(.*)\s*$!) {
1445                        $r->{$1}->{svm} = {};
1446                } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
1447                        $r->{$1}->{url} = $2;
1448                } elsif (m!^(.+)\.(branches|tags)=
1449                           (.*):refs/remotes/(.+)\s*$/!x) {
1450                        my ($p, $g) = ($3, $4);
1451                        my $rs = $r->{$1}->{$2} = {
1452                                          t => $2,
1453                                          remote => $1,
1454                                          path => Git::SVN::GlobSpec->new($p),
1455                                          ref => Git::SVN::GlobSpec->new($g) };
1456                        if (length($rs->{ref}->{right}) != 0) {
1457                                die "The '*' glob character must be the last ",
1458                                    "character of '$g'\n";
1459                        }
1460                }
1461        }
1462
1463        map {
1464                if (defined $r->{$_}->{svm}) {
1465                        my $svm;
1466                        eval {
1467                                my $section = "svn-remote.$_";
1468                                $svm = {
1469                                        source => tmp_config('--get',
1470                                            "$section.svm-source"),
1471                                        replace => tmp_config('--get',
1472                                            "$section.svm-replace"),
1473                                }
1474                        };
1475                        $r->{$_}->{svm} = $svm;
1476                }
1477        } keys %$r;
1478
1479        $r;
1480}
1481
1482sub init_vars {
1483        $_gc_nr = $_gc_period = 1000;
1484        if (defined $_repack || defined $_repack_flags) {
1485               warn "Repack options are obsolete; they have no effect.\n";
1486        }
1487}
1488
1489sub verify_remotes_sanity {
1490        return unless -d $ENV{GIT_DIR};
1491        my %seen;
1492        foreach (command(qw/config -l/)) {
1493                if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
1494                        if ($seen{$1}) {
1495                                die "Remote ref refs/remote/$1 is tracked by",
1496                                    "\n  \"$_\"\nand\n  \"$seen{$1}\"\n",
1497                                    "Please resolve this ambiguity in ",
1498                                    "your git configuration file before ",
1499                                    "continuing\n";
1500                        }
1501                        $seen{$1} = $_;
1502                }
1503        }
1504}
1505
1506sub find_existing_remote {
1507        my ($url, $remotes) = @_;
1508        return undef if $no_reuse_existing;
1509        my $existing;
1510        foreach my $repo_id (keys %$remotes) {
1511                my $u = $remotes->{$repo_id}->{url} or next;
1512                next if $u ne $url;
1513                $existing = $repo_id;
1514                last;
1515        }
1516        $existing;
1517}
1518
1519sub init_remote_config {
1520        my ($self, $url, $no_write) = @_;
1521        $url =~ s!/+$!!; # strip trailing slash
1522        my $r = read_all_remotes();
1523        my $existing = find_existing_remote($url, $r);
1524        if ($existing) {
1525                unless ($no_write) {
1526                        print STDERR "Using existing ",
1527                                     "[svn-remote \"$existing\"]\n";
1528                }
1529                $self->{repo_id} = $existing;
1530        } elsif ($_minimize_url) {
1531                my $min_url = Git::SVN::Ra->new($url)->minimize_url;
1532                $existing = find_existing_remote($min_url, $r);
1533                if ($existing) {
1534                        unless ($no_write) {
1535                                print STDERR "Using existing ",
1536                                             "[svn-remote \"$existing\"]\n";
1537                        }
1538                        $self->{repo_id} = $existing;
1539                }
1540                if ($min_url ne $url) {
1541                        unless ($no_write) {
1542                                print STDERR "Using higher level of URL: ",
1543                                             "$url => $min_url\n";
1544                        }
1545                        my $old_path = $self->{path};
1546                        $self->{path} = $url;
1547                        $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
1548                        if (length $old_path) {
1549                                $self->{path} .= "/$old_path";
1550                        }
1551                        $url = $min_url;
1552                }
1553        }
1554        my $orig_url;
1555        if (!$existing) {
1556                # verify that we aren't overwriting anything:
1557                $orig_url = eval {
1558                        command_oneline('config', '--get',
1559                                        "svn-remote.$self->{repo_id}.url")
1560                };
1561                if ($orig_url && ($orig_url ne $url)) {
1562                        die "svn-remote.$self->{repo_id}.url already set: ",
1563                            "$orig_url\nwanted to set to: $url\n";
1564                }
1565        }
1566        my ($xrepo_id, $xpath) = find_ref($self->refname);
1567        if (defined $xpath) {
1568                die "svn-remote.$xrepo_id.fetch already set to track ",
1569                    "$xpath:refs/remotes/", $self->refname, "\n";
1570        }
1571        unless ($no_write) {
1572                command_noisy('config',
1573                              "svn-remote.$self->{repo_id}.url", $url);
1574                $self->{path} =~ s{^/}{};
1575                command_noisy('config', '--add',
1576                              "svn-remote.$self->{repo_id}.fetch",
1577                              "$self->{path}:".$self->refname);
1578        }
1579        $self->{url} = $url;
1580}
1581
1582sub find_by_url { # repos_root and, path are optional
1583        my ($class, $full_url, $repos_root, $path) = @_;
1584
1585        return undef unless defined $full_url;
1586        remove_username($full_url);
1587        remove_username($repos_root) if defined $repos_root;
1588        my $remotes = read_all_remotes();
1589        if (defined $full_url && defined $repos_root && !defined $path) {
1590                $path = $full_url;
1591                $path =~ s#^\Q$repos_root\E(?:/|$)##;
1592        }
1593        foreach my $repo_id (keys %$remotes) {
1594                my $u = $remotes->{$repo_id}->{url} or next;
1595                remove_username($u);
1596                next if defined $repos_root && $repos_root ne $u;
1597
1598                my $fetch = $remotes->{$repo_id}->{fetch} || {};
1599                foreach (qw/branches tags/) {
1600                        resolve_local_globs($u, $fetch,
1601                                            $remotes->{$repo_id}->{$_});
1602                }
1603                my $p = $path;
1604                my $rwr = rewrite_root({repo_id => $repo_id});
1605                my $svm = $remotes->{$repo_id}->{svm}
1606                        if defined $remotes->{$repo_id}->{svm};
1607                unless (defined $p) {
1608                        $p = $full_url;
1609                        my $z = $u;
1610                        my $prefix = '';
1611                        if ($rwr) {
1612                                $z = $rwr;
1613                        } elsif (defined $svm) {
1614                                $z = $svm->{source};
1615                                $prefix = $svm->{replace};
1616                                $prefix =~ s#^\Q$u\E(?:/|$)##;
1617                                $prefix =~ s#/$##;
1618                        }
1619                        $p =~ s#^\Q$z\E(?:/|$)#$prefix# or next;
1620                }
1621                foreach my $f (keys %$fetch) {
1622                        next if $f ne $p;
1623                        return Git::SVN->new($fetch->{$f}, $repo_id, $f);
1624                }
1625        }
1626        undef;
1627}
1628
1629sub init {
1630        my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
1631        my $self = _new($class, $repo_id, $ref_id, $path);
1632        if (defined $url) {
1633                $self->init_remote_config($url, $no_write);
1634        }
1635        $self;
1636}
1637
1638sub find_ref {
1639        my ($ref_id) = @_;
1640        foreach (command(qw/config -l/)) {
1641                next unless m!^svn-remote\.(.+)\.fetch=
1642                              \s*(.*)\s*:\s*refs/remotes/(.+)\s*$!x;
1643                my ($repo_id, $path, $ref) = ($1, $2, $3);
1644                if ($ref eq $ref_id) {
1645                        $path = '' if ($path =~ m#^\./?#);
1646                        return ($repo_id, $path);
1647                }
1648        }
1649        (undef, undef, undef);
1650}
1651
1652sub new {
1653        my ($class, $ref_id, $repo_id, $path) = @_;
1654        if (defined $ref_id && !defined $repo_id && !defined $path) {
1655                ($repo_id, $path) = find_ref($ref_id);
1656                if (!defined $repo_id) {
1657                        die "Could not find a \"svn-remote.*.fetch\" key ",
1658                            "in the repository configuration matching: ",
1659                            "refs/remotes/$ref_id\n";
1660                }
1661        }
1662        my $self = _new($class, $repo_id, $ref_id, $path);
1663        if (!defined $self->{path} || !length $self->{path}) {
1664                my $fetch = command_oneline('config', '--get',
1665                                            "svn-remote.$repo_id.fetch",
1666                                            ":refs/remotes/$ref_id\$") or
1667                     die "Failed to read \"svn-remote.$repo_id.fetch\" ",
1668                         "\":refs/remotes/$ref_id\$\" in config\n";
1669                ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
1670        }
1671        $self->{url} = command_oneline('config', '--get',
1672                                       "svn-remote.$repo_id.url") or
1673                  die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
1674        $self->rebuild;
1675        $self;
1676}
1677
1678sub refname {
1679        my ($refname) = "refs/remotes/$_[0]->{ref_id}" ;
1680
1681        # It cannot end with a slash /, we'll throw up on this because
1682        # SVN can't have directories with a slash in their name, either:
1683        if ($refname =~ m{/$}) {
1684                die "ref: '$refname' ends with a trailing slash, this is ",
1685                    "not permitted by git nor Subversion\n";
1686        }
1687
1688        # It cannot have ASCII control character space, tilde ~, caret ^,
1689        # colon :, question-mark ?, asterisk *, space, or open bracket [
1690        # anywhere.
1691        #
1692        # Additionally, % must be escaped because it is used for escaping
1693        # and we want our escaped refname to be reversible
1694        $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
1695
1696        # no slash-separated component can begin with a dot .
1697        # /.* becomes /%2E*
1698        $refname =~ s{/\.}{/%2E}g;
1699
1700        # It cannot have two consecutive dots .. anywhere
1701        # .. becomes %2E%2E
1702        $refname =~ s{\.\.}{%2E%2E}g;
1703
1704        return $refname;
1705}
1706
1707sub desanitize_refname {
1708        my ($refname) = @_;
1709        $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
1710        return $refname;
1711}
1712
1713sub svm_uuid {
1714        my ($self) = @_;
1715        return $self->{svm}->{uuid} if $self->svm;
1716        $self->ra;
1717        unless ($self->{svm}) {
1718                die "SVM UUID not cached, and reading remotely failed\n";
1719        }
1720        $self->{svm}->{uuid};
1721}
1722
1723sub svm {
1724        my ($self) = @_;
1725        return $self->{svm} if $self->{svm};
1726        my $svm;
1727        # see if we have it in our config, first:
1728        eval {
1729                my $section = "svn-remote.$self->{repo_id}";
1730                $svm = {
1731                  source => tmp_config('--get', "$section.svm-source"),
1732                  uuid => tmp_config('--get', "$section.svm-uuid"),
1733                  replace => tmp_config('--get', "$section.svm-replace"),
1734                }
1735        };
1736        if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
1737                $self->{svm} = $svm;
1738        }
1739        $self->{svm};
1740}
1741
1742sub _set_svm_vars {
1743        my ($self, $ra) = @_;
1744        return $ra if $self->svm;
1745
1746        my @err = ( "useSvmProps set, but failed to read SVM properties\n",
1747                    "(svm:source, svm:uuid) ",
1748                    "from the following URLs:\n" );
1749        sub read_svm_props {
1750                my ($self, $ra, $path, $r) = @_;
1751                my $props = ($ra->get_dir($path, $r))[2];
1752                my $src = $props->{'svm:source'};
1753                my $uuid = $props->{'svm:uuid'};
1754                return undef if (!$src || !$uuid);
1755
1756                chomp($src, $uuid);
1757
1758                $uuid =~ m{^[0-9a-f\-]{30,}$}
1759                    or die "doesn't look right - svm:uuid is '$uuid'\n";
1760
1761                # the '!' is used to mark the repos_root!/relative/path
1762                $src =~ s{/?!/?}{/};
1763                $src =~ s{/+$}{}; # no trailing slashes please
1764                # username is of no interest
1765                $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
1766
1767                my $replace = $ra->{url};
1768                $replace .= "/$path" if length $path;
1769
1770                my $section = "svn-remote.$self->{repo_id}";
1771                tmp_config("$section.svm-source", $src);
1772                tmp_config("$section.svm-replace", $replace);
1773                tmp_config("$section.svm-uuid", $uuid);
1774                $self->{svm} = {
1775                        source => $src,
1776                        uuid => $uuid,
1777                        replace => $replace
1778                };
1779        }
1780
1781        my $r = $ra->get_latest_revnum;
1782        my $path = $self->{path};
1783        my %tried;
1784        while (length $path) {
1785                unless ($tried{"$self->{url}/$path"}) {
1786                        return $ra if $self->read_svm_props($ra, $path, $r);
1787                        $tried{"$self->{url}/$path"} = 1;
1788                }
1789                $path =~ s#/?[^/]+$##;
1790        }
1791        die "Path: '$path' should be ''\n" if $path ne '';
1792        return $ra if $self->read_svm_props($ra, $path, $r);
1793        $tried{"$self->{url}/$path"} = 1;
1794
1795        if ($ra->{repos_root} eq $self->{url}) {
1796                die @err, (map { "  $_\n" } keys %tried), "\n";
1797        }
1798
1799        # nope, make sure we're connected to the repository root:
1800        my $ok;
1801        my @tried_b;
1802        $path = $ra->{svn_path};
1803        $ra = Git::SVN::Ra->new($ra->{repos_root});
1804        while (length $path) {
1805                unless ($tried{"$ra->{url}/$path"}) {
1806                        $ok = $self->read_svm_props($ra, $path, $r);
1807                        last if $ok;
1808                        $tried{"$ra->{url}/$path"} = 1;
1809                }
1810                $path =~ s#/?[^/]+$##;
1811        }
1812        die "Path: '$path' should be ''\n" if $path ne '';
1813        $ok ||= $self->read_svm_props($ra, $path, $r);
1814        $tried{"$ra->{url}/$path"} = 1;
1815        if (!$ok) {
1816                die @err, (map { "  $_\n" } keys %tried), "\n";
1817        }
1818        Git::SVN::Ra->new($self->{url});
1819}
1820
1821sub svnsync {
1822        my ($self) = @_;
1823        return $self->{svnsync} if $self->{svnsync};
1824
1825        if ($self->no_metadata) {
1826                die "Can't have both 'noMetadata' and ",
1827                    "'useSvnsyncProps' options set!\n";
1828        }
1829        if ($self->rewrite_root) {
1830                die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
1831                    "options set!\n";
1832        }
1833
1834        my $svnsync;
1835        # see if we have it in our config, first:
1836        eval {
1837                my $section = "svn-remote.$self->{repo_id}";
1838
1839                my $url = tmp_config('--get', "$section.svnsync-url");
1840                ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
1841                   die "doesn't look right - svn:sync-from-url is '$url'\n";
1842
1843                my $uuid = tmp_config('--get', "$section.svnsync-uuid");
1844                ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}) or
1845                   die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
1846
1847                $svnsync = { url => $url, uuid => $uuid }
1848        };
1849        if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
1850                return $self->{svnsync} = $svnsync;
1851        }
1852
1853        my $err = "useSvnsyncProps set, but failed to read " .
1854                  "svnsync property: svn:sync-from-";
1855        my $rp = $self->ra->rev_proplist(0);
1856
1857        my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
1858        ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
1859                   die "doesn't look right - svn:sync-from-url is '$url'\n";
1860
1861        my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
1862        ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}) or
1863                   die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
1864
1865        my $section = "svn-remote.$self->{repo_id}";
1866        tmp_config('--add', "$section.svnsync-uuid", $uuid);
1867        tmp_config('--add', "$section.svnsync-url", $url);
1868        return $self->{svnsync} = { url => $url, uuid => $uuid };
1869}
1870
1871# this allows us to memoize our SVN::Ra UUID locally and avoid a
1872# remote lookup (useful for 'git svn log').
1873sub ra_uuid {
1874        my ($self) = @_;
1875        unless ($self->{ra_uuid}) {
1876                my $key = "svn-remote.$self->{repo_id}.uuid";
1877                my $uuid = eval { tmp_config('--get', $key) };
1878                if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/) {
1879                        $self->{ra_uuid} = $uuid;
1880                } else {
1881                        die "ra_uuid called without URL\n" unless $self->{url};
1882                        $self->{ra_uuid} = $self->ra->get_uuid;
1883                        tmp_config('--add', $key, $self->{ra_uuid});
1884                }
1885        }
1886        $self->{ra_uuid};
1887}
1888
1889sub _set_repos_root {
1890        my ($self, $repos_root) = @_;
1891        my $k = "svn-remote.$self->{repo_id}.reposRoot";
1892        $repos_root ||= $self->ra->{repos_root};
1893        tmp_config($k, $repos_root);
1894        $repos_root;
1895}
1896
1897sub repos_root {
1898        my ($self) = @_;
1899        my $k = "svn-remote.$self->{repo_id}.reposRoot";
1900        eval { tmp_config('--get', $k) } || $self->_set_repos_root;
1901}
1902
1903sub ra {
1904        my ($self) = shift;
1905        my $ra = Git::SVN::Ra->new($self->{url});
1906        $self->_set_repos_root($ra->{repos_root});
1907        if ($self->use_svm_props && !$self->{svm}) {
1908                if ($self->no_metadata) {
1909                        die "Can't have both 'noMetadata' and ",
1910                            "'useSvmProps' options set!\n";
1911                } elsif ($self->use_svnsync_props) {
1912                        die "Can't have both 'useSvnsyncProps' and ",
1913                            "'useSvmProps' options set!\n";
1914                }
1915                $ra = $self->_set_svm_vars($ra);
1916                $self->{-want_revprops} = 1;
1917        }
1918        $ra;
1919}
1920
1921sub rel_path {
1922        my ($self) = @_;
1923        my $repos_root = $self->ra->{repos_root};
1924        return $self->{path} if ($self->{url} eq $repos_root);
1925        my $url = $self->{url} .
1926                  (length $self->{path} ? "/$self->{path}" : $self->{path});
1927        $url =~ s!^\Q$repos_root\E(?:/+|$)!!g;
1928        $url;
1929}
1930
1931# prop_walk(PATH, REV, SUB)
1932# -------------------------
1933# Recursively traverse PATH at revision REV and invoke SUB for each
1934# directory that contains a SVN property.  SUB will be invoked as
1935# follows:  &SUB(gs, path, props);  where `gs' is this instance of
1936# Git::SVN, `path' the path to the directory where the properties
1937# `props' were found.  The `path' will be relative to point of checkout,
1938# that is, if url://repo/trunk is the current Git branch, and that
1939# directory contains a sub-directory `d', SUB will be invoked with `/d/'
1940# as `path' (note the trailing `/').
1941sub prop_walk {
1942        my ($self, $path, $rev, $sub) = @_;
1943
1944        $path =~ s#^/##;
1945        my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev);
1946        $path =~ s#^/*#/#g;
1947        my $p = $path;
1948        # Strip the irrelevant part of the path.
1949        $p =~ s#^/+\Q$self->{path}\E(/|$)#/#;
1950        # Ensure the path is terminated by a `/'.
1951        $p =~ s#/*$#/#;
1952
1953        # The properties contain all the internal SVN stuff nobody
1954        # (usually) cares about.
1955        my $interesting_props = 0;
1956        foreach (keys %{$props}) {
1957                # If it doesn't start with `svn:', it must be a
1958                # user-defined property.
1959                ++$interesting_props and next if $_ !~ /^svn:/;
1960                # FIXME: Fragile, if SVN adds new public properties,
1961                # this needs to be updated.
1962                ++$interesting_props if /^svn:(?:ignore|keywords|executable
1963                                                 |eol-style|mime-type
1964                                                 |externals|needs-lock)$/x;
1965        }
1966        &$sub($self, $p, $props) if $interesting_props;
1967
1968        foreach (sort keys %$dirent) {
1969                next if $dirent->{$_}->{kind} != $SVN::Node::dir;
1970                $self->prop_walk($self->{path} . $p . $_, $rev, $sub);
1971        }
1972}
1973
1974sub last_rev { ($_[0]->last_rev_commit)[0] }
1975sub last_commit { ($_[0]->last_rev_commit)[1] }
1976
1977# returns the newest SVN revision number and newest commit SHA1
1978sub last_rev_commit {
1979        my ($self) = @_;
1980        if (defined $self->{last_rev} && defined $self->{last_commit}) {
1981                return ($self->{last_rev}, $self->{last_commit});
1982        }
1983        my $c = ::verify_ref($self->refname.'^0');
1984        if ($c && !$self->use_svm_props && !$self->no_metadata) {
1985                my $rev = (::cmt_metadata($c))[1];
1986                if (defined $rev) {
1987                        ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
1988                        return ($rev, $c);
1989                }
1990        }
1991        my $map_path = $self->map_path;
1992        unless (-e $map_path) {
1993                ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
1994                return (undef, undef);
1995        }
1996        my ($rev, $commit) = $self->rev_map_max(1);
1997        ($self->{last_rev}, $self->{last_commit}) = ($rev, $commit);
1998        return ($rev, $commit);
1999}
2000
2001sub get_fetch_range {
2002        my ($self, $min, $max) = @_;
2003        $max ||= $self->ra->get_latest_revnum;
2004        $min ||= $self->rev_map_max;
2005        (++$min, $max);
2006}
2007
2008sub tmp_config {
2009        my (@args) = @_;
2010        my $old_def_config = "$ENV{GIT_DIR}/svn/config";
2011        my $config = "$ENV{GIT_DIR}/svn/.metadata";
2012        if (! -f $config && -f $old_def_config) {
2013                rename $old_def_config, $config or
2014                       die "Failed rename $old_def_config => $config: $!\n";
2015        }
2016        my $old_config = $ENV{GIT_CONFIG};
2017        $ENV{GIT_CONFIG} = $config;
2018        $@ = undef;
2019        my @ret = eval {
2020                unless (-f $config) {
2021                        mkfile($config);
2022                        open my $fh, '>', $config or
2023                            die "Can't open $config: $!\n";
2024                        print $fh "; This file is used internally by ",
2025                                  "git-svn\n" or die
2026                                  "Couldn't write to $config: $!\n";
2027                        print $fh "; You should not have to edit it\n" or
2028                              die "Couldn't write to $config: $!\n";
2029                        close $fh or die "Couldn't close $config: $!\n";
2030                }
2031                command('config', @args);
2032        };
2033        my $err = $@;
2034        if (defined $old_config) {
2035                $ENV{GIT_CONFIG} = $old_config;
2036        } else {
2037                delete $ENV{GIT_CONFIG};
2038        }
2039        die $err if $err;
2040        wantarray ? @ret : $ret[0];
2041}
2042
2043sub tmp_index_do {
2044        my ($self, $sub) = @_;
2045        my $old_index = $ENV{GIT_INDEX_FILE};
2046        $ENV{GIT_INDEX_FILE} = $self->{index};
2047        $@ = undef;
2048        my @ret = eval {
2049                my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
2050                mkpath([$dir]) unless -d $dir;
2051                &$sub;
2052        };
2053        my $err = $@;
2054        if (defined $old_index) {
2055                $ENV{GIT_INDEX_FILE} = $old_index;
2056        } else {
2057                delete $ENV{GIT_INDEX_FILE};
2058        }
2059        die $err if $err;
2060        wantarray ? @ret : $ret[0];
2061}
2062
2063sub assert_index_clean {
2064        my ($self, $treeish) = @_;
2065
2066        $self->tmp_index_do(sub {
2067                command_noisy('read-tree', $treeish) unless -e $self->{index};
2068                my $x = command_oneline('write-tree');
2069                my ($y) = (command(qw/cat-file commit/, $treeish) =~
2070                           /^tree ($::sha1)/mo);
2071                return if $y eq $x;
2072
2073                warn "Index mismatch: $y != $x\nrereading $treeish\n";
2074                unlink $self->{index} or die "unlink $self->{index}: $!\n";
2075                command_noisy('read-tree', $treeish);
2076                $x = command_oneline('write-tree');
2077                if ($y ne $x) {
2078                        ::fatal "trees ($treeish) $y != $x\n",
2079                                "Something is seriously wrong...";
2080                }
2081        });
2082}
2083
2084sub get_commit_parents {
2085        my ($self, $log_entry) = @_;
2086        my (%seen, @ret, @tmp);
2087        # legacy support for 'set-tree'; this is only used by set_tree_cb:
2088        if (my $ip = $self->{inject_parents}) {
2089                if (my $commit = delete $ip->{$log_entry->{revision}}) {
2090                        push @tmp, $commit;
2091                }
2092        }
2093        if (my $cur = ::verify_ref($self->refname.'^0')) {
2094                push @tmp, $cur;
2095        }
2096        if (my $ipd = $self->{inject_parents_dcommit}) {
2097                if (my $commit = delete $ipd->{$log_entry->{revision}}) {
2098                        push @tmp, @$commit;
2099                }
2100        }
2101        push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
2102        while (my $p = shift @tmp) {
2103                next if $seen{$p};
2104                $seen{$p} = 1;
2105                push @ret, $p;
2106                # MAXPARENT is defined to 16 in commit-tree.c:
2107                last if @ret >= 16;
2108        }
2109        if (@tmp) {
2110                die "r$log_entry->{revision}: No room for parents:\n\t",
2111                    join("\n\t", @tmp), "\n";
2112        }
2113        @ret;
2114}
2115
2116sub rewrite_root {
2117        my ($self) = @_;
2118        return $self->{-rewrite_root} if exists $self->{-rewrite_root};
2119        my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
2120        my $rwr = eval { command_oneline(qw/config --get/, $k) };
2121        if ($rwr) {
2122                $rwr =~ s#/+$##;
2123                if ($rwr !~ m#^[a-z\+]+://#) {
2124                        die "$rwr is not a valid URL (key: $k)\n";
2125                }
2126        }
2127        $self->{-rewrite_root} = $rwr;
2128}
2129
2130sub metadata_url {
2131        my ($self) = @_;
2132        ($self->rewrite_root || $self->{url}) .
2133           (length $self->{path} ? '/' . $self->{path} : '');
2134}
2135
2136sub full_url {
2137        my ($self) = @_;
2138        $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
2139}
2140
2141
2142sub set_commit_header_env {
2143        my ($log_entry) = @_;
2144        my %env;
2145        foreach my $ned (qw/NAME EMAIL DATE/) {
2146                foreach my $ac (qw/AUTHOR COMMITTER/) {
2147                        $env{"GIT_${ac}_${ned}"} = $ENV{"GIT_${ac}_${ned}"};
2148                }
2149        }
2150
2151        $ENV{GIT_AUTHOR_NAME} = $log_entry->{name};
2152        $ENV{GIT_AUTHOR_EMAIL} = $log_entry->{email};
2153        $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
2154
2155        $ENV{GIT_COMMITTER_NAME} = (defined $log_entry->{commit_name})
2156                                                ? $log_entry->{commit_name}
2157                                                : $log_entry->{name};
2158        $ENV{GIT_COMMITTER_EMAIL} = (defined $log_entry->{commit_email})
2159                                                ? $log_entry->{commit_email}
2160                                                : $log_entry->{email};
2161        \%env;
2162}
2163
2164sub restore_commit_header_env {
2165        my ($env) = @_;
2166        foreach my $ned (qw/NAME EMAIL DATE/) {
2167                foreach my $ac (qw/AUTHOR COMMITTER/) {
2168                        my $k = "GIT_${ac}_${ned}";
2169                        if (defined $env->{$k}) {
2170                                $ENV{$k} = $env->{$k};
2171                        } else {
2172                                delete $ENV{$k};
2173                        }
2174                }
2175        }
2176}
2177
2178sub gc {
2179        command_noisy('gc', '--auto');
2180};
2181
2182sub do_git_commit {
2183        my ($self, $log_entry) = @_;
2184        my $lr = $self->last_rev;
2185        if (defined $lr && $lr >= $log_entry->{revision}) {
2186                die "Last fetched revision of ", $self->refname,
2187                    " was r$lr, but we are about to fetch: ",
2188                    "r$log_entry->{revision}!\n";
2189        }
2190        if (my $c = $self->rev_map_get($log_entry->{revision})) {
2191                croak "$log_entry->{revision} = $c already exists! ",
2192                      "Why are we refetching it?\n";
2193        }
2194        my $old_env = set_commit_header_env($log_entry);
2195        my $tree = $log_entry->{tree};
2196        if (!defined $tree) {
2197                $tree = $self->tmp_index_do(sub {
2198                                            command_oneline('write-tree') });
2199        }
2200        die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
2201
2202        my @exec = ('git-commit-tree', $tree);
2203        foreach ($self->get_commit_parents($log_entry)) {
2204                push @exec, '-p', $_;
2205        }
2206        defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
2207                                                                   or croak $!;
2208        print $msg_fh $log_entry->{log} or croak $!;
2209        restore_commit_header_env($old_env);
2210        unless ($self->no_metadata) {
2211                print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
2212                              or croak $!;
2213        }
2214        $msg_fh->flush == 0 or croak $!;
2215        close $msg_fh or croak $!;
2216        chomp(my $commit = do { local $/; <$out_fh> });
2217        close $out_fh or croak $!;
2218        waitpid $pid, 0;
2219        croak $? if $?;
2220        if ($commit !~ /^$::sha1$/o) {
2221                die "Failed to commit, invalid sha1: $commit\n";
2222        }
2223
2224        $self->rev_map_set($log_entry->{revision}, $commit, 1);
2225
2226        $self->{last_rev} = $log_entry->{revision};
2227        $self->{last_commit} = $commit;
2228        print "r$log_entry->{revision}";
2229        if (defined $log_entry->{svm_revision}) {
2230                 print " (\@$log_entry->{svm_revision})";
2231                 $self->rev_map_set($log_entry->{svm_revision}, $commit,
2232                                   0, $self->svm_uuid);
2233        }
2234        print " = $commit ($self->{ref_id})\n";
2235        if (--$_gc_nr == 0) {
2236                $_gc_nr = $_gc_period;
2237                gc();
2238        }
2239        return $commit;
2240}
2241
2242sub match_paths {
2243        my ($self, $paths, $r) = @_;
2244        return 1 if $self->{path} eq '';
2245        if (my $path = $paths->{"/$self->{path}"}) {
2246                return ($path->{action} eq 'D') ? 0 : 1;
2247        }
2248        $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
2249        if (grep /$self->{path_regex}/, keys %$paths) {
2250                return 1;
2251        }
2252        my $c = '';
2253        foreach (split m#/#, $self->{path}) {
2254                $c .= "/$_";
2255                next unless ($paths->{$c} &&
2256                             ($paths->{$c}->{action} =~ /^[AR]$/));
2257                if ($self->ra->check_path($self->{path}, $r) ==
2258                    $SVN::Node::dir) {
2259                        return 1;
2260                }
2261        }
2262        return 0;
2263}
2264
2265sub find_parent_branch {
2266        my ($self, $paths, $rev) = @_;
2267        return undef unless $self->follow_parent;
2268        unless (defined $paths) {
2269                my $err_handler = $SVN::Error::handler;
2270                $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
2271                $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1, sub {
2272                                   $paths =
2273                                      Git::SVN::Ra::dup_changed_paths($_[0]) });
2274                $SVN::Error::handler = $err_handler;
2275        }
2276        return undef unless defined $paths;
2277
2278        # look for a parent from another branch:
2279        my @b_path_components = split m#/#, $self->rel_path;
2280        my @a_path_components;
2281        my $i;
2282        while (@b_path_components) {
2283                $i = $paths->{'/'.join('/', @b_path_components)};
2284                last if $i && defined $i->{copyfrom_path};
2285                unshift(@a_path_components, pop(@b_path_components));
2286        }
2287        return undef unless defined $i && defined $i->{copyfrom_path};
2288        my $branch_from = $i->{copyfrom_path};
2289        if (@a_path_components) {
2290                print STDERR "branch_from: $branch_from => ";
2291                $branch_from .= '/'.join('/', @a_path_components);
2292                print STDERR $branch_from, "\n";
2293        }
2294        my $r = $i->{copyfrom_rev};
2295        my $repos_root = $self->ra->{repos_root};
2296        my $url = $self->ra->{url};
2297        my $new_url = $repos_root . $branch_from;
2298        print STDERR  "Found possible branch point: ",
2299                      "$new_url => ", $self->full_url, ", $r\n";
2300        $branch_from =~ s#^/##;
2301        my $gs = Git::SVN->find_by_url($new_url, $repos_root, $branch_from);
2302        unless ($gs) {
2303                my $ref_id = $self->{ref_id};
2304                $ref_id =~ s/\@\d+$//;
2305                $ref_id .= "\@$r";
2306                # just grow a tail if we're not unique enough :x
2307                $ref_id .= '-' while find_ref($ref_id);
2308                print STDERR "Initializing parent: $ref_id\n";
2309                my ($u, $p, $repo_id) = ($new_url, '', $ref_id);
2310                if ($u =~ s#^\Q$url\E(/|$)##) {
2311                        $p = $u;
2312                        $u = $url;
2313                        $repo_id = $self->{repo_id};
2314                }
2315                $gs = Git::SVN->init($u, $p, $repo_id, $ref_id, 1);
2316        }
2317        my ($r0, $parent) = $gs->find_rev_before($r, 1);
2318        if (!defined $r0 || !defined $parent) {
2319                my ($base, $head) = parse_revision_argument(0, $r);
2320                if ($base <= $r) {
2321                        $gs->fetch($base, $r);
2322                }
2323                ($r0, $parent) = $gs->last_rev_commit;
2324        }
2325        if (defined $r0 && defined $parent) {
2326                print STDERR "Found branch parent: ($self->{ref_id}) $parent\n";
2327                my $ed;
2328                if ($self->ra->can_do_switch) {
2329                        $self->assert_index_clean($parent);
2330                        print STDERR "Following parent with do_switch\n";
2331                        # do_switch works with svn/trunk >= r22312, but that
2332                        # is not included with SVN 1.4.3 (the latest version
2333                        # at the moment), so we can't rely on it
2334                        $self->{last_commit} = $parent;
2335                        $ed = SVN::Git::Fetcher->new($self);
2336                        $gs->ra->gs_do_switch($r0, $rev, $gs,
2337                                              $self->full_url, $ed)
2338                          or die "SVN connection failed somewhere...\n";
2339                } elsif ($self->ra->trees_match($new_url, $r0,
2340                                                $self->full_url, $rev)) {
2341                        print STDERR "Trees match:\n",
2342                                     "  $new_url\@$r0\n",
2343                                     "  ${\$self->full_url}\@$rev\n",
2344                                     "Following parent with no changes\n";
2345                        $self->tmp_index_do(sub {
2346                            command_noisy('read-tree', $parent);
2347                        });
2348                        $self->{last_commit} = $parent;
2349                } else {
2350                        print STDERR "Following parent with do_update\n";
2351                        $ed = SVN::Git::Fetcher->new($self);
2352                        $self->ra->gs_do_update($rev, $rev, $self, $ed)
2353                          or die "SVN connection failed somewhere...\n";
2354                }
2355                print STDERR "Successfully followed parent\n";
2356                return $self->make_log_entry($rev, [$parent], $ed);
2357        }
2358        return undef;
2359}
2360
2361sub do_fetch {
2362        my ($self, $paths, $rev) = @_;
2363        my $ed;
2364        my ($last_rev, @parents);
2365        if (my $lc = $self->last_commit) {
2366                # we can have a branch that was deleted, then re-added
2367                # under the same name but copied from another path, in
2368                # which case we'll have multiple parents (we don't
2369                # want to break the original ref, nor lose copypath info):
2370                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2371                        push @{$log_entry->{parents}}, $lc;
2372                        return $log_entry;
2373                }
2374                $ed = SVN::Git::Fetcher->new($self);
2375                $last_rev = $self->{last_rev};
2376                $ed->{c} = $lc;
2377                @parents = ($lc);
2378        } else {
2379                $last_rev = $rev;
2380                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2381                        return $log_entry;
2382                }
2383                $ed = SVN::Git::Fetcher->new($self);
2384        }
2385        unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
2386                die "SVN connection failed somewhere...\n";
2387        }
2388        $self->make_log_entry($rev, \@parents, $ed);
2389}
2390
2391sub get_untracked {
2392        my ($self, $ed) = @_;
2393        my @out;
2394        my $h = $ed->{empty};
2395        foreach (sort keys %$h) {
2396                my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
2397                push @out, "  $act: " . uri_encode($_);
2398                warn "W: $act: $_\n";
2399        }
2400        foreach my $t (qw/dir_prop file_prop/) {
2401                $h = $ed->{$t} or next;
2402                foreach my $path (sort keys %$h) {
2403                        my $ppath = $path eq '' ? '.' : $path;
2404                        foreach my $prop (sort keys %{$h->{$path}}) {
2405                                next if $SKIP_PROP{$prop};
2406                                my $v = $h->{$path}->{$prop};
2407                                my $t_ppath_prop = "$t: " .
2408                                                    uri_encode($ppath) . ' ' .
2409                                                    uri_encode($prop);
2410                                if (defined $v) {
2411                                        push @out, "  +$t_ppath_prop " .
2412                                                   uri_encode($v);
2413                                } else {
2414                                        push @out, "  -$t_ppath_prop";
2415                                }
2416                        }
2417                }
2418        }
2419        foreach my $t (qw/absent_file absent_directory/) {
2420                $h = $ed->{$t} or next;
2421                foreach my $parent (sort keys %$h) {
2422                        foreach my $path (sort @{$h->{$parent}}) {
2423                                push @out, "  $t: " .
2424                                           uri_encode("$parent/$path");
2425                                warn "W: $t: $parent/$path ",
2426                                     "Insufficient permissions?\n";
2427                        }
2428                }
2429        }
2430        \@out;
2431}
2432
2433sub parse_svn_date {
2434        my $date = shift || return '+0000 1970-01-01 00:00:00';
2435        my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2436                                            (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x) or
2437                                         croak "Unable to parse date: $date\n";
2438        "+0000 $Y-$m-$d $H:$M:$S";
2439}
2440
2441sub check_author {
2442        my ($author) = @_;
2443        if (!defined $author || length $author == 0) {
2444                $author = '(no author)';
2445        } elsif (defined $::_authors && ! defined $::users{$author}) {
2446                die "Author: $author not defined in $::_authors file\n";
2447        }
2448        $author;
2449}
2450
2451sub make_log_entry {
2452        my ($self, $rev, $parents, $ed) = @_;
2453        my $untracked = $self->get_untracked($ed);
2454
2455        open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
2456        print $un "r$rev\n" or croak $!;
2457        print $un $_, "\n" foreach @$untracked;
2458        my %log_entry = ( parents => $parents || [], revision => $rev,
2459                          log => '');
2460
2461        my $headrev;
2462        my $logged = delete $self->{logged_rev_props};
2463        if (!$logged || $self->{-want_revprops}) {
2464                my $rp = $self->ra->rev_proplist($rev);
2465                foreach (sort keys %$rp) {
2466                        my $v = $rp->{$_};
2467                        if (/^svn:(author|date|log)$/) {
2468                                $log_entry{$1} = $v;
2469                        } elsif ($_ eq 'svm:headrev') {
2470                                $headrev = $v;
2471                        } else {
2472                                print $un "  rev_prop: ", uri_encode($_), ' ',
2473                                          uri_encode($v), "\n";
2474                        }
2475                }
2476        } else {
2477                map { $log_entry{$_} = $logged->{$_} } keys %$logged;
2478        }
2479        close $un or croak $!;
2480
2481        $log_entry{date} = parse_svn_date($log_entry{date});
2482        $log_entry{log} .= "\n";
2483        my $author = $log_entry{author} = check_author($log_entry{author});
2484        my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
2485                                                       : ($author, undef);
2486
2487        my ($commit_name, $commit_email) = ($name, $email);
2488        if ($_use_log_author) {
2489                my $name_field;
2490                if ($log_entry{log} =~ /From:\s+(.*\S)\s*\n/i) {
2491                        $name_field = $1;
2492                } elsif ($log_entry{log} =~ /Signed-off-by:\s+(.*\S)\s*\n/i) {
2493                        $name_field = $1;
2494                }
2495                if (!defined $name_field) {
2496                        if (!defined $email) {
2497                                $email = $name;
2498                        }
2499                } elsif ($name_field =~ /(.*?)\s+<(.*)>/) {
2500                        ($name, $email) = ($1, $2);
2501                } elsif ($name_field =~ /(.*)@/) {
2502                        ($name, $email) = ($1, $name_field);
2503                } else {
2504                        ($name, $email) = ($name_field, $name_field);
2505                }
2506        }
2507        if (defined $headrev && $self->use_svm_props) {
2508                if ($self->rewrite_root) {
2509                        die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
2510                            "options set!\n";
2511                }
2512                my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$};
2513                # we don't want "SVM: initializing mirror for junk" ...
2514                return undef if $r == 0;
2515                my $svm = $self->svm;
2516                if ($uuid ne $svm->{uuid}) {
2517                        die "UUID mismatch on SVM path:\n",
2518                            "expected: $svm->{uuid}\n",
2519                            "     got: $uuid\n";
2520                }
2521                my $full_url = $self->full_url;
2522                $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
2523                             die "Failed to replace '$svm->{replace}' with ",
2524                                 "'$svm->{source}' in $full_url\n";
2525                # throw away username for storing in records
2526                remove_username($full_url);
2527                $log_entry{metadata} = "$full_url\@$r $uuid";
2528                $log_entry{svm_revision} = $r;
2529                $email ||= "$author\@$uuid";
2530                $commit_email ||= "$author\@$uuid";
2531        } elsif ($self->use_svnsync_props) {
2532                my $full_url = $self->svnsync->{url};
2533                $full_url .= "/$self->{path}" if length $self->{path};
2534                remove_username($full_url);
2535                my $uuid = $self->svnsync->{uuid};
2536                $log_entry{metadata} = "$full_url\@$rev $uuid";
2537                $email ||= "$author\@$uuid";
2538                $commit_email ||= "$author\@$uuid";
2539        } else {
2540                my $url = $self->metadata_url;
2541                remove_username($url);
2542                $log_entry{metadata} = "$url\@$rev " .
2543                                       $self->ra->get_uuid;
2544                $email ||= "$author\@" . $self->ra->get_uuid;
2545                $commit_email ||= "$author\@" . $self->ra->get_uuid;
2546        }
2547        $log_entry{name} = $name;
2548        $log_entry{email} = $email;
2549        $log_entry{commit_name} = $commit_name;
2550        $log_entry{commit_email} = $commit_email;
2551        \%log_entry;
2552}
2553
2554sub fetch {
2555        my ($self, $min_rev, $max_rev, @parents) = @_;
2556        my ($last_rev, $last_commit) = $self->last_rev_commit;
2557        my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
2558        $self->ra->gs_fetch_loop_common($base, $head, [$self]);
2559}
2560
2561sub set_tree_cb {
2562        my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
2563        $self->{inject_parents} = { $rev => $tree };
2564        $self->fetch(undef, undef);
2565}
2566
2567sub set_tree {
2568        my ($self, $tree) = (shift, shift);
2569        my $log_entry = ::get_commit_entry($tree);
2570        unless ($self->{last_rev}) {
2571                fatal("Must have an existing revision to commit");
2572        }
2573        my %ed_opts = ( r => $self->{last_rev},
2574                        log => $log_entry->{log},
2575                        ra => $self->ra,
2576                        tree_a => $self->{last_commit},
2577                        tree_b => $tree,
2578                        editor_cb => sub {
2579                               $self->set_tree_cb($log_entry, $tree, @_) },
2580                        svn_path => $self->{path} );
2581        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
2582                print "No changes\nr$self->{last_rev} = $tree\n";
2583        }
2584}
2585
2586sub rebuild_from_rev_db {
2587        my ($self, $path) = @_;
2588        my $r = -1;
2589        open my $fh, '<', $path or croak "open: $!";
2590        binmode $fh or croak "binmode: $!";
2591        while (<$fh>) {
2592                length($_) == 41 or croak "inconsistent size in ($_) != 41";
2593                chomp($_);
2594                ++$r;
2595                next if $_ eq ('0' x 40);
2596                $self->rev_map_set($r, $_);
2597                print "r$r = $_\n";
2598        }
2599        close $fh or croak "close: $!";
2600        unlink $path or croak "unlink: $!";
2601}
2602
2603sub rebuild {
2604        my ($self) = @_;
2605        my $map_path = $self->map_path;
2606        return if (-e $map_path && ! -z $map_path);
2607        return unless ::verify_ref($self->refname.'^0');
2608        if ($self->use_svm_props || $self->no_metadata) {
2609                my $rev_db = $self->rev_db_path;
2610                $self->rebuild_from_rev_db($rev_db);
2611                if ($self->use_svm_props) {
2612                        my $svm_rev_db = $self->rev_db_path($self->svm_uuid);
2613                        $self->rebuild_from_rev_db($svm_rev_db);
2614                }
2615                $self->unlink_rev_db_symlink;
2616                return;
2617        }
2618        print "Rebuilding $map_path ...\n";
2619        my ($log, $ctx) =
2620            command_output_pipe(qw/rev-list --pretty=raw --no-color --reverse/,
2621                                $self->refname, '--');
2622        my $metadata_url = $self->metadata_url;
2623        remove_username($metadata_url);
2624        my $svn_uuid = $self->ra_uuid;
2625        my $c;
2626        while (<$log>) {
2627                if ( m{^commit ($::sha1)$} ) {
2628                        $c = $1;
2629                        next;
2630                }
2631                next unless s{^\s*(git-svn-id:)}{$1};
2632                my ($url, $rev, $uuid) = ::extract_metadata($_);
2633                remove_username($url);
2634
2635                # ignore merges (from set-tree)
2636                next if (!defined $rev || !$uuid);
2637
2638                # if we merged or otherwise started elsewhere, this is
2639                # how we break out of it
2640                if (($uuid ne $svn_uuid) ||
2641                    ($metadata_url && $url && ($url ne $metadata_url))) {
2642                        next;
2643                }
2644
2645                $self->rev_map_set($rev, $c);
2646                print "r$rev = $c\n";
2647        }
2648        command_close_pipe($log, $ctx);
2649        print "Done rebuilding $map_path\n";
2650        my $rev_db_path = $self->rev_db_path;
2651        if (-f $self->rev_db_path) {
2652                unlink $self->rev_db_path or croak "unlink: $!";
2653        }
2654        $self->unlink_rev_db_symlink;
2655}
2656
2657# rev_map:
2658# Tie::File seems to be prone to offset errors if revisions get sparse,
2659# it's not that fast, either.  Tie::File is also not in Perl 5.6.  So
2660# one of my favorite modules is out :<  Next up would be one of the DBM
2661# modules, but I'm not sure which is most portable...
2662#
2663# This is the replacement for the rev_db format, which was too big
2664# and inefficient for large repositories with a lot of sparse history
2665# (mainly tags)
2666#
2667# The format is this:
2668#   - 24 bytes for every record,
2669#     * 4 bytes for the integer representing an SVN revision number
2670#     * 20 bytes representing the sha1 of a git commit
2671#   - No empty padding records like the old format
2672#     (except the last record, which can be overwritten)
2673#   - new records are written append-only since SVN revision numbers
2674#     increase monotonically
2675#   - lookups on SVN revision number are done via a binary search
2676#   - Piping the file to xxd -c24 is a good way of dumping it for
2677#     viewing or editing (piped back through xxd -r), should the need
2678#     ever arise.
2679#   - The last record can be padding revision with an all-zero sha1
2680#     This is used to optimize fetch performance when using multiple
2681#     "fetch" directives in .git/config
2682#
2683# These files are disposable unless noMetadata or useSvmProps is set
2684
2685sub _rev_map_set {
2686        my ($fh, $rev, $commit) = @_;
2687
2688        binmode $fh or croak "binmode: $!";
2689        my $size = (stat($fh))[7];
2690        ($size % 24) == 0 or croak "inconsistent size: $size";
2691
2692        my $wr_offset = 0;
2693        if ($size > 0) {
2694                sysseek($fh, -24, SEEK_END) or croak "seek: $!";
2695                my $read = sysread($fh, my $buf, 24) or croak "read: $!";
2696                $read == 24 or croak "read only $read bytes (!= 24)";
2697                my ($last_rev, $last_commit) = unpack(rev_map_fmt, $buf);
2698                if ($last_commit eq ('0' x40)) {
2699                        if ($size >= 48) {
2700                                sysseek($fh, -48, SEEK_END) or croak "seek: $!";
2701                                $read = sysread($fh, $buf, 24) or
2702                                    croak "read: $!";
2703                                $read == 24 or
2704                                    croak "read only $read bytes (!= 24)";
2705                                ($last_rev, $last_commit) =
2706                                    unpack(rev_map_fmt, $buf);
2707                                if ($last_commit eq ('0' x40)) {
2708                                        croak "inconsistent .rev_map\n";
2709                                }
2710                        }
2711                        if ($last_rev >= $rev) {
2712                                croak "last_rev is higher!: $last_rev >= $rev";
2713                        }
2714                        $wr_offset = -24;
2715                }
2716        }
2717        sysseek($fh, $wr_offset, SEEK_END) or croak "seek: $!";
2718        syswrite($fh, pack(rev_map_fmt, $rev, $commit), 24) == 24 or
2719          croak "write: $!";
2720}
2721
2722sub mkfile {
2723        my ($path) = @_;
2724        unless (-e $path) {
2725                my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
2726                mkpath([$dir]) unless -d $dir;
2727                open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
2728                close $fh or die "Couldn't close (create) $path: $!\n";
2729        }
2730}
2731
2732sub rev_map_set {
2733        my ($self, $rev, $commit, $update_ref, $uuid) = @_;
2734        length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
2735        my $db = $self->map_path($uuid);
2736        my $db_lock = "$db.lock";
2737        my $sig;
2738        if ($update_ref) {
2739                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
2740                            $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
2741        }
2742        mkfile($db);
2743
2744        $LOCKFILES{$db_lock} = 1;
2745        my $sync;
2746        # both of these options make our .rev_db file very, very important
2747        # and we can't afford to lose it because rebuild() won't work
2748        if ($self->use_svm_props || $self->no_metadata) {
2749                $sync = 1;
2750                copy($db, $db_lock) or die "rev_map_set(@_): ",
2751                                           "Failed to copy: ",
2752                                           "$db => $db_lock ($!)\n";
2753        } else {
2754                rename $db, $db_lock or die "rev_map_set(@_): ",
2755                                            "Failed to rename: ",
2756                                            "$db => $db_lock ($!)\n";
2757        }
2758
2759        sysopen(my $fh, $db_lock, O_RDWR | O_CREAT)
2760             or croak "Couldn't open $db_lock: $!\n";
2761        _rev_map_set($fh, $rev, $commit);
2762        if ($sync) {
2763                $fh->flush or die "Couldn't flush $db_lock: $!\n";
2764                $fh->sync or die "Couldn't sync $db_lock: $!\n";
2765        }
2766        close $fh or croak $!;
2767        if ($update_ref) {
2768                $_head = $self;
2769                command_noisy('update-ref', '-m', "r$rev",
2770                              $self->refname, $commit);
2771        }
2772        rename $db_lock, $db or die "rev_map_set(@_): ", "Failed to rename: ",
2773                                    "$db_lock => $db ($!)\n";
2774        delete $LOCKFILES{$db_lock};
2775        if ($update_ref) {
2776                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
2777                            $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
2778                kill $sig, $$ if defined $sig;
2779        }
2780}
2781
2782# If want_commit, this will return an array of (rev, commit) where
2783# commit _must_ be a valid commit in the archive.
2784# Otherwise, it'll return the max revision (whether or not the
2785# commit is valid or just a 0x40 placeholder).
2786sub rev_map_max {
2787        my ($self, $want_commit) = @_;
2788        $self->rebuild;
2789        my $map_path = $self->map_path;
2790        stat $map_path or return $want_commit ? (0, undef) : 0;
2791        sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
2792        binmode $fh or croak "binmode: $!";
2793        my $size = (stat($fh))[7];
2794        ($size % 24) == 0 or croak "inconsistent size: $size";
2795
2796        if ($size == 0) {
2797                close $fh or croak "close: $!";
2798                return $want_commit ? (0, undef) : 0;
2799        }
2800
2801        sysseek($fh, -24, SEEK_END) or croak "seek: $!";
2802        sysread($fh, my $buf, 24) == 24 or croak "read: $!";
2803        my ($r, $c) = unpack(rev_map_fmt, $buf);
2804        if ($want_commit && $c eq ('0' x40)) {
2805                if ($size < 48) {
2806                        return $want_commit ? (0, undef) : 0;
2807                }
2808                sysseek($fh, -48, SEEK_END) or croak "seek: $!";
2809                sysread($fh, $buf, 24) == 24 or croak "read: $!";
2810                ($r, $c) = unpack(rev_map_fmt, $buf);
2811                if ($c eq ('0'x40)) {
2812                        croak "Penultimate record is all-zeroes in $map_path";
2813                }
2814        }
2815        close $fh or croak "close: $!";
2816        $want_commit ? ($r, $c) : $r;
2817}
2818
2819sub rev_map_get {
2820        my ($self, $rev, $uuid) = @_;
2821        my $map_path = $self->map_path($uuid);
2822        return undef unless -e $map_path;
2823
2824        sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
2825        binmode $fh or croak "binmode: $!";
2826        my $size = (stat($fh))[7];
2827        ($size % 24) == 0 or croak "inconsistent size: $size";
2828
2829        if ($size == 0) {
2830                close $fh or croak "close: $fh";
2831                return undef;
2832        }
2833
2834        my ($l, $u) = (0, $size - 24);
2835        my ($r, $c, $buf);
2836
2837        while ($l <= $u) {
2838                my $i = int(($l/24 + $u/24) / 2) * 24;
2839                sysseek($fh, $i, SEEK_SET) or croak "seek: $!";
2840                sysread($fh, my $buf, 24) == 24 or croak "read: $!";
2841                my ($r, $c) = unpack('NH40', $buf);
2842
2843                if ($r < $rev) {
2844                        $l = $i + 24;
2845                } elsif ($r > $rev) {
2846                        $u = $i - 24;
2847                } else { # $r == $rev
2848                        close($fh) or croak "close: $!";
2849                        return $c eq ('0' x 40) ? undef : $c;
2850                }
2851        }
2852        close($fh) or croak "close: $!";
2853        undef;
2854}
2855
2856# Finds the first svn revision that exists on (if $eq_ok is true) or
2857# before $rev for the current branch.  It will not search any lower
2858# than $min_rev.  Returns the git commit hash and svn revision number
2859# if found, else (undef, undef).
2860sub find_rev_before {
2861        my ($self, $rev, $eq_ok, $min_rev) = @_;
2862        --$rev unless $eq_ok;
2863        $min_rev ||= 1;
2864        while ($rev >= $min_rev) {
2865                if (my $c = $self->rev_map_get($rev)) {
2866                        return ($rev, $c);
2867                }
2868                --$rev;
2869        }
2870        return (undef, undef);
2871}
2872
2873# Finds the first svn revision that exists on (if $eq_ok is true) or
2874# after $rev for the current branch.  It will not search any higher
2875# than $max_rev.  Returns the git commit hash and svn revision number
2876# if found, else (undef, undef).
2877sub find_rev_after {
2878        my ($self, $rev, $eq_ok, $max_rev) = @_;
2879        ++$rev unless $eq_ok;
2880        $max_rev ||= $self->rev_map_max;
2881        while ($rev <= $max_rev) {
2882                if (my $c = $self->rev_map_get($rev)) {
2883                        return ($rev, $c);
2884                }
2885                ++$rev;
2886        }
2887        return (undef, undef);
2888}
2889
2890sub _new {
2891        my ($class, $repo_id, $ref_id, $path) = @_;
2892        unless (defined $repo_id && length $repo_id) {
2893                $repo_id = $Git::SVN::default_repo_id;
2894        }
2895        unless (defined $ref_id && length $ref_id) {
2896                $_[2] = $ref_id = $Git::SVN::default_ref_id;
2897        }
2898        $_[1] = $repo_id;
2899        my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
2900        $_[3] = $path = '' unless (defined $path);
2901        mkpath(["$ENV{GIT_DIR}/svn"]);
2902        bless {
2903                ref_id => $ref_id, dir => $dir, index => "$dir/index",
2904                path => $path, config => "$ENV{GIT_DIR}/svn/config",
2905                map_root => "$dir/.rev_map", repo_id => $repo_id }, $class;
2906}
2907
2908# for read-only access of old .rev_db formats
2909sub unlink_rev_db_symlink {
2910        my ($self) = @_;
2911        my $link = $self->rev_db_path;
2912        $link =~ s/\.[\w-]+$// or croak "missing UUID at the end of $link";
2913        if (-l $link) {
2914                unlink $link or croak "unlink: $link failed!";
2915        }
2916}
2917
2918sub rev_db_path {
2919        my ($self, $uuid) = @_;
2920        my $db_path = $self->map_path($uuid);
2921        $db_path =~ s{/\.rev_map\.}{/\.rev_db\.}
2922            or croak "map_path: $db_path does not contain '/.rev_map.' !";
2923        $db_path;
2924}
2925
2926# the new replacement for .rev_db
2927sub map_path {
2928        my ($self, $uuid) = @_;
2929        $uuid ||= $self->ra_uuid;
2930        "$self->{map_root}.$uuid";
2931}
2932
2933sub uri_encode {
2934        my ($f) = @_;
2935        $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
2936        $f
2937}
2938
2939sub remove_username {
2940        $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
2941}
2942
2943package Git::SVN::Prompt;
2944use strict;
2945use warnings;
2946require SVN::Core;
2947use vars qw/$_no_auth_cache $_username/;
2948
2949sub simple {
2950        my ($cred, $realm, $default_username, $may_save, $pool) = @_;
2951        $may_save = undef if $_no_auth_cache;
2952        $default_username = $_username if defined $_username;
2953        if (defined $default_username && length $default_username) {
2954                if (defined $realm && length $realm) {
2955                        print STDERR "Authentication realm: $realm\n";
2956                        STDERR->flush;
2957                }
2958                $cred->username($default_username);
2959        } else {
2960                username($cred, $realm, $may_save, $pool);
2961        }
2962        $cred->password(_read_password("Password for '" .
2963                                       $cred->username . "': ", $realm));
2964        $cred->may_save($may_save);
2965        $SVN::_Core::SVN_NO_ERROR;
2966}
2967
2968sub ssl_server_trust {
2969        my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
2970        $may_save = undef if $_no_auth_cache;
2971        print STDERR "Error validating server certificate for '$realm':\n";
2972        {
2973                no warnings 'once';
2974                # All variables SVN::Auth::SSL::* are used only once,
2975                # so we're shutting up Perl warnings about this.
2976                if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
2977                        print STDERR " - The certificate is not issued ",
2978                            "by a trusted authority. Use the\n",
2979                            "   fingerprint to validate ",
2980                            "the certificate manually!\n";
2981                }
2982                if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
2983                        print STDERR " - The certificate hostname ",
2984                            "does not match.\n";
2985                }
2986                if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
2987                        print STDERR " - The certificate is not yet valid.\n";
2988                }
2989                if ($failures & $SVN::Auth::SSL::EXPIRED) {
2990                        print STDERR " - The certificate has expired.\n";
2991                }
2992                if ($failures & $SVN::Auth::SSL::OTHER) {
2993                        print STDERR " - The certificate has ",
2994                            "an unknown error.\n";
2995                }
2996        } # no warnings 'once'
2997        printf STDERR
2998                "Certificate information:\n".
2999                " - Hostname: %s\n".
3000                " - Valid: from %s until %s\n".
3001                " - Issuer: %s\n".
3002                " - Fingerprint: %s\n",
3003                map $cert_info->$_, qw(hostname valid_from valid_until
3004                                       issuer_dname fingerprint);
3005        my $choice;
3006prompt:
3007        print STDERR $may_save ?
3008              "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
3009              "(R)eject or accept (t)emporarily? ";
3010        STDERR->flush;
3011        $choice = lc(substr(<STDIN> || 'R', 0, 1));
3012        if ($choice =~ /^t$/i) {
3013                $cred->may_save(undef);
3014        } elsif ($choice =~ /^r$/i) {
3015                return -1;
3016        } elsif ($may_save && $choice =~ /^p$/i) {
3017                $cred->may_save($may_save);
3018        } else {
3019                goto prompt;
3020        }
3021        $cred->accepted_failures($failures);
3022        $SVN::_Core::SVN_NO_ERROR;
3023}
3024
3025sub ssl_client_cert {
3026        my ($cred, $realm, $may_save, $pool) = @_;
3027        $may_save = undef if $_no_auth_cache;
3028        print STDERR "Client certificate filename: ";
3029        STDERR->flush;
3030        chomp(my $filename = <STDIN>);
3031        $cred->cert_file($filename);
3032        $cred->may_save($may_save);
3033        $SVN::_Core::SVN_NO_ERROR;
3034}
3035
3036sub ssl_client_cert_pw {
3037        my ($cred, $realm, $may_save, $pool) = @_;
3038        $may_save = undef if $_no_auth_cache;
3039        $cred->password(_read_password("Password: ", $realm));
3040        $cred->may_save($may_save);
3041        $SVN::_Core::SVN_NO_ERROR;
3042}
3043
3044sub username {
3045        my ($cred, $realm, $may_save, $pool) = @_;
3046        $may_save = undef if $_no_auth_cache;
3047        if (defined $realm && length $realm) {
3048                print STDERR "Authentication realm: $realm\n";
3049        }
3050        my $username;
3051        if (defined $_username) {
3052                $username = $_username;
3053        } else {
3054                print STDERR "Username: ";
3055                STDERR->flush;
3056                chomp($username = <STDIN>);
3057        }
3058        $cred->username($username);
3059        $cred->may_save($may_save);
3060        $SVN::_Core::SVN_NO_ERROR;
3061}
3062
3063sub _read_password {
3064        my ($prompt, $realm) = @_;
3065        print STDERR $prompt;
3066        STDERR->flush;
3067        require Term::ReadKey;
3068        Term::ReadKey::ReadMode('noecho');
3069        my $password = '';
3070        while (defined(my $key = Term::ReadKey::ReadKey(0))) {
3071                last if $key =~ /[\012\015]/; # \n\r
3072                $password .= $key;
3073        }
3074        Term::ReadKey::ReadMode('restore');
3075        print STDERR "\n";
3076        STDERR->flush;
3077        $password;
3078}
3079
3080package SVN::Git::Fetcher;
3081use vars qw/@ISA/;
3082use strict;
3083use warnings;
3084use Carp qw/croak/;
3085use File::Temp qw/tempfile/;
3086use IO::File qw//;
3087
3088# file baton members: path, mode_a, mode_b, pool, fh, blob, base
3089sub new {
3090        my ($class, $git_svn) = @_;
3091        my $self = SVN::Delta::Editor->new;
3092        bless $self, $class;
3093        $self->{c} = $git_svn->{last_commit} if exists $git_svn->{last_commit};
3094        $self->{empty} = {};
3095        $self->{dir_prop} = {};
3096        $self->{file_prop} = {};
3097        $self->{absent_dir} = {};
3098        $self->{absent_file} = {};
3099        $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
3100        $self;
3101}
3102
3103sub set_path_strip {
3104        my ($self, $path) = @_;
3105        $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
3106}
3107
3108sub open_root {
3109        { path => '' };
3110}
3111
3112sub open_directory {
3113        my ($self, $path, $pb, $rev) = @_;
3114        { path => $path };
3115}
3116
3117sub git_path {
3118        my ($self, $path) = @_;
3119        if ($self->{path_strip}) {
3120                $path =~ s!$self->{path_strip}!! or
3121                  die "Failed to strip path '$path' ($self->{path_strip})\n";
3122        }
3123        $path;
3124}
3125
3126sub delete_entry {
3127        my ($self, $path, $rev, $pb) = @_;
3128
3129        my $gpath = $self->git_path($path);
3130        return undef if ($gpath eq '');
3131
3132        # remove entire directories.
3133        if (command('ls-tree', $self->{c}, '--', $gpath) =~ /^040000 tree/) {
3134                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3135                                                     -r --name-only -z/,
3136                                                     $self->{c}, '--', $gpath);
3137                local $/ = "\0";
3138                while (<$ls>) {
3139                        chomp;
3140                        $self->{gii}->remove($_);
3141                        print "\tD\t$_\n" unless $::_q;
3142                }
3143                print "\tD\t$gpath/\n" unless $::_q;
3144                command_close_pipe($ls, $ctx);
3145                $self->{empty}->{$path} = 0
3146        } else {
3147                $self->{gii}->remove($gpath);
3148                print "\tD\t$gpath\n" unless $::_q;
3149        }
3150        undef;
3151}
3152
3153sub open_file {
3154        my ($self, $path, $pb, $rev) = @_;
3155        my $gpath = $self->git_path($path);
3156        my ($mode, $blob) = (command('ls-tree', $self->{c}, '--', $gpath)
3157                             =~ /^(\d{6}) blob ([a-f\d]{40})\t/);
3158        unless (defined $mode && defined $blob) {
3159                die "$path was not found in commit $self->{c} (r$rev)\n";
3160        }
3161        { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
3162          pool => SVN::Pool->new, action => 'M' };
3163}
3164
3165sub add_file {
3166        my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
3167        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3168        delete $self->{empty}->{$dir};
3169        { path => $path, mode_a => 100644, mode_b => 100644,
3170          pool => SVN::Pool->new, action => 'A' };
3171}
3172
3173sub add_directory {
3174        my ($self, $path, $cp_path, $cp_rev) = @_;
3175        my $gpath = $self->git_path($path);
3176        if ($gpath eq '') {
3177                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3178                                                     -r --name-only -z/,
3179                                                     $self->{c});
3180                local $/ = "\0";
3181                while (<$ls>) {
3182                        chomp;
3183                        $self->{gii}->remove($_);
3184                        print "\tD\t$_\n" unless $::_q;
3185                }
3186                command_close_pipe($ls, $ctx);
3187                $self->{empty}->{$path} = 0;
3188        }
3189        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3190        delete $self->{empty}->{$dir};
3191        $self->{empty}->{$path} = 1;
3192        { path => $path };
3193}
3194
3195sub change_dir_prop {
3196        my ($self, $db, $prop, $value) = @_;
3197        $self->{dir_prop}->{$db->{path}} ||= {};
3198        $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
3199        undef;
3200}
3201
3202sub absent_directory {
3203        my ($self, $path, $pb) = @_;
3204        $self->{absent_dir}->{$pb->{path}} ||= [];
3205        push @{$self->{absent_dir}->{$pb->{path}}}, $path;
3206        undef;
3207}
3208
3209sub absent_file {
3210        my ($self, $path, $pb) = @_;
3211        $self->{absent_file}->{$pb->{path}} ||= [];
3212        push @{$self->{absent_file}->{$pb->{path}}}, $path;
3213        undef;
3214}
3215
3216sub change_file_prop {
3217        my ($self, $fb, $prop, $value) = @_;
3218        if ($prop eq 'svn:executable') {
3219                if ($fb->{mode_b} != 120000) {
3220                        $fb->{mode_b} = defined $value ? 100755 : 100644;
3221                }
3222        } elsif ($prop eq 'svn:special') {
3223                $fb->{mode_b} = defined $value ? 120000 : 100644;
3224        } else {
3225                $self->{file_prop}->{$fb->{path}} ||= {};
3226                $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
3227        }
3228        undef;
3229}
3230
3231sub apply_textdelta {
3232        my ($self, $fb, $exp) = @_;
3233        my $fh = IO::File->new_tmpfile;
3234        $fh->autoflush(1);
3235        # $fh gets auto-closed() by SVN::TxDelta::apply(),
3236        # (but $base does not,) so dup() it for reading in close_file
3237        open my $dup, '<&', $fh or croak $!;
3238        my $base = IO::File->new_tmpfile;
3239        $base->autoflush(1);
3240        if ($fb->{blob}) {
3241                print $base 'link ' if ($fb->{mode_a} == 120000);
3242                my $size = $::_repository->cat_blob($fb->{blob}, $base);
3243                die "Failed to read object $fb->{blob}" if ($size < 0);
3244
3245                if (defined $exp) {
3246                        seek $base, 0, 0 or croak $!;
3247                        my $got = ::md5sum($base);
3248                        die "Checksum mismatch: $fb->{path} $fb->{blob}\n",
3249                            "expected: $exp\n",
3250                            "     got: $got\n" if ($got ne $exp);
3251                }
3252        }
3253        seek $base, 0, 0 or croak $!;
3254        $fb->{fh} = $dup;
3255        $fb->{base} = $base;
3256        [ SVN::TxDelta::apply($base, $fh, undef, $fb->{path}, $fb->{pool}) ];
3257}
3258
3259sub close_file {
3260        my ($self, $fb, $exp) = @_;
3261        my $hash;
3262        my $path = $self->git_path($fb->{path});
3263        if (my $fh = $fb->{fh}) {
3264                if (defined $exp) {
3265                        seek($fh, 0, 0) or croak $!;
3266                        my $got = ::md5sum($fh);
3267                        if ($got ne $exp) {
3268                                die "Checksum mismatch: $path\n",
3269                                    "expected: $exp\n    got: $got\n";
3270                        }
3271                }
3272                sysseek($fh, 0, 0) or croak $!;
3273                if ($fb->{mode_b} == 120000) {
3274                        eval {
3275                                sysread($fh, my $buf, 5) == 5 or croak $!;
3276                                $buf eq 'link ' or die "$path has mode 120000",
3277                                                       " but is not a link";
3278                        };
3279                        if ($@) {
3280                                warn "$@\n";
3281                                sysseek($fh, 0, 0) or croak $!;
3282                        }
3283                }
3284
3285                my ($tmp_fh, $tmp_filename) = File::Temp::tempfile(UNLINK => 1);
3286                my $result;
3287                while ($result = sysread($fh, my $string, 1024)) {
3288                        my $wrote = syswrite($tmp_fh, $string, $result);
3289                        defined($wrote) && $wrote == $result
3290                                or croak("write $tmp_filename: $!\n");
3291                }
3292                defined $result or croak $!;
3293                close $tmp_fh or croak $!;
3294
3295                close $fh or croak $!;
3296
3297                $hash = $::_repository->hash_and_insert_object($tmp_filename);
3298                unlink($tmp_filename);
3299                $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
3300                close $fb->{base} or croak $!;
3301        } else {
3302                $hash = $fb->{blob} or die "no blob information\n";
3303        }
3304        $fb->{pool}->clear;
3305        $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
3306        print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
3307        undef;
3308}
3309
3310sub abort_edit {
3311        my $self = shift;
3312        $self->{nr} = $self->{gii}->{nr};
3313        delete $self->{gii};
3314        $self->SUPER::abort_edit(@_);
3315}
3316
3317sub close_edit {
3318        my $self = shift;
3319        $self->{git_commit_ok} = 1;
3320        $self->{nr} = $self->{gii}->{nr};
3321        delete $self->{gii};
3322        $self->SUPER::close_edit(@_);
3323}
3324
3325package SVN::Git::Editor;
3326use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
3327use strict;
3328use warnings;
3329use Carp qw/croak/;
3330use IO::File;
3331
3332sub new {
3333        my ($class, $opts) = @_;
3334        foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
3335                die "$_ required!\n" unless (defined $opts->{$_});
3336        }
3337
3338        my $pool = SVN::Pool->new;
3339        my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
3340        my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
3341                                     $opts->{r}, $mods);
3342
3343        # $opts->{ra} functions should not be used after this:
3344        my @ce  = $opts->{ra}->get_commit_editor($opts->{log},
3345                                                $opts->{editor_cb}, $pool);
3346        my $self = SVN::Delta::Editor->new(@ce, $pool);
3347        bless $self, $class;
3348        foreach (qw/svn_path r tree_a tree_b/) {
3349                $self->{$_} = $opts->{$_};
3350        }
3351        $self->{url} = $opts->{ra}->{url};
3352        $self->{mods} = $mods;
3353        $self->{types} = $types;
3354        $self->{pool} = $pool;
3355        $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
3356        $self->{rm} = { };
3357        $self->{path_prefix} = length $self->{svn_path} ?
3358                               "$self->{svn_path}/" : '';
3359        $self->{config} = $opts->{config};
3360        return $self;
3361}
3362
3363sub generate_diff {
3364        my ($tree_a, $tree_b) = @_;
3365        my @diff_tree = qw(diff-tree -z -r);
3366        if ($_cp_similarity) {
3367                push @diff_tree, "-C$_cp_similarity";
3368        } else {
3369                push @diff_tree, '-C';
3370        }
3371        push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
3372        push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
3373        push @diff_tree, $tree_a, $tree_b;
3374        my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
3375        local $/ = "\0";
3376        my $state = 'meta';
3377        my @mods;
3378        while (<$diff_fh>) {
3379                chomp $_; # this gets rid of the trailing "\0"
3380                if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
3381                                        $::sha1\s($::sha1)\s
3382                                        ([MTCRAD])\d*$/xo) {
3383                        push @mods, {   mode_a => $1, mode_b => $2,
3384                                        sha1_b => $3, chg => $4 };
3385                        if ($4 =~ /^(?:C|R)$/) {
3386                                $state = 'file_a';
3387                        } else {
3388                                $state = 'file_b';
3389                        }
3390                } elsif ($state eq 'file_a') {
3391                        my $x = $mods[$#mods] or croak "Empty array\n";
3392                        if ($x->{chg} !~ /^(?:C|R)$/) {
3393                                croak "Error parsing $_, $x->{chg}\n";
3394                        }
3395                        $x->{file_a} = $_;
3396                        $state = 'file_b';
3397                } elsif ($state eq 'file_b') {
3398                        my $x = $mods[$#mods] or croak "Empty array\n";
3399                        if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
3400                                croak "Error parsing $_, $x->{chg}\n";
3401                        }
3402                        if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
3403                                croak "Error parsing $_, $x->{chg}\n";
3404                        }
3405                        $x->{file_b} = $_;
3406                        $state = 'meta';
3407                } else {
3408                        croak "Error parsing $_\n";
3409                }
3410        }
3411        command_close_pipe($diff_fh, $ctx);
3412        \@mods;
3413}
3414
3415sub check_diff_paths {
3416        my ($ra, $pfx, $rev, $mods) = @_;
3417        my %types;
3418        $pfx .= '/' if length $pfx;
3419
3420        sub type_diff_paths {
3421                my ($ra, $types, $path, $rev) = @_;
3422                my @p = split m#/+#, $path;
3423                my $c = shift @p;
3424                unless (defined $types->{$c}) {
3425                        $types->{$c} = $ra->check_path($c, $rev);
3426                }
3427                while (@p) {
3428                        $c .= '/' . shift @p;
3429                        next if defined $types->{$c};
3430                        $types->{$c} = $ra->check_path($c, $rev);
3431                }
3432        }
3433
3434        foreach my $m (@$mods) {
3435                foreach my $f (qw/file_a file_b/) {
3436                        next unless defined $m->{$f};
3437                        my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
3438                        if (length $pfx.$dir && ! defined $types{$dir}) {
3439                                type_diff_paths($ra, \%types, $pfx.$dir, $rev);
3440                        }
3441                }
3442        }
3443        \%types;
3444}
3445
3446sub split_path {
3447        return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
3448}
3449
3450sub repo_path {
3451        my ($self, $path) = @_;
3452        $self->{path_prefix}.(defined $path ? $path : '');
3453}
3454
3455sub url_path {
3456        my ($self, $path) = @_;
3457        if ($self->{url} =~ m#^https?://#) {
3458                $path =~ s/([^a-zA-Z0-9_.-])/uc sprintf("%%%02x",ord($1))/eg;
3459        }
3460        $self->{url} . '/' . $self->repo_path($path);
3461}
3462
3463sub rmdirs {
3464        my ($self) = @_;
3465        my $rm = $self->{rm};
3466        delete $rm->{''}; # we never delete the url we're tracking
3467        return unless %$rm;
3468
3469        foreach (keys %$rm) {
3470                my @d = split m#/#, $_;
3471                my $c = shift @d;
3472                $rm->{$c} = 1;
3473                while (@d) {
3474                        $c .= '/' . shift @d;
3475                        $rm->{$c} = 1;
3476                }
3477        }
3478        delete $rm->{$self->{svn_path}};
3479        delete $rm->{''}; # we never delete the url we're tracking
3480        return unless %$rm;
3481
3482        my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
3483                                             $self->{tree_b});
3484        local $/ = "\0";
3485        while (<$fh>) {
3486                chomp;
3487                my @dn = split m#/#, $_;
3488                while (pop @dn) {
3489                        delete $rm->{join '/', @dn};
3490                }
3491                unless (%$rm) {
3492                        close $fh;
3493                        return;
3494                }
3495        }
3496        command_close_pipe($fh, $ctx);
3497
3498        my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
3499        foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
3500                $self->close_directory($bat->{$d}, $p);
3501                my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
3502                print "\tD+\t$d/\n" unless $::_q;
3503                $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
3504                delete $bat->{$d};
3505        }
3506}
3507
3508sub open_or_add_dir {
3509        my ($self, $full_path, $baton) = @_;
3510        my $t = $self->{types}->{$full_path};
3511        if (!defined $t) {
3512                die "$full_path not known in r$self->{r} or we have a bug!\n";
3513        }
3514        {
3515                no warnings 'once';
3516                # SVN::Node::none and SVN::Node::file are used only once,
3517                # so we're shutting up Perl's warnings about them.
3518                if ($t == $SVN::Node::none) {
3519                        return $self->add_directory($full_path, $baton,
3520                            undef, -1, $self->{pool});
3521                } elsif ($t == $SVN::Node::dir) {
3522                        return $self->open_directory($full_path, $baton,
3523                            $self->{r}, $self->{pool});
3524                } # no warnings 'once'
3525                print STDERR "$full_path already exists in repository at ",
3526                    "r$self->{r} and it is not a directory (",
3527                    ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
3528        } # no warnings 'once'
3529        exit 1;
3530}
3531
3532sub ensure_path {
3533        my ($self, $path) = @_;
3534        my $bat = $self->{bat};
3535        my $repo_path = $self->repo_path($path);
3536        return $bat->{''} unless (length $repo_path);
3537        my @p = split m#/+#, $repo_path;
3538        my $c = shift @p;
3539        $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
3540        while (@p) {
3541                my $c0 = $c;
3542                $c .= '/' . shift @p;
3543                $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
3544        }
3545        return $bat->{$c};
3546}
3547
3548# Subroutine to convert a globbing pattern to a regular expression.
3549# From perl cookbook.
3550sub glob2pat {
3551        my $globstr = shift;
3552        my %patmap = ('*' => '.*', '?' => '.', '[' => '[', ']' => ']');
3553        $globstr =~ s{(.)} { $patmap{$1} || "\Q$1" }ge;
3554        return '^' . $globstr . '$';
3555}
3556
3557sub check_autoprop {
3558        my ($self, $pattern, $properties, $file, $fbat) = @_;
3559        # Convert the globbing pattern to a regular expression.
3560        my $regex = glob2pat($pattern);
3561        # Check if the pattern matches the file name.
3562        if($file =~ m/($regex)/) {
3563                # Parse the list of properties to set.
3564                my @props = split(/;/, $properties);
3565                foreach my $prop (@props) {
3566                        # Parse 'name=value' syntax and set the property.
3567                        if ($prop =~ /([^=]+)=(.*)/) {
3568                                my ($n,$v) = ($1,$2);
3569                                for ($n, $v) {
3570                                        s/^\s+//; s/\s+$//;
3571                                }
3572                                $self->change_file_prop($fbat, $n, $v);
3573                        }
3574                }
3575        }
3576}
3577
3578sub apply_autoprops {
3579        my ($self, $file, $fbat) = @_;
3580        my $conf_t = ${$self->{config}}{'config'};
3581        no warnings 'once';
3582        # Check [miscellany]/enable-auto-props in svn configuration.
3583        if (SVN::_Core::svn_config_get_bool(
3584                $conf_t,
3585                $SVN::_Core::SVN_CONFIG_SECTION_MISCELLANY,
3586                $SVN::_Core::SVN_CONFIG_OPTION_ENABLE_AUTO_PROPS,
3587                0)) {
3588                # Auto-props are enabled.  Enumerate them to look for matches.
3589                my $callback = sub {
3590                        $self->check_autoprop($_[0], $_[1], $file, $fbat);
3591                };
3592                SVN::_Core::svn_config_enumerate(
3593                        $conf_t,
3594                        $SVN::_Core::SVN_CONFIG_SECTION_AUTO_PROPS,
3595                        $callback);
3596        }
3597}
3598
3599sub A {
3600        my ($self, $m) = @_;
3601        my ($dir, $file) = split_path($m->{file_b});
3602        my $pbat = $self->ensure_path($dir);
3603        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3604                                        undef, -1);
3605        print "\tA\t$m->{file_b}\n" unless $::_q;
3606        $self->apply_autoprops($file, $fbat);
3607        $self->chg_file($fbat, $m);
3608        $self->close_file($fbat,undef,$self->{pool});
3609}
3610
3611sub C {
3612        my ($self, $m) = @_;
3613        my ($dir, $file) = split_path($m->{file_b});
3614        my $pbat = $self->ensure_path($dir);
3615        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3616                                $self->url_path($m->{file_a}), $self->{r});
3617        print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
3618        $self->chg_file($fbat, $m);
3619        $self->close_file($fbat,undef,$self->{pool});
3620}
3621
3622sub delete_entry {
3623        my ($self, $path, $pbat) = @_;
3624        my $rpath = $self->repo_path($path);
3625        my ($dir, $file) = split_path($rpath);
3626        $self->{rm}->{$dir} = 1;
3627        $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
3628}
3629
3630sub R {
3631        my ($self, $m) = @_;
3632        my ($dir, $file) = split_path($m->{file_b});
3633        my $pbat = $self->ensure_path($dir);
3634        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3635                                $self->url_path($m->{file_a}), $self->{r});
3636        print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
3637        $self->chg_file($fbat, $m);
3638        $self->close_file($fbat,undef,$self->{pool});
3639
3640        ($dir, $file) = split_path($m->{file_a});
3641        $pbat = $self->ensure_path($dir);
3642        $self->delete_entry($m->{file_a}, $pbat);
3643}
3644
3645sub M {
3646        my ($self, $m) = @_;
3647        my ($dir, $file) = split_path($m->{file_b});
3648        my $pbat = $self->ensure_path($dir);
3649        my $fbat = $self->open_file($self->repo_path($m->{file_b}),
3650                                $pbat,$self->{r},$self->{pool});
3651        print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
3652        $self->chg_file($fbat, $m);
3653        $self->close_file($fbat,undef,$self->{pool});
3654}
3655
3656sub T { shift->M(@_) }
3657
3658sub change_file_prop {
3659        my ($self, $fbat, $pname, $pval) = @_;
3660        $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
3661}
3662
3663sub chg_file {
3664        my ($self, $fbat, $m) = @_;
3665        if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
3666                $self->change_file_prop($fbat,'svn:executable','*');
3667        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
3668                $self->change_file_prop($fbat,'svn:executable',undef);
3669        }
3670        my $fh = IO::File->new_tmpfile or croak $!;
3671        if ($m->{mode_b} =~ /^120/) {
3672                print $fh 'link ' or croak $!;
3673                $self->change_file_prop($fbat,'svn:special','*');
3674        } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
3675                $self->change_file_prop($fbat,'svn:special',undef);
3676        }
3677        my $size = $::_repository->cat_blob($m->{sha1_b}, $fh);
3678        croak "Failed to read object $m->{sha1_b}" if ($size < 0);
3679        $fh->flush == 0 or croak $!;
3680        seek $fh, 0, 0 or croak $!;
3681
3682        my $exp = ::md5sum($fh);
3683        seek $fh, 0, 0 or croak $!;
3684
3685        my $pool = SVN::Pool->new;
3686        my $atd = $self->apply_textdelta($fbat, undef, $pool);
3687        my $got = SVN::TxDelta::send_stream($fh, @$atd, $pool);
3688        die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
3689        $pool->clear;
3690
3691        close $fh or croak $!;
3692}
3693
3694sub D {
3695        my ($self, $m) = @_;
3696        my ($dir, $file) = split_path($m->{file_b});
3697        my $pbat = $self->ensure_path($dir);
3698        print "\tD\t$m->{file_b}\n" unless $::_q;
3699        $self->delete_entry($m->{file_b}, $pbat);
3700}
3701
3702sub close_edit {
3703        my ($self) = @_;
3704        my ($p,$bat) = ($self->{pool}, $self->{bat});
3705        foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
3706                next if $_ eq '';
3707                $self->close_directory($bat->{$_}, $p);
3708        }
3709        $self->close_directory($bat->{''}, $p);
3710        $self->SUPER::close_edit($p);
3711        $p->clear;
3712}
3713
3714sub abort_edit {
3715        my ($self) = @_;
3716        $self->SUPER::abort_edit($self->{pool});
3717}
3718
3719sub DESTROY {
3720        my $self = shift;
3721        $self->SUPER::DESTROY(@_);
3722        $self->{pool}->clear;
3723}
3724
3725# this drives the editor
3726sub apply_diff {
3727        my ($self) = @_;
3728        my $mods = $self->{mods};
3729        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
3730        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
3731                my $f = $m->{chg};
3732                if (defined $o{$f}) {
3733                        $self->$f($m);
3734                } else {
3735                        fatal("Invalid change type: $f");
3736                }
3737        }
3738        $self->rmdirs if $_rmdir;
3739        if (@$mods == 0) {
3740                $self->abort_edit;
3741        } else {
3742                $self->close_edit;
3743        }
3744        return scalar @$mods;
3745}
3746
3747package Git::SVN::Ra;
3748use vars qw/@ISA $config_dir $_log_window_size/;
3749use strict;
3750use warnings;
3751my ($ra_invalid, $can_do_switch, %ignored_err, $RA);
3752
3753BEGIN {
3754        # enforce temporary pool usage for some simple functions
3755        no strict 'refs';
3756        for my $f (qw/rev_proplist get_latest_revnum get_uuid get_repos_root/) {
3757                my $SUPER = "SUPER::$f";
3758                *$f = sub {
3759                        my $self = shift;
3760                        my $pool = SVN::Pool->new;
3761                        my @ret = $self->$SUPER(@_,$pool);
3762                        $pool->clear;
3763                        wantarray ? @ret : $ret[0];
3764                };
3765        }
3766}
3767
3768sub _auth_providers () {
3769        [
3770          SVN::Client::get_simple_provider(),
3771          SVN::Client::get_ssl_server_trust_file_provider(),
3772          SVN::Client::get_simple_prompt_provider(
3773            \&Git::SVN::Prompt::simple, 2),
3774          SVN::Client::get_ssl_client_cert_file_provider(),
3775          SVN::Client::get_ssl_client_cert_prompt_provider(
3776            \&Git::SVN::Prompt::ssl_client_cert, 2),
3777          SVN::Client::get_ssl_client_cert_pw_file_provider(),
3778          SVN::Client::get_ssl_client_cert_pw_prompt_provider(
3779            \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
3780          SVN::Client::get_username_provider(),
3781          SVN::Client::get_ssl_server_trust_prompt_provider(
3782            \&Git::SVN::Prompt::ssl_server_trust),
3783          SVN::Client::get_username_prompt_provider(
3784            \&Git::SVN::Prompt::username, 2)
3785        ]
3786}
3787
3788sub escape_uri_only {
3789        my ($uri) = @_;
3790        my @tmp;
3791        foreach (split m{/}, $uri) {
3792                s/([^\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
3793                push @tmp, $_;
3794        }
3795        join('/', @tmp);
3796}
3797
3798sub escape_url {
3799        my ($url) = @_;
3800        if ($url =~ m#^(https?)://([^/]+)(.*)$#) {
3801                my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
3802                $url = "$scheme://$domain$uri";
3803        }
3804        $url;
3805}
3806
3807sub new {
3808        my ($class, $url) = @_;
3809        $url =~ s!/+$!!;
3810        return $RA if ($RA && $RA->{url} eq $url);
3811
3812        SVN::_Core::svn_config_ensure($config_dir, undef);
3813        my ($baton, $callbacks) = SVN::Core::auth_open_helper(_auth_providers);
3814        my $config = SVN::Core::config_get_config($config_dir);
3815        $RA = undef;
3816        my $dont_store_passwords = 1;
3817        my $conf_t = ${$config}{'config'};
3818        {
3819                no warnings 'once';
3820                # The usage of $SVN::_Core::SVN_CONFIG_* variables
3821                # produces warnings that variables are used only once.
3822                # I had not found the better way to shut them up, so
3823                # the warnings of type 'once' are disabled in this block.
3824                if (SVN::_Core::svn_config_get_bool($conf_t,
3825                    $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
3826                    $SVN::_Core::SVN_CONFIG_OPTION_STORE_PASSWORDS,
3827                    1) == 0) {
3828                        SVN::_Core::svn_auth_set_parameter($baton,
3829                            $SVN::_Core::SVN_AUTH_PARAM_DONT_STORE_PASSWORDS,
3830                            bless (\$dont_store_passwords, "_p_void"));
3831                }
3832                if (SVN::_Core::svn_config_get_bool($conf_t,
3833                    $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
3834                    $SVN::_Core::SVN_CONFIG_OPTION_STORE_AUTH_CREDS,
3835                    1) == 0) {
3836                        $Git::SVN::Prompt::_no_auth_cache = 1;
3837                }
3838        } # no warnings 'once'
3839        my $self = SVN::Ra->new(url => escape_url($url), auth => $baton,
3840                              config => $config,
3841                              pool => SVN::Pool->new,
3842                              auth_provider_callbacks => $callbacks);
3843        $self->{url} = $url;
3844        $self->{svn_path} = $url;
3845        $self->{repos_root} = $self->get_repos_root;
3846        $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
3847        $self->{cache} = { check_path => { r => 0, data => {} },
3848                           get_dir => { r => 0, data => {} } };
3849        $RA = bless $self, $class;
3850}
3851
3852sub check_path {
3853        my ($self, $path, $r) = @_;
3854        my $cache = $self->{cache}->{check_path};
3855        if ($r == $cache->{r} && exists $cache->{data}->{$path}) {
3856                return $cache->{data}->{$path};
3857        }
3858        my $pool = SVN::Pool->new;
3859        my $t = $self->SUPER::check_path($path, $r, $pool);
3860        $pool->clear;
3861        if ($r != $cache->{r}) {
3862                %{$cache->{data}} = ();
3863                $cache->{r} = $r;
3864        }
3865        $cache->{data}->{$path} = $t;
3866}
3867
3868sub get_dir {
3869        my ($self, $dir, $r) = @_;
3870        my $cache = $self->{cache}->{get_dir};
3871        if ($r == $cache->{r}) {
3872                if (my $x = $cache->{data}->{$dir}) {
3873                        return wantarray ? @$x : $x->[0];
3874                }
3875        }
3876        my $pool = SVN::Pool->new;
3877        my ($d, undef, $props) = $self->SUPER::get_dir($dir, $r, $pool);
3878        my %dirents = map { $_ => { kind => $d->{$_}->kind } } keys %$d;
3879        $pool->clear;
3880        if ($r != $cache->{r}) {
3881                %{$cache->{data}} = ();
3882                $cache->{r} = $r;
3883        }
3884        $cache->{data}->{$dir} = [ \%dirents, $r, $props ];
3885        wantarray ? (\%dirents, $r, $props) : \%dirents;
3886}
3887
3888sub DESTROY {
3889        # do not call the real DESTROY since we store ourselves in $RA
3890}
3891
3892sub get_log {
3893        my ($self, @args) = @_;
3894        my $pool = SVN::Pool->new;
3895        splice(@args, 3, 1) if ($SVN::Core::VERSION le '1.2.0');
3896        my $ret = $self->SUPER::get_log(@args, $pool);
3897        $pool->clear;
3898        $ret;
3899}
3900
3901sub trees_match {
3902        my ($self, $url1, $rev1, $url2, $rev2) = @_;
3903        my $ctx = SVN::Client->new(auth => _auth_providers);
3904        my $out = IO::File->new_tmpfile;
3905
3906        # older SVN (1.1.x) doesn't take $pool as the last parameter for
3907        # $ctx->diff(), so we'll create a default one
3908        my $pool = SVN::Pool->new_default_sub;
3909
3910        $ra_invalid = 1; # this will open a new SVN::Ra connection to $url1
3911        $ctx->diff([], $url1, $rev1, $url2, $rev2, 1, 1, 0, $out, $out);
3912        $out->flush;
3913        my $ret = (($out->stat)[7] == 0);
3914        close $out or croak $!;
3915
3916        $ret;
3917}
3918
3919sub get_commit_editor {
3920        my ($self, $log, $cb, $pool) = @_;
3921        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
3922        $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
3923}
3924
3925sub gs_do_update {
3926        my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
3927        my $new = ($rev_a == $rev_b);
3928        my $path = $gs->{path};
3929
3930        if ($new && -e $gs->{index}) {
3931                unlink $gs->{index} or die
3932                  "Couldn't unlink index: $gs->{index}: $!\n";
3933        }
3934        my $pool = SVN::Pool->new;
3935        $editor->set_path_strip($path);
3936        my (@pc) = split m#/#, $path;
3937        my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
3938                                        1, $editor, $pool);
3939        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
3940
3941        # Since we can't rely on svn_ra_reparent being available, we'll
3942        # just have to do some magic with set_path to make it so
3943        # we only want a partial path.
3944        my $sp = '';
3945        my $final = join('/', @pc);
3946        while (@pc) {
3947                $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
3948                $sp .= '/' if length $sp;
3949                $sp .= shift @pc;
3950        }
3951        die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
3952
3953        $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
3954
3955        $reporter->finish_report($pool);
3956        $pool->clear;
3957        $editor->{git_commit_ok};
3958}
3959
3960# this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
3961# svn_ra_reparent didn't work before 1.4)
3962sub gs_do_switch {
3963        my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
3964        my $path = $gs->{path};
3965        my $pool = SVN::Pool->new;
3966
3967        my $full_url = $self->{url};
3968        my $old_url = $full_url;
3969        $full_url .= '/' . escape_uri_only($path) if length $path;
3970        my ($ra, $reparented);
3971        if ($old_url ne $full_url) {
3972                if ($old_url !~ m#^svn(\+ssh)?://#) {
3973                        SVN::_Ra::svn_ra_reparent($self->{session}, $full_url,
3974                                                  $pool);
3975                        $self->{url} = $full_url;
3976                        $reparented = 1;
3977                } else {
3978                        $_[0] = undef;
3979                        $self = undef;
3980                        $RA = undef;
3981                        $ra = Git::SVN::Ra->new($full_url);
3982                        $ra_invalid = 1;
3983                }
3984        }
3985        $ra ||= $self;
3986        my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
3987        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
3988        $reporter->set_path('', $rev_a, 0, @lock, $pool);
3989        $reporter->finish_report($pool);
3990
3991        if ($reparented) {
3992                SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
3993                $self->{url} = $old_url;
3994        }
3995
3996        $pool->clear;
3997        $editor->{git_commit_ok};
3998}
3999
4000sub longest_common_path {
4001        my ($gsv, $globs) = @_;
4002        my %common;
4003        my $common_max = scalar @$gsv;
4004
4005        foreach my $gs (@$gsv) {
4006                my @tmp = split m#/#, $gs->{path};
4007                my $p = '';
4008                foreach (@tmp) {
4009                        $p .= length($p) ? "/$_" : $_;
4010                        $common{$p} ||= 0;
4011                        $common{$p}++;
4012                }
4013        }
4014        $globs ||= [];
4015        $common_max += scalar @$globs;
4016        foreach my $glob (@$globs) {
4017                my @tmp = split m#/#, $glob->{path}->{left};
4018                my $p = '';
4019                foreach (@tmp) {
4020                        $p .= length($p) ? "/$_" : $_;
4021                        $common{$p} ||= 0;
4022                        $common{$p}++;
4023                }
4024        }
4025
4026        my $longest_path = '';
4027        foreach (sort {length $b <=> length $a} keys %common) {
4028                if ($common{$_} == $common_max) {
4029                        $longest_path = $_;
4030                        last;
4031                }
4032        }
4033        $longest_path;
4034}
4035
4036sub gs_fetch_loop_common {
4037        my ($self, $base, $head, $gsv, $globs) = @_;
4038        return if ($base > $head);
4039        my $inc = $_log_window_size;
4040        my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
4041        my $longest_path = longest_common_path($gsv, $globs);
4042        my $ra_url = $self->{url};
4043        while (1) {
4044                my %revs;
4045                my $err;
4046                my $err_handler = $SVN::Error::handler;
4047                $SVN::Error::handler = sub {
4048                        ($err) = @_;
4049                        skip_unknown_revs($err);
4050                };
4051                sub _cb {
4052                        my ($paths, $r, $author, $date, $log) = @_;
4053                        [ dup_changed_paths($paths),
4054                          { author => $author, date => $date, log => $log } ];
4055                }
4056                $self->get_log([$longest_path], $min, $max, 0, 1, 1,
4057                               sub { $revs{$_[1]} = _cb(@_) });
4058                if ($err && $max >= $head) {
4059                        print STDERR "Path '$longest_path' ",
4060                                     "was probably deleted:\n",
4061                                     $err->expanded_message,
4062                                     "\nWill attempt to follow ",
4063                                     "revisions r$min .. r$max ",
4064                                     "committed before the deletion\n";
4065                        my $hi = $max;
4066                        while (--$hi >= $min) {
4067                                my $ok;
4068                                $self->get_log([$longest_path], $min, $hi,
4069                                               0, 1, 1, sub {
4070                                               $ok ||= $_[1];
4071                                               $revs{$_[1]} = _cb(@_) });
4072                                if ($ok) {
4073                                        print STDERR "r$min .. r$ok OK\n";
4074                                        last;
4075                                }
4076                        }
4077                }
4078                $SVN::Error::handler = $err_handler;
4079
4080                my %exists = map { $_->{path} => $_ } @$gsv;
4081                foreach my $r (sort {$a <=> $b} keys %revs) {
4082                        my ($paths, $logged) = @{$revs{$r}};
4083
4084                        foreach my $gs ($self->match_globs(\%exists, $paths,
4085                                                           $globs, $r)) {
4086                                if ($gs->rev_map_max >= $r) {
4087                                        next;
4088                                }
4089                                next unless $gs->match_paths($paths, $r);
4090                                $gs->{logged_rev_props} = $logged;
4091                                if (my $last_commit = $gs->last_commit) {
4092                                        $gs->assert_index_clean($last_commit);
4093                                }
4094                                my $log_entry = $gs->do_fetch($paths, $r);
4095                                if ($log_entry) {
4096                                        $gs->do_git_commit($log_entry);
4097                                }
4098                                $INDEX_FILES{$gs->{index}} = 1;
4099                        }
4100                        foreach my $g (@$globs) {
4101                                my $k = "svn-remote.$g->{remote}." .
4102                                        "$g->{t}-maxRev";
4103                                Git::SVN::tmp_config($k, $r);
4104                        }
4105                        if ($ra_invalid) {
4106                                $_[0] = undef;
4107                                $self = undef;
4108                                $RA = undef;
4109                                $self = Git::SVN::Ra->new($ra_url);
4110                                $ra_invalid = undef;
4111                        }
4112                }
4113                # pre-fill the .rev_db since it'll eventually get filled in
4114                # with '0' x40 if something new gets committed
4115                foreach my $gs (@$gsv) {
4116                        next if $gs->rev_map_max >= $max;
4117                        next if defined $gs->rev_map_get($max);
4118                        $gs->rev_map_set($max, 0 x40);
4119                }
4120                foreach my $g (@$globs) {
4121                        my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
4122                        Git::SVN::tmp_config($k, $max);
4123                }
4124                last if $max >= $head;
4125                $min = $max + 1;
4126                $max += $inc;
4127                $max = $head if ($max > $head);
4128        }
4129        Git::SVN::gc();
4130}
4131
4132sub get_dir_globbed {
4133        my ($self, $left, $depth, $r) = @_;
4134
4135        my @x = eval { $self->get_dir($left, $r) };
4136        return unless scalar @x == 3;
4137        my $dirents = $x[0];
4138        my @finalents;
4139        foreach my $de (keys %$dirents) {
4140                next if $dirents->{$de}->{kind} != $SVN::Node::dir;
4141                if ($depth > 1) {
4142                        my @args = ("$left/$de", $depth - 1, $r);
4143                        foreach my $dir ($self->get_dir_globbed(@args)) {
4144                                push @finalents, "$de/$dir";
4145                        }
4146                } else {
4147                        push @finalents, $de;
4148                }
4149        }
4150        @finalents;
4151}
4152
4153sub match_globs {
4154        my ($self, $exists, $paths, $globs, $r) = @_;
4155
4156        sub get_dir_check {
4157                my ($self, $exists, $g, $r) = @_;
4158
4159                my @dirs = $self->get_dir_globbed($g->{path}->{left},
4160                                                  $g->{path}->{depth},
4161                                                  $r);
4162
4163                foreach my $de (@dirs) {
4164                        my $p = $g->{path}->full_path($de);
4165                        next if $exists->{$p};
4166                        next if (length $g->{path}->{right} &&
4167                                 ($self->check_path($p, $r) !=
4168                                  $SVN::Node::dir));
4169                        $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
4170                                         $g->{ref}->full_path($de), 1);
4171                }
4172        }
4173        foreach my $g (@$globs) {
4174                if (my $path = $paths->{"/$g->{path}->{left}"}) {
4175                        if ($path->{action} =~ /^[AR]$/) {
4176                                get_dir_check($self, $exists, $g, $r);
4177                        }
4178                }
4179                foreach (keys %$paths) {
4180                        if (/$g->{path}->{left_regex}/ &&
4181                            !/$g->{path}->{regex}/) {
4182                                next if $paths->{$_}->{action} !~ /^[AR]$/;
4183                                get_dir_check($self, $exists, $g, $r);
4184                        }
4185                        next unless /$g->{path}->{regex}/;
4186                        my $p = $1;
4187                        my $pathname = $g->{path}->full_path($p);
4188                        next if $exists->{$pathname};
4189                        next if ($self->check_path($pathname, $r) !=
4190                                 $SVN::Node::dir);
4191                        $exists->{$pathname} = Git::SVN->init(
4192                                              $self->{url}, $pathname, undef,
4193                                              $g->{ref}->full_path($p), 1);
4194                }
4195                my $c = '';
4196                foreach (split m#/#, $g->{path}->{left}) {
4197                        $c .= "/$_";
4198                        next unless ($paths->{$c} &&
4199                                     ($paths->{$c}->{action} =~ /^[AR]$/));
4200                        get_dir_check($self, $exists, $g, $r);
4201                }
4202        }
4203        values %$exists;
4204}
4205
4206sub minimize_url {
4207        my ($self) = @_;
4208        return $self->{url} if ($self->{url} eq $self->{repos_root});
4209        my $url = $self->{repos_root};
4210        my @components = split(m!/!, $self->{svn_path});
4211        my $c = '';
4212        do {
4213                $url .= "/$c" if length $c;
4214                eval { (ref $self)->new($url)->get_latest_revnum };
4215        } while ($@ && ($c = shift @components));
4216        $url;
4217}
4218
4219sub can_do_switch {
4220        my $self = shift;
4221        unless (defined $can_do_switch) {
4222                my $pool = SVN::Pool->new;
4223                my $rep = eval {
4224                        $self->do_switch(1, '', 0, $self->{url},
4225                                         SVN::Delta::Editor->new, $pool);
4226                };
4227                if ($@) {
4228                        $can_do_switch = 0;
4229                } else {
4230                        $rep->abort_report($pool);
4231                        $can_do_switch = 1;
4232                }
4233                $pool->clear;
4234        }
4235        $can_do_switch;
4236}
4237
4238sub skip_unknown_revs {
4239        my ($err) = @_;
4240        my $errno = $err->apr_err();
4241        # Maybe the branch we're tracking didn't
4242        # exist when the repo started, so it's
4243        # not an error if it doesn't, just continue
4244        #
4245        # Wonderfully consistent library, eh?
4246        # 160013 - svn:// and file://
4247        # 175002 - http(s)://
4248        # 175007 - http(s):// (this repo required authorization, too...)
4249        #   More codes may be discovered later...
4250        if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
4251                my $err_key = $err->expanded_message;
4252                # revision numbers change every time, filter them out
4253                $err_key =~ s/\d+/\0/g;
4254                $err_key = "$errno\0$err_key";
4255                unless ($ignored_err{$err_key}) {
4256                        warn "W: Ignoring error from SVN, path probably ",
4257                             "does not exist: ($errno): ",
4258                             $err->expanded_message,"\n";
4259                        warn "W: Do not be alarmed at the above message ",
4260                             "git-svn is just searching aggressively for ",
4261                             "old history.\n",
4262                             "This may take a while on large repositories\n";
4263                        $ignored_err{$err_key} = 1;
4264                }
4265                return;
4266        }
4267        die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
4268}
4269
4270# svn_log_changed_path_t objects passed to get_log are likely to be
4271# overwritten even if only the refs are copied to an external variable,
4272# so we should dup the structures in their entirety.  Using an externally
4273# passed pool (instead of our temporary and quickly cleared pool in
4274# Git::SVN::Ra) does not help matters at all...
4275sub dup_changed_paths {
4276        my ($paths) = @_;
4277        return undef unless $paths;
4278        my %ret;
4279        foreach my $p (keys %$paths) {
4280                my $i = $paths->{$p};
4281                my %s = map { $_ => $i->$_ }
4282                              qw/copyfrom_path copyfrom_rev action/;
4283                $ret{$p} = \%s;
4284        }
4285        \%ret;
4286}
4287
4288package Git::SVN::Log;
4289use strict;
4290use warnings;
4291use POSIX qw/strftime/;
4292use constant commit_log_separator => ('-' x 72) . "\n";
4293use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
4294            %rusers $show_commit $incremental/;
4295my $l_fmt;
4296
4297sub cmt_showable {
4298        my ($c) = @_;
4299        return 1 if defined $c->{r};
4300
4301        # big commit message got truncated by the 16k pretty buffer in rev-list
4302        if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
4303                                $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
4304                @{$c->{l}} = ();
4305                my @log = command(qw/cat-file commit/, $c->{c});
4306
4307                # shift off the headers
4308                shift @log while ($log[0] ne '');
4309                shift @log;
4310
4311                # TODO: make $c->{l} not have a trailing newline in the future
4312                @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
4313
4314                (undef, $c->{r}, undef) = ::extract_metadata(
4315                                (grep(/^git-svn-id: /, @log))[-1]);
4316        }
4317        return defined $c->{r};
4318}
4319
4320sub log_use_color {
4321        return $color || Git->repository->get_colorbool('color.diff');
4322}
4323
4324sub git_svn_log_cmd {
4325        my ($r_min, $r_max, @args) = @_;
4326        my $head = 'HEAD';
4327        my (@files, @log_opts);
4328        foreach my $x (@args) {
4329                if ($x eq '--' || @files) {
4330                        push @files, $x;
4331                } else {
4332                        if (::verify_ref("$x^0")) {
4333                                $head = $x;
4334                        } else {
4335                                push @log_opts, $x;
4336                        }
4337                }
4338        }
4339
4340        my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
4341        $gs ||= Git::SVN->_new;
4342        my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
4343                   $gs->refname);
4344        push @cmd, '-r' unless $non_recursive;
4345        push @cmd, qw/--raw --name-status/ if $verbose;
4346        push @cmd, '--color' if log_use_color();
4347        push @cmd, @log_opts;
4348        if (defined $r_max && $r_max == $r_min) {
4349                push @cmd, '--max-count=1';
4350                if (my $c = $gs->rev_map_get($r_max)) {
4351                        push @cmd, $c;
4352                }
4353        } elsif (defined $r_max) {
4354                if ($r_max < $r_min) {
4355                        ($r_min, $r_max) = ($r_max, $r_min);
4356                }
4357                my (undef, $c_max) = $gs->find_rev_before($r_max, 1, $r_min);
4358                my (undef, $c_min) = $gs->find_rev_after($r_min, 1, $r_max);
4359                # If there are no commits in the range, both $c_max and $c_min
4360                # will be undefined.  If there is at least 1 commit in the
4361                # range, both will be defined.
4362                return () if !defined $c_min || !defined $c_max;
4363                if ($c_min eq $c_max) {
4364                        push @cmd, '--max-count=1', $c_min;
4365                } else {
4366                        push @cmd, '--boundary', "$c_min..$c_max";
4367                }
4368        }
4369        return (@cmd, @files);
4370}
4371
4372# adapted from pager.c
4373sub config_pager {
4374        $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
4375        if (!defined $pager) {
4376                $pager = 'less';
4377        } elsif (length $pager == 0 || $pager eq 'cat') {
4378                $pager = undef;
4379        }
4380        $ENV{GIT_PAGER_IN_USE} = defined($pager);
4381}
4382
4383sub run_pager {
4384        return unless -t *STDOUT && defined $pager;
4385        pipe my $rfd, my $wfd or return;
4386        defined(my $pid = fork) or ::fatal "Can't fork: $!";
4387        if (!$pid) {
4388                open STDOUT, '>&', $wfd or
4389                                     ::fatal "Can't redirect to stdout: $!";
4390                return;
4391        }
4392        open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!";
4393        $ENV{LESS} ||= 'FRSX';
4394        exec $pager or ::fatal "Can't run pager: $! ($pager)";
4395}
4396
4397sub format_svn_date {
4398        return strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)", localtime(shift));
4399}
4400
4401sub parse_git_date {
4402        my ($t, $tz) = @_;
4403        # Date::Parse isn't in the standard Perl distro :(
4404        if ($tz =~ s/^\+//) {
4405                $t += tz_to_s_offset($tz);
4406        } elsif ($tz =~ s/^\-//) {
4407                $t -= tz_to_s_offset($tz);
4408        }
4409        return $t;
4410}
4411
4412sub set_local_timezone {
4413        if (defined $TZ) {
4414                $ENV{TZ} = $TZ;
4415        } else {
4416                delete $ENV{TZ};
4417        }
4418}
4419
4420sub tz_to_s_offset {
4421        my ($tz) = @_;
4422        $tz =~ s/(\d\d)$//;
4423        return ($1 * 60) + ($tz * 3600);
4424}
4425
4426sub get_author_info {
4427        my ($dest, $author, $t, $tz) = @_;
4428        $author =~ s/(?:^\s*|\s*$)//g;
4429        $dest->{a_raw} = $author;
4430        my $au;
4431        if ($::_authors) {
4432                $au = $rusers{$author} || undef;
4433        }
4434        if (!$au) {
4435                ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
4436        }
4437        $dest->{t} = $t;
4438        $dest->{tz} = $tz;
4439        $dest->{a} = $au;
4440        $dest->{t_utc} = parse_git_date($t, $tz);
4441}
4442
4443sub process_commit {
4444        my ($c, $r_min, $r_max, $defer) = @_;
4445        if (defined $r_min && defined $r_max) {
4446                if ($r_min == $c->{r} && $r_min == $r_max) {
4447                        show_commit($c);
4448                        return 0;
4449                }
4450                return 1 if $r_min == $r_max;
4451                if ($r_min < $r_max) {
4452                        # we need to reverse the print order
4453                        return 0 if (defined $limit && --$limit < 0);
4454                        push @$defer, $c;
4455                        return 1;
4456                }
4457                if ($r_min != $r_max) {
4458                        return 1 if ($r_min < $c->{r});
4459                        return 1 if ($r_max > $c->{r});
4460                }
4461        }
4462        return 0 if (defined $limit && --$limit < 0);
4463        show_commit($c);
4464        return 1;
4465}
4466
4467sub show_commit {
4468        my $c = shift;
4469        if ($oneline) {
4470                my $x = "\n";
4471                if (my $l = $c->{l}) {
4472                        while ($l->[0] =~ /^\s*$/) { shift @$l }
4473                        $x = $l->[0];
4474                }
4475                $l_fmt ||= 'A' . length($c->{r});
4476                print 'r',pack($l_fmt, $c->{r}),' | ';
4477                print "$c->{c} | " if $show_commit;
4478                print $x;
4479        } else {
4480                show_commit_normal($c);
4481        }
4482}
4483
4484sub show_commit_changed_paths {
4485        my ($c) = @_;
4486        return unless $c->{changed};
4487        print "Changed paths:\n", @{$c->{changed}};
4488}
4489
4490sub show_commit_normal {
4491        my ($c) = @_;
4492        print commit_log_separator, "r$c->{r} | ";
4493        print "$c->{c} | " if $show_commit;
4494        print "$c->{a} | ", format_svn_date($c->{t_utc}), ' | ';
4495        my $nr_line = 0;
4496
4497        if (my $l = $c->{l}) {
4498                while ($l->[$#$l] eq "\n" && $#$l > 0
4499                                          && $l->[($#$l - 1)] eq "\n") {
4500                        pop @$l;
4501                }
4502                $nr_line = scalar @$l;
4503                if (!$nr_line) {
4504                        print "1 line\n\n\n";
4505                } else {
4506                        if ($nr_line == 1) {
4507                                $nr_line = '1 line';
4508                        } else {
4509                                $nr_line .= ' lines';
4510                        }
4511                        print $nr_line, "\n";
4512                        show_commit_changed_paths($c);
4513                        print "\n";
4514                        print $_ foreach @$l;
4515                }
4516        } else {
4517                print "1 line\n";
4518                show_commit_changed_paths($c);
4519                print "\n";
4520
4521        }
4522        foreach my $x (qw/raw stat diff/) {
4523                if ($c->{$x}) {
4524                        print "\n";
4525                        print $_ foreach @{$c->{$x}}
4526                }
4527        }
4528}
4529
4530sub cmd_show_log {
4531        my (@args) = @_;
4532        my ($r_min, $r_max);
4533        my $r_last = -1; # prevent dupes
4534        set_local_timezone();
4535        if (defined $::_revision) {
4536                if ($::_revision =~ /^(\d+):(\d+)$/) {
4537                        ($r_min, $r_max) = ($1, $2);
4538                } elsif ($::_revision =~ /^\d+$/) {
4539                        $r_min = $r_max = $::_revision;
4540                } else {
4541                        ::fatal "-r$::_revision is not supported, use ",
4542                                "standard 'git log' arguments instead";
4543                }
4544        }
4545
4546        config_pager();
4547        @args = git_svn_log_cmd($r_min, $r_max, @args);
4548        if (!@args) {
4549                print commit_log_separator unless $incremental || $oneline;
4550                return;
4551        }
4552        my $log = command_output_pipe(@args);
4553        run_pager();
4554        my (@k, $c, $d, $stat);
4555        my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
4556        while (<$log>) {
4557                if (/^${esc_color}commit -?($::sha1_short)/o) {
4558                        my $cmt = $1;
4559                        if ($c && cmt_showable($c) && $c->{r} != $r_last) {
4560                                $r_last = $c->{r};
4561                                process_commit($c, $r_min, $r_max, \@k) or
4562                                                                goto out;
4563                        }
4564                        $d = undef;
4565                        $c = { c => $cmt };
4566                } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
4567                        get_author_info($c, $1, $2, $3);
4568                } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
4569                        # ignore
4570                } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
4571                        push @{$c->{raw}}, $_;
4572                } elsif (/^${esc_color}[ACRMDT]\t/) {
4573                        # we could add $SVN->{svn_path} here, but that requires
4574                        # remote access at the moment (repo_path_split)...
4575                        s#^(${esc_color})([ACRMDT])\t#$1   $2 #o;
4576                        push @{$c->{changed}}, $_;
4577                } elsif (/^${esc_color}diff /o) {
4578                        $d = 1;
4579                        push @{$c->{diff}}, $_;
4580                } elsif ($d) {
4581                        push @{$c->{diff}}, $_;
4582                } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
4583                          $esc_color*[\+\-]*$esc_color$/x) {
4584                        $stat = 1;
4585                        push @{$c->{stat}}, $_;
4586                } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
4587                        push @{$c->{stat}}, $_;
4588                        $stat = undef;
4589                } elsif (/^${esc_color}    (git-svn-id:.+)$/o) {
4590                        ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
4591                } elsif (s/^${esc_color}    //o) {
4592                        push @{$c->{l}}, $_;
4593                }
4594        }
4595        if ($c && defined $c->{r} && $c->{r} != $r_last) {
4596                $r_last = $c->{r};
4597                process_commit($c, $r_min, $r_max, \@k);
4598        }
4599        if (@k) {
4600                ($r_min, $r_max) = ($r_max, $r_min);
4601                process_commit($_, $r_min, $r_max) foreach reverse @k;
4602        }
4603out:
4604        close $log;
4605        print commit_log_separator unless $incremental || $oneline;
4606}
4607
4608sub cmd_blame {
4609        my $path = pop;
4610
4611        config_pager();
4612        run_pager();
4613
4614        my ($fh, $ctx, $rev);
4615
4616        if ($_git_format) {
4617                ($fh, $ctx) = command_output_pipe('blame', @_, $path);
4618                while (my $line = <$fh>) {
4619                        if ($line =~ /^\^?([[:xdigit:]]+)\s/) {
4620                                # Uncommitted edits show up as a rev ID of
4621                                # all zeros, which we can't look up with
4622                                # cmt_metadata
4623                                if ($1 !~ /^0+$/) {
4624                                        (undef, $rev, undef) =
4625                                                ::cmt_metadata($1);
4626                                        $rev = '0' if (!$rev);
4627                                } else {
4628                                        $rev = '0';
4629                                }
4630                                $rev = sprintf('%-10s', $rev);
4631                                $line =~ s/^\^?[[:xdigit:]]+(\s)/$rev$1/;
4632                        }
4633                        print $line;
4634                }
4635        } else {
4636                ($fh, $ctx) = command_output_pipe('blame', '-p', @_, 'HEAD',
4637                                                  '--', $path);
4638                my ($sha1);
4639                my %authors;
4640                while (my $line = <$fh>) {
4641                        if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
4642                                $sha1 = $1;
4643                                (undef, $rev, undef) = ::cmt_metadata($1);
4644                                $rev = '0' if (!$rev);
4645                        }
4646                        elsif ($line =~ /^author (.*)/) {
4647                                $authors{$rev} = $1;
4648                                $authors{$rev} =~ s/\s/_/g;
4649                        }
4650                        elsif ($line =~ /^\t(.*)$/) {
4651                                printf("%6s %10s %s\n", $rev, $authors{$rev}, $1);
4652                        }
4653                }
4654        }
4655        command_close_pipe($fh, $ctx);
4656}
4657
4658package Git::SVN::Migration;
4659# these version numbers do NOT correspond to actual version numbers
4660# of git nor git-svn.  They are just relative.
4661#
4662# v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
4663#
4664# v1 layout: .git/$id/info/url, refs/remotes/$id
4665#
4666# v2 layout: .git/svn/$id/info/url, refs/remotes/$id
4667#
4668# v3 layout: .git/svn/$id, refs/remotes/$id
4669#            - info/url may remain for backwards compatibility
4670#            - this is what we migrate up to this layout automatically,
4671#            - this will be used by git svn init on single branches
4672# v3.1 layout (auto migrated):
4673#            - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
4674#              for backwards compatibility
4675#
4676# v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
4677#            - this is only created for newly multi-init-ed
4678#              repositories.  Similar in spirit to the
4679#              --use-separate-remotes option in git-clone (now default)
4680#            - we do not automatically migrate to this (following
4681#              the example set by core git)
4682#
4683# v5 layout: .rev_db.$UUID => .rev_map.$UUID
4684#            - newer, more-efficient format that uses 24-bytes per record
4685#              with no filler space.
4686#            - use xxd -c24 < .rev_map.$UUID to view and debug
4687#            - This is a one-way migration, repositories updated to the
4688#              new format will not be able to use old git-svn without
4689#              rebuilding the .rev_db.  Rebuilding the rev_db is not
4690#              possible if noMetadata or useSvmProps are set; but should
4691#              be no problem for users that use the (sensible) defaults.
4692use strict;
4693use warnings;
4694use Carp qw/croak/;
4695use File::Path qw/mkpath/;
4696use File::Basename qw/dirname basename/;
4697use vars qw/$_minimize/;
4698
4699sub migrate_from_v0 {
4700        my $git_dir = $ENV{GIT_DIR};
4701        return undef unless -d $git_dir;
4702        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
4703        my $migrated = 0;
4704        while (<$fh>) {
4705                chomp;
4706                my ($id, $orig_ref) = ($_, $_);
4707                next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
4708                next unless -f "$git_dir/$id/info/url";
4709                my $new_ref = "refs/remotes/$id";
4710                if (::verify_ref("$new_ref^0")) {
4711                        print STDERR "W: $orig_ref is probably an old ",
4712                                     "branch used by an ancient version of ",
4713                                     "git-svn.\n",
4714                                     "However, $new_ref also exists.\n",
4715                                     "We will not be able ",
4716                                     "to use this branch until this ",
4717                                     "ambiguity is resolved.\n";
4718                        next;
4719                }
4720                print STDERR "Migrating from v0 layout...\n" if !$migrated;
4721                print STDERR "Renaming ref: $orig_ref => $new_ref\n";
4722                command_noisy('update-ref', $new_ref, $orig_ref);
4723                command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
4724                $migrated++;
4725        }
4726        command_close_pipe($fh, $ctx);
4727        print STDERR "Done migrating from v0 layout...\n" if $migrated;
4728        $migrated;
4729}
4730
4731sub migrate_from_v1 {
4732        my $git_dir = $ENV{GIT_DIR};
4733        my $migrated = 0;
4734        return $migrated unless -d $git_dir;
4735        my $svn_dir = "$git_dir/svn";
4736
4737        # just in case somebody used 'svn' as their $id at some point...
4738        return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
4739
4740        print STDERR "Migrating from a git-svn v1 layout...\n";
4741        mkpath([$svn_dir]);
4742        print STDERR "Data from a previous version of git-svn exists, but\n\t",
4743                     "$svn_dir\n\t(required for this version ",
4744                     "($::VERSION) of git-svn) does not exist.\n";
4745        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
4746        while (<$fh>) {
4747                my $x = $_;
4748                next unless $x =~ s#^refs/remotes/##;
4749                chomp $x;
4750                next unless -f "$git_dir/$x/info/url";
4751                my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
4752                next unless $u;
4753                my $dn = dirname("$git_dir/svn/$x");
4754                mkpath([$dn]) unless -d $dn;
4755                if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
4756                        mkpath(["$git_dir/svn/svn"]);
4757                        print STDERR " - $git_dir/$x/info => ",
4758                                        "$git_dir/svn/$x/info\n";
4759                        rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
4760                               croak "$!: $x";
4761                        # don't worry too much about these, they probably
4762                        # don't exist with repos this old (save for index,
4763                        # and we can easily regenerate that)
4764                        foreach my $f (qw/unhandled.log index .rev_db/) {
4765                                rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
4766                        }
4767                } else {
4768                        print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
4769                        rename "$git_dir/$x", "$git_dir/svn/$x" or
4770                               croak "$!: $x";
4771                }
4772                $migrated++;
4773        }
4774        command_close_pipe($fh, $ctx);
4775        print STDERR "Done migrating from a git-svn v1 layout\n";
4776        $migrated;
4777}
4778
4779sub read_old_urls {
4780        my ($l_map, $pfx, $path) = @_;
4781        my @dir;
4782        foreach (<$path/*>) {
4783                if (-r "$_/info/url") {
4784                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
4785                        my $ref_id = $pfx . basename $_;
4786                        my $url = ::file_to_s("$_/info/url");
4787                        $l_map->{$ref_id} = $url;
4788                } elsif (-d $_) {
4789                        push @dir, $_;
4790                }
4791        }
4792        foreach (@dir) {
4793                my $x = $_;
4794                $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
4795                read_old_urls($l_map, $x, $_);
4796        }
4797}
4798
4799sub migrate_from_v2 {
4800        my @cfg = command(qw/config -l/);
4801        return if grep /^svn-remote\..+\.url=/, @cfg;
4802        my %l_map;
4803        read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
4804        my $migrated = 0;
4805
4806        foreach my $ref_id (sort keys %l_map) {
4807                eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
4808                if ($@) {
4809                        Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
4810                }
4811                $migrated++;
4812        }
4813        $migrated;
4814}
4815
4816sub minimize_connections {
4817        my $r = Git::SVN::read_all_remotes();
4818        my $new_urls = {};
4819        my $root_repos = {};
4820        foreach my $repo_id (keys %$r) {
4821                my $url = $r->{$repo_id}->{url} or next;
4822                my $fetch = $r->{$repo_id}->{fetch} or next;
4823                my $ra = Git::SVN::Ra->new($url);
4824
4825                # skip existing cases where we already connect to the root
4826                if (($ra->{url} eq $ra->{repos_root}) ||
4827                    ($ra->{repos_root} eq $repo_id)) {
4828                        $root_repos->{$ra->{url}} = $repo_id;
4829                        next;
4830                }
4831
4832                my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
4833                my $root_path = $ra->{url};
4834                $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
4835                foreach my $path (keys %$fetch) {
4836                        my $ref_id = $fetch->{$path};
4837                        my $gs = Git::SVN->new($ref_id, $repo_id, $path);
4838
4839                        # make sure we can read when connecting to
4840                        # a higher level of a repository
4841                        my ($last_rev, undef) = $gs->last_rev_commit;
4842                        if (!defined $last_rev) {
4843                                $last_rev = eval {
4844                                        $root_ra->get_latest_revnum;
4845                                };
4846                                next if $@;
4847                        }
4848                        my $new = $root_path;
4849                        $new .= length $path ? "/$path" : '';
4850                        eval {
4851                                $root_ra->get_log([$new], $last_rev, $last_rev,
4852                                                  0, 0, 1, sub { });
4853                        };
4854                        next if $@;
4855                        $new_urls->{$ra->{repos_root}}->{$new} =
4856                                { ref_id => $ref_id,
4857                                  old_repo_id => $repo_id,
4858                                  old_path => $path };
4859                }
4860        }
4861
4862        my @emptied;
4863        foreach my $url (keys %$new_urls) {
4864                # see if we can re-use an existing [svn-remote "repo_id"]
4865                # instead of creating a(n ugly) new section:
4866                my $repo_id = $root_repos->{$url} || $url;
4867
4868                my $fetch = $new_urls->{$url};
4869                foreach my $path (keys %$fetch) {
4870                        my $x = $fetch->{$path};
4871                        Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
4872                        my $pfx = "svn-remote.$x->{old_repo_id}";
4873
4874                        my $old_fetch = quotemeta("$x->{old_path}:".
4875                                                  "refs/remotes/$x->{ref_id}");
4876                        command_noisy(qw/config --unset/,
4877                                      "$pfx.fetch", '^'. $old_fetch . '$');
4878                        delete $r->{$x->{old_repo_id}}->
4879                               {fetch}->{$x->{old_path}};
4880                        if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
4881                                command_noisy(qw/config --unset/,
4882                                              "$pfx.url");
4883                                push @emptied, $x->{old_repo_id}
4884                        }
4885                }
4886        }
4887        if (@emptied) {
4888                my $file = $ENV{GIT_CONFIG} || $ENV{GIT_CONFIG_LOCAL} ||
4889                           "$ENV{GIT_DIR}/config";
4890                print STDERR <<EOF;
4891The following [svn-remote] sections in your config file ($file) are empty
4892and can be safely removed:
4893EOF
4894                print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
4895        }
4896}
4897
4898sub migration_check {
4899        migrate_from_v0();
4900        migrate_from_v1();
4901        migrate_from_v2();
4902        minimize_connections() if $_minimize;
4903}
4904
4905package Git::IndexInfo;
4906use strict;
4907use warnings;
4908use Git qw/command_input_pipe command_close_pipe/;
4909
4910sub new {
4911        my ($class) = @_;
4912        my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
4913        bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
4914}
4915
4916sub remove {
4917        my ($self, $path) = @_;
4918        if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
4919                return ++$self->{nr};
4920        }
4921        undef;
4922}
4923
4924sub update {
4925        my ($self, $mode, $hash, $path) = @_;
4926        if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
4927                return ++$self->{nr};
4928        }
4929        undef;
4930}
4931
4932sub DESTROY {
4933        my ($self) = @_;
4934        command_close_pipe($self->{gui}, $self->{ctx});
4935}
4936
4937package Git::SVN::GlobSpec;
4938use strict;
4939use warnings;
4940
4941sub new {
4942        my ($class, $glob) = @_;
4943        my $re = $glob;
4944        $re =~ s!/+$!!g; # no need for trailing slashes
4945        $re =~ m!^([^*]*)(\*(?:/\*)*)([^*]*)$!;
4946        my $temp = $re;
4947        my ($left, $right) = ($1, $3);
4948        $re = $2;
4949        my $depth = $re =~ tr/*/*/;
4950        if ($depth != $temp =~ tr/*/*/) {
4951                die "Only one set of wildcard directories " .
4952                        "(e.g. '*' or '*/*/*') is supported: '$glob'\n";
4953        }
4954        if ($depth == 0) {
4955                die "One '*' is needed for glob: '$glob'\n";
4956        }
4957        $re =~ s!\*!\[^/\]*!g;
4958        $re = quotemeta($left) . "($re)" . quotemeta($right);
4959        if (length $left && !($left =~ s!/+$!!g)) {
4960                die "Missing trailing '/' on left side of: '$glob' ($left)\n";
4961        }
4962        if (length $right && !($right =~ s!^/+!!g)) {
4963                die "Missing leading '/' on right side of: '$glob' ($right)\n";
4964        }
4965        my $left_re = qr/^\/\Q$left\E(\/|$)/;
4966        bless { left => $left, right => $right, left_regex => $left_re,
4967                regex => qr/$re/, glob => $glob, depth => $depth }, $class;
4968}
4969
4970sub full_path {
4971        my ($self, $path) = @_;
4972        return (length $self->{left} ? "$self->{left}/" : '') .
4973               $path . (length $self->{right} ? "/$self->{right}" : '');
4974}
4975
4976__END__
4977
4978Data structures:
4979
4980
4981$remotes = { # returned by read_all_remotes()
4982        'svn' => {
4983                # svn-remote.svn.url=https://svn.musicpd.org
4984                url => 'https://svn.musicpd.org',
4985                # svn-remote.svn.fetch=mpd/trunk:trunk
4986                fetch => {
4987                        'mpd/trunk' => 'trunk',
4988                },
4989                # svn-remote.svn.tags=mpd/tags/*:tags/*
4990                tags => {
4991                        path => {
4992                                left => 'mpd/tags',
4993                                right => '',
4994                                regex => qr!mpd/tags/([^/]+)$!,
4995                                glob => 'tags/*',
4996                        },
4997                        ref => {
4998                                left => 'tags',
4999                                right => '',
5000                                regex => qr!tags/([^/]+)$!,
5001                                glob => 'tags/*',
5002                        },
5003                }
5004        }
5005};
5006
5007$log_entry hashref as returned by libsvn_log_entry()
5008{
5009        log => 'whitespace-formatted log entry
5010',                                              # trailing newline is preserved
5011        revision => '8',                        # integer
5012        date => '2004-02-24T17:01:44.108345Z',  # commit date
5013        author => 'committer name'
5014};
5015
5016
5017# this is generated by generate_diff();
5018@mods = array of diff-index line hashes, each element represents one line
5019        of diff-index output
5020
5021diff-index line ($m hash)
5022{
5023        mode_a => first column of diff-index output, no leading ':',
5024        mode_b => second column of diff-index output,
5025        sha1_b => sha1sum of the final blob,
5026        chg => change type [MCRADT],
5027        file_a => original file name of a file (iff chg is 'C' or 'R')
5028        file_b => new/current file name of a file (any chg)
5029}
5030;
5031
5032# retval of read_url_paths{,_all}();
5033$l_map = {
5034        # repository root url
5035        'https://svn.musicpd.org' => {
5036                # repository path               # GIT_SVN_ID
5037                'mpd/trunk'             =>      'trunk',
5038                'mpd/tags/0.11.5'       =>      'tags/0.11.5',
5039        },
5040}
5041
5042Notes:
5043        I don't trust the each() function on unless I created %hash myself
5044        because the internal iterator may not have started at base.