git-svn.perlon commit Merge branch 'jc/maint-add-sync-stat' (c78a249)
   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
   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 IO::File qw//;
  39use File::Basename qw/dirname basename/;
  40use File::Path qw/mkpath/;
  41use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev/;
  42use IPC::Open3;
  43use Git;
  44
  45BEGIN {
  46        # import functions from Git into our packages, en masse
  47        no strict 'refs';
  48        foreach (qw/command command_oneline command_noisy command_output_pipe
  49                    command_input_pipe command_close_pipe/) {
  50                for my $package ( qw(SVN::Git::Editor SVN::Git::Fetcher
  51                        Git::SVN::Migration Git::SVN::Log Git::SVN),
  52                        __PACKAGE__) {
  53                        *{"${package}::$_"} = \&{"Git::$_"};
  54                }
  55        }
  56}
  57
  58my ($SVN);
  59
  60$sha1 = qr/[a-f\d]{40}/;
  61$sha1_short = qr/[a-f\d]{4,40}/;
  62my ($_stdin, $_help, $_edit,
  63        $_message, $_file,
  64        $_template, $_shared,
  65        $_version, $_fetch_all, $_no_rebase,
  66        $_merge, $_strategy, $_dry_run, $_local,
  67        $_prefix, $_no_checkout, $_verbose);
  68$Git::SVN::_follow_parent = 1;
  69my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
  70                    'config-dir=s' => \$Git::SVN::Ra::config_dir,
  71                    'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache );
  72my %fc_opts = ( 'follow-parent|follow!' => \$Git::SVN::_follow_parent,
  73                'authors-file|A=s' => \$_authors,
  74                'repack:i' => \$Git::SVN::_repack,
  75                'noMetadata' => \$Git::SVN::_no_metadata,
  76                'useSvmProps' => \$Git::SVN::_use_svm_props,
  77                'useSvnsyncProps' => \$Git::SVN::_use_svnsync_props,
  78                'log-window-size=i' => \$Git::SVN::Ra::_log_window_size,
  79                'no-checkout' => \$_no_checkout,
  80                'quiet|q' => \$_q,
  81                'repack-flags|repack-args|repack-opts=s' =>
  82                   \$Git::SVN::_repack_flags,
  83                %remote_opts );
  84
  85my ($_trunk, $_tags, $_branches, $_stdlayout);
  86my %icv;
  87my %init_opts = ( 'template=s' => \$_template, 'shared:s' => \$_shared,
  88                  'trunk|T=s' => \$_trunk, 'tags|t=s' => \$_tags,
  89                  'branches|b=s' => \$_branches, 'prefix=s' => \$_prefix,
  90                  'stdlayout|s' => \$_stdlayout,
  91                  'minimize-url|m' => \$Git::SVN::_minimize_url,
  92                  'no-metadata' => sub { $icv{noMetadata} = 1 },
  93                  'use-svm-props' => sub { $icv{useSvmProps} = 1 },
  94                  'use-svnsync-props' => sub { $icv{useSvnsyncProps} = 1 },
  95                  'rewrite-root=s' => sub { $icv{rewriteRoot} = $_[1] },
  96                  %remote_opts );
  97my %cmt_opts = ( 'edit|e' => \$_edit,
  98                'rmdir' => \$SVN::Git::Editor::_rmdir,
  99                'find-copies-harder' => \$SVN::Git::Editor::_find_copies_harder,
 100                'l=i' => \$SVN::Git::Editor::_rename_limit,
 101                'copy-similarity|C=i'=> \$SVN::Git::Editor::_cp_similarity
 102);
 103
 104my %cmd = (
 105        fetch => [ \&cmd_fetch, "Download new revisions from SVN",
 106                        { 'revision|r=s' => \$_revision,
 107                          'fetch-all|all' => \$_fetch_all,
 108                           %fc_opts } ],
 109        clone => [ \&cmd_clone, "Initialize and fetch revisions",
 110                        { 'revision|r=s' => \$_revision,
 111                           %fc_opts, %init_opts } ],
 112        init => [ \&cmd_init, "Initialize a repo for tracking" .
 113                          " (requires URL argument)",
 114                          \%init_opts ],
 115        'multi-init' => [ \&cmd_multi_init,
 116                          "Deprecated alias for ".
 117                          "'$0 init -T<trunk> -b<branches> -t<tags>'",
 118                          \%init_opts ],
 119        dcommit => [ \&cmd_dcommit,
 120                     'Commit several diffs to merge with upstream',
 121                        { 'merge|m|M' => \$_merge,
 122                          'strategy|s=s' => \$_strategy,
 123                          'verbose|v' => \$_verbose,
 124                          'dry-run|n' => \$_dry_run,
 125                          'fetch-all|all' => \$_fetch_all,
 126                          'no-rebase' => \$_no_rebase,
 127                        %cmt_opts, %fc_opts } ],
 128        'set-tree' => [ \&cmd_set_tree,
 129                        "Set an SVN repository to a git tree-ish",
 130                        { 'stdin|' => \$_stdin, %cmt_opts, %fc_opts, } ],
 131        'create-ignore' => [ \&cmd_create_ignore,
 132                             'Create a .gitignore per svn:ignore',
 133                             { 'revision|r=i' => \$_revision
 134                             } ],
 135        'propget' => [ \&cmd_propget,
 136                       'Print the value of a property on a file or directory',
 137                       { 'revision|r=i' => \$_revision } ],
 138        'proplist' => [ \&cmd_proplist,
 139                       'List all properties of a file or directory',
 140                       { 'revision|r=i' => \$_revision } ],
 141        'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
 142                        { 'revision|r=i' => \$_revision
 143                        } ],
 144        'multi-fetch' => [ \&cmd_multi_fetch,
 145                           "Deprecated alias for $0 fetch --all",
 146                           { 'revision|r=s' => \$_revision, %fc_opts } ],
 147        'migrate' => [ sub { },
 148                       # no-op, we automatically run this anyways,
 149                       'Migrate configuration/metadata/layout from
 150                        previous versions of git-svn',
 151                       { 'minimize' => \$Git::SVN::Migration::_minimize,
 152                         %remote_opts } ],
 153        'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
 154                        { 'limit=i' => \$Git::SVN::Log::limit,
 155                          'revision|r=s' => \$_revision,
 156                          'verbose|v' => \$Git::SVN::Log::verbose,
 157                          'incremental' => \$Git::SVN::Log::incremental,
 158                          'oneline' => \$Git::SVN::Log::oneline,
 159                          'show-commit' => \$Git::SVN::Log::show_commit,
 160                          'non-recursive' => \$Git::SVN::Log::non_recursive,
 161                          'authors-file|A=s' => \$_authors,
 162                          'color' => \$Git::SVN::Log::color,
 163                          'pager=s' => \$Git::SVN::Log::pager
 164                        } ],
 165        'find-rev' => [ \&cmd_find_rev, "Translate between SVN revision numbers and tree-ish",
 166                        {} ],
 167        'rebase' => [ \&cmd_rebase, "Fetch and rebase your working directory",
 168                        { 'merge|m|M' => \$_merge,
 169                          'verbose|v' => \$_verbose,
 170                          'strategy|s=s' => \$_strategy,
 171                          'local|l' => \$_local,
 172                          'fetch-all|all' => \$_fetch_all,
 173                          %fc_opts } ],
 174        'commit-diff' => [ \&cmd_commit_diff,
 175                           'Commit a diff between two trees',
 176                        { 'message|m=s' => \$_message,
 177                          'file|F=s' => \$_file,
 178                          'revision|r=s' => \$_revision,
 179                        %cmt_opts } ],
 180);
 181
 182my $cmd;
 183for (my $i = 0; $i < @ARGV; $i++) {
 184        if (defined $cmd{$ARGV[$i]}) {
 185                $cmd = $ARGV[$i];
 186                splice @ARGV, $i, 1;
 187                last;
 188        }
 189};
 190
 191my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
 192
 193read_repo_config(\%opts);
 194Getopt::Long::Configure('pass_through') if ($cmd && $cmd eq 'log');
 195my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
 196                    'minimize-connections' => \$Git::SVN::Migration::_minimize,
 197                    'id|i=s' => \$Git::SVN::default_ref_id,
 198                    'svn-remote|remote|R=s' => sub {
 199                       $Git::SVN::no_reuse_existing = 1;
 200                       $Git::SVN::default_repo_id = $_[1] });
 201exit 1 if (!$rv && $cmd && $cmd ne 'log');
 202
 203usage(0) if $_help;
 204version() if $_version;
 205usage(1) unless defined $cmd;
 206load_authors() if $_authors;
 207
 208# make sure we're always running
 209unless ($cmd =~ /(?:clone|init|multi-init)$/) {
 210        unless (-d $ENV{GIT_DIR}) {
 211                if ($git_dir_user_set) {
 212                        die "GIT_DIR=$ENV{GIT_DIR} explicitly set, ",
 213                            "but it is not a directory\n";
 214                }
 215                my $git_dir = delete $ENV{GIT_DIR};
 216                chomp(my $cdup = command_oneline(qw/rev-parse --show-cdup/));
 217                unless (length $cdup) {
 218                        die "Already at toplevel, but $git_dir ",
 219                            "not found '$cdup'\n";
 220                }
 221                chdir $cdup or die "Unable to chdir up to '$cdup'\n";
 222                unless (-d $git_dir) {
 223                        die "$git_dir still not found after going to ",
 224                            "'$cdup'\n";
 225                }
 226                $ENV{GIT_DIR} = $git_dir;
 227        }
 228}
 229unless ($cmd =~ /^(?:clone|init|multi-init|commit-diff)$/) {
 230        Git::SVN::Migration::migration_check();
 231}
 232Git::SVN::init_vars();
 233eval {
 234        Git::SVN::verify_remotes_sanity();
 235        $cmd{$cmd}->[0]->(@ARGV);
 236};
 237fatal $@ if $@;
 238post_fetch_checkout();
 239exit 0;
 240
 241####################### primary functions ######################
 242sub usage {
 243        my $exit = shift || 0;
 244        my $fd = $exit ? \*STDERR : \*STDOUT;
 245        print $fd <<"";
 246git-svn - bidirectional operations between a single Subversion tree and git
 247Usage: $0 <command> [options] [arguments]\n
 248
 249        print $fd "Available commands:\n" unless $cmd;
 250
 251        foreach (sort keys %cmd) {
 252                next if $cmd && $cmd ne $_;
 253                next if /^multi-/; # don't show deprecated commands
 254                print $fd '  ',pack('A17',$_),$cmd{$_}->[1],"\n";
 255                foreach (sort keys %{$cmd{$_}->[2]}) {
 256                        # mixed-case options are for .git/config only
 257                        next if /[A-Z]/ && /^[a-z]+$/i;
 258                        # prints out arguments as they should be passed:
 259                        my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
 260                        print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
 261                                                        "--$_" : "-$_" }
 262                                                split /\|/,$_)," $x\n";
 263                }
 264        }
 265        print $fd <<"";
 266\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
 267arbitrary identifier if you're tracking multiple SVN branches/repositories in
 268one git repository and want to keep them separate.  See git-svn(1) for more
 269information.
 270
 271        exit $exit;
 272}
 273
 274sub version {
 275        print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
 276        exit 0;
 277}
 278
 279sub do_git_init_db {
 280        unless (-d $ENV{GIT_DIR}) {
 281                my @init_db = ('init');
 282                push @init_db, "--template=$_template" if defined $_template;
 283                if (defined $_shared) {
 284                        if ($_shared =~ /[a-z]/) {
 285                                push @init_db, "--shared=$_shared";
 286                        } else {
 287                                push @init_db, "--shared";
 288                        }
 289                }
 290                command_noisy(@init_db);
 291        }
 292        my $set;
 293        my $pfx = "svn-remote.$Git::SVN::default_repo_id";
 294        foreach my $i (keys %icv) {
 295                die "'$set' and '$i' cannot both be set\n" if $set;
 296                next unless defined $icv{$i};
 297                command_noisy('config', "$pfx.$i", $icv{$i});
 298                $set = $i;
 299        }
 300}
 301
 302sub init_subdir {
 303        my $repo_path = shift or return;
 304        mkpath([$repo_path]) unless -d $repo_path;
 305        chdir $repo_path or die "Couldn't chdir to $repo_path: $!\n";
 306        $ENV{GIT_DIR} = '.git';
 307}
 308
 309sub cmd_clone {
 310        my ($url, $path) = @_;
 311        if (!defined $path &&
 312            (defined $_trunk || defined $_branches || defined $_tags ||
 313             defined $_stdlayout) &&
 314            $url !~ m#^[a-z\+]+://#) {
 315                $path = $url;
 316        }
 317        $path = basename($url) if !defined $path || !length $path;
 318        cmd_init($url, $path);
 319        Git::SVN::fetch_all($Git::SVN::default_repo_id);
 320}
 321
 322sub cmd_init {
 323        if (defined $_stdlayout) {
 324                $_trunk = 'trunk' if (!defined $_trunk);
 325                $_tags = 'tags' if (!defined $_tags);
 326                $_branches = 'branches' if (!defined $_branches);
 327        }
 328        if (defined $_trunk || defined $_branches || defined $_tags) {
 329                return cmd_multi_init(@_);
 330        }
 331        my $url = shift or die "SVN repository location required ",
 332                               "as a command-line argument\n";
 333        init_subdir(@_);
 334        do_git_init_db();
 335
 336        Git::SVN->init($url);
 337}
 338
 339sub cmd_fetch {
 340        if (grep /^\d+=./, @_) {
 341                die "'<rev>=<commit>' fetch arguments are ",
 342                    "no longer supported.\n";
 343        }
 344        my ($remote) = @_;
 345        if (@_ > 1) {
 346                die "Usage: $0 fetch [--all] [svn-remote]\n";
 347        }
 348        $remote ||= $Git::SVN::default_repo_id;
 349        if ($_fetch_all) {
 350                cmd_multi_fetch();
 351        } else {
 352                Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
 353        }
 354}
 355
 356sub cmd_set_tree {
 357        my (@commits) = @_;
 358        if ($_stdin || !@commits) {
 359                print "Reading from stdin...\n";
 360                @commits = ();
 361                while (<STDIN>) {
 362                        if (/\b($sha1_short)\b/o) {
 363                                unshift @commits, $1;
 364                        }
 365                }
 366        }
 367        my @revs;
 368        foreach my $c (@commits) {
 369                my @tmp = command('rev-parse',$c);
 370                if (scalar @tmp == 1) {
 371                        push @revs, $tmp[0];
 372                } elsif (scalar @tmp > 1) {
 373                        push @revs, reverse(command('rev-list',@tmp));
 374                } else {
 375                        fatal "Failed to rev-parse $c";
 376                }
 377        }
 378        my $gs = Git::SVN->new;
 379        my ($r_last, $cmt_last) = $gs->last_rev_commit;
 380        $gs->fetch;
 381        if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
 382                fatal "There are new revisions that were fetched ",
 383                      "and need to be merged (or acknowledged) ",
 384                      "before committing.\nlast rev: $r_last\n",
 385                      " current: $gs->{last_rev}";
 386        }
 387        $gs->set_tree($_) foreach @revs;
 388        print "Done committing ",scalar @revs," revisions to SVN\n";
 389}
 390
 391sub cmd_dcommit {
 392        my $head = shift;
 393        git_cmd_try { command_oneline(qw/diff-index --quiet HEAD/) }
 394                'Cannot dcommit with a dirty index.  Commit your changes first'
 395                . "or stash them with `git stash'.\n";
 396        $head ||= 'HEAD';
 397        my @refs;
 398        my ($url, $rev, $uuid, $gs) = working_head_info($head, \@refs);
 399        print "Committing to $url ...\n";
 400        unless ($gs) {
 401                die "Unable to determine upstream SVN information from ",
 402                    "$head history\n";
 403        }
 404        my $last_rev;
 405        my ($linear_refs, $parents) = linearize_history($gs, \@refs);
 406        if ($_no_rebase && scalar(@$linear_refs) > 1) {
 407                warn "Attempting to commit more than one change while ",
 408                     "--no-rebase is enabled.\n",
 409                     "If these changes depend on each other, re-running ",
 410                     "without --no-rebase will be required."
 411        }
 412        while (1) {
 413                my $d = shift @$linear_refs or last;
 414                unless (defined $last_rev) {
 415                        (undef, $last_rev, undef) = cmt_metadata("$d~1");
 416                        unless (defined $last_rev) {
 417                                fatal "Unable to extract revision information ",
 418                                      "from commit $d~1";
 419                        }
 420                }
 421                if ($_dry_run) {
 422                        print "diff-tree $d~1 $d\n";
 423                } else {
 424                        my $cmt_rev;
 425                        my %ed_opts = ( r => $last_rev,
 426                                        log => get_commit_entry($d)->{log},
 427                                        ra => Git::SVN::Ra->new($gs->full_url),
 428                                        tree_a => "$d~1",
 429                                        tree_b => $d,
 430                                        editor_cb => sub {
 431                                               print "Committed r$_[0]\n";
 432                                               $cmt_rev = $_[0];
 433                                        },
 434                                        svn_path => '');
 435                        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
 436                                print "No changes\n$d~1 == $d\n";
 437                        } elsif ($parents->{$d} && @{$parents->{$d}}) {
 438                                $gs->{inject_parents_dcommit}->{$cmt_rev} =
 439                                                               $parents->{$d};
 440                        }
 441                        $_fetch_all ? $gs->fetch_all : $gs->fetch;
 442                        next if $_no_rebase;
 443
 444                        # we always want to rebase against the current HEAD,
 445                        # not any head that was passed to us
 446                        my @diff = command('diff-tree', $d,
 447                                           $gs->refname, '--');
 448                        my @finish;
 449                        if (@diff) {
 450                                @finish = rebase_cmd();
 451                                print STDERR "W: $d and ", $gs->refname,
 452                                             " differ, using @finish:\n",
 453                                             join("\n", @diff), "\n";
 454                        } else {
 455                                print "No changes between current HEAD and ",
 456                                      $gs->refname,
 457                                      "\nResetting to the latest ",
 458                                      $gs->refname, "\n";
 459                                @finish = qw/reset --mixed/;
 460                        }
 461                        command_noisy(@finish, $gs->refname);
 462                        if (@diff) {
 463                                @refs = ();
 464                                my ($url_, $rev_, $uuid_, $gs_) =
 465                                              working_head_info($head, \@refs);
 466                                my ($linear_refs_, $parents_) =
 467                                              linearize_history($gs_, \@refs);
 468                                if (scalar(@$linear_refs) !=
 469                                    scalar(@$linear_refs_)) {
 470                                        fatal "# of revisions changed ",
 471                                          "\nbefore:\n",
 472                                          join("\n", @$linear_refs),
 473                                          "\n\nafter:\n",
 474                                          join("\n", @$linear_refs_), "\n",
 475                                          'If you are attempting to commit ',
 476                                          "merges, try running:\n\t",
 477                                          'git rebase --interactive',
 478                                          '--preserve-merges ',
 479                                          $gs->refname,
 480                                          "\nBefore dcommitting";
 481                                }
 482                                if ($url_ ne $url) {
 483                                        fatal "URL mismatch after rebase: ",
 484                                              "$url_ != $url";
 485                                }
 486                                if ($uuid_ ne $uuid) {
 487                                        fatal "uuid mismatch after rebase: ",
 488                                              "$uuid_ != $uuid";
 489                                }
 490                                # remap parents
 491                                my (%p, @l, $i);
 492                                for ($i = 0; $i < scalar @$linear_refs; $i++) {
 493                                        my $new = $linear_refs_->[$i] or next;
 494                                        $p{$new} =
 495                                                $parents->{$linear_refs->[$i]};
 496                                        push @l, $new;
 497                                }
 498                                $parents = \%p;
 499                                $linear_refs = \@l;
 500                        }
 501                        $last_rev = $cmt_rev;
 502                }
 503        }
 504}
 505
 506sub cmd_find_rev {
 507        my $revision_or_hash = shift;
 508        my $result;
 509        if ($revision_or_hash =~ /^r\d+$/) {
 510                my $head = shift;
 511                $head ||= 'HEAD';
 512                my @refs;
 513                my (undef, undef, undef, $gs) = working_head_info($head, \@refs);
 514                unless ($gs) {
 515                        die "Unable to determine upstream SVN information from ",
 516                            "$head history\n";
 517                }
 518                my $desired_revision = substr($revision_or_hash, 1);
 519                $result = $gs->rev_db_get($desired_revision);
 520        } else {
 521                my (undef, $rev, undef) = cmt_metadata($revision_or_hash);
 522                $result = $rev;
 523        }
 524        print "$result\n" if $result;
 525}
 526
 527sub cmd_rebase {
 528        command_noisy(qw/update-index --refresh/);
 529        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 530        unless ($gs) {
 531                die "Unable to determine upstream SVN information from ",
 532                    "working tree history\n";
 533        }
 534        if (command(qw/diff-index HEAD --/)) {
 535                print STDERR "Cannot rebase with uncommited changes:\n";
 536                command_noisy('status');
 537                exit 1;
 538        }
 539        unless ($_local) {
 540                $_fetch_all ? $gs->fetch_all : $gs->fetch;
 541        }
 542        command_noisy(rebase_cmd(), $gs->refname);
 543}
 544
 545sub cmd_show_ignore {
 546        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 547        $gs ||= Git::SVN->new;
 548        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 549        $gs->prop_walk($gs->{path}, $r, sub {
 550                my ($gs, $path, $props) = @_;
 551                print STDOUT "\n# $path\n";
 552                my $s = $props->{'svn:ignore'} or return;
 553                $s =~ s/[\r\n]+/\n/g;
 554                chomp $s;
 555                $s =~ s#^#$path#gm;
 556                print STDOUT "$s\n";
 557        });
 558}
 559
 560sub cmd_create_ignore {
 561        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 562        $gs ||= Git::SVN->new;
 563        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 564        $gs->prop_walk($gs->{path}, $r, sub {
 565                my ($gs, $path, $props) = @_;
 566                # $path is of the form /path/to/dir/
 567                my $ignore = '.' . $path . '.gitignore';
 568                my $s = $props->{'svn:ignore'} or return;
 569                open(GITIGNORE, '>', $ignore)
 570                  or fatal("Failed to open `$ignore' for writing: $!");
 571                $s =~ s/[\r\n]+/\n/g;
 572                chomp $s;
 573                # Prefix all patterns so that the ignore doesn't apply
 574                # to sub-directories.
 575                $s =~ s#^#/#gm;
 576                print GITIGNORE "$s\n";
 577                close(GITIGNORE)
 578                  or fatal("Failed to close `$ignore': $!");
 579                command_noisy('add', $ignore);
 580        });
 581}
 582
 583# get_svnprops(PATH)
 584# ------------------
 585# Helper for cmd_propget and cmd_proplist below.
 586sub get_svnprops {
 587        my $path = shift;
 588        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 589        $gs ||= Git::SVN->new;
 590
 591        # prefix THE PATH by the sub-directory from which the user
 592        # invoked us.
 593        $path = $cmd_dir_prefix . $path;
 594        fatal("No such file or directory: $path") unless -e $path;
 595        my $is_dir = -d $path ? 1 : 0;
 596        $path = $gs->{path} . '/' . $path;
 597
 598        # canonicalize the path (otherwise libsvn will abort or fail to
 599        # find the file)
 600        # File::Spec->canonpath doesn't collapse x/../y into y (for a
 601        # good reason), so let's do this manually.
 602        $path =~ s#/+#/#g;
 603        $path =~ s#/\.(?:/|$)#/#g;
 604        $path =~ s#/[^/]+/\.\.##g;
 605        $path =~ s#/$##g;
 606
 607        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 608        my $props;
 609        if ($is_dir) {
 610                (undef, undef, $props) = $gs->ra->get_dir($path, $r);
 611        }
 612        else {
 613                (undef, $props) = $gs->ra->get_file($path, $r, undef);
 614        }
 615        return $props;
 616}
 617
 618# cmd_propget (PROP, PATH)
 619# ------------------------
 620# Print the SVN property PROP for PATH.
 621sub cmd_propget {
 622        my ($prop, $path) = @_;
 623        $path = '.' if not defined $path;
 624        usage(1) if not defined $prop;
 625        my $props = get_svnprops($path);
 626        if (not defined $props->{$prop}) {
 627                fatal("`$path' does not have a `$prop' SVN property.");
 628        }
 629        print $props->{$prop} . "\n";
 630}
 631
 632# cmd_proplist (PATH)
 633# -------------------
 634# Print the list of SVN properties for PATH.
 635sub cmd_proplist {
 636        my $path = shift;
 637        $path = '.' if not defined $path;
 638        my $props = get_svnprops($path);
 639        print "Properties on '$path':\n";
 640        foreach (sort keys %{$props}) {
 641                print "  $_\n";
 642        }
 643}
 644
 645sub cmd_multi_init {
 646        my $url = shift;
 647        unless (defined $_trunk || defined $_branches || defined $_tags) {
 648                usage(1);
 649        }
 650
 651        # there are currently some bugs that prevent multi-init/multi-fetch
 652        # setups from working well without this.
 653        $Git::SVN::_minimize_url = 1;
 654
 655        $_prefix = '' unless defined $_prefix;
 656        if (defined $url) {
 657                $url =~ s#/+$##;
 658                init_subdir(@_);
 659        }
 660        do_git_init_db();
 661        if (defined $_trunk) {
 662                my $trunk_ref = $_prefix . 'trunk';
 663                # try both old-style and new-style lookups:
 664                my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
 665                unless ($gs_trunk) {
 666                        my ($trunk_url, $trunk_path) =
 667                                              complete_svn_url($url, $_trunk);
 668                        $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
 669                                                   undef, $trunk_ref);
 670                }
 671        }
 672        return unless defined $_branches || defined $_tags;
 673        my $ra = $url ? Git::SVN::Ra->new($url) : undef;
 674        complete_url_ls_init($ra, $_branches, '--branches/-b', $_prefix);
 675        complete_url_ls_init($ra, $_tags, '--tags/-t', $_prefix . 'tags/');
 676}
 677
 678sub cmd_multi_fetch {
 679        my $remotes = Git::SVN::read_all_remotes();
 680        foreach my $repo_id (sort keys %$remotes) {
 681                if ($remotes->{$repo_id}->{url}) {
 682                        Git::SVN::fetch_all($repo_id, $remotes);
 683                }
 684        }
 685}
 686
 687# this command is special because it requires no metadata
 688sub cmd_commit_diff {
 689        my ($ta, $tb, $url) = @_;
 690        my $usage = "Usage: $0 commit-diff -r<revision> ".
 691                    "<tree-ish> <tree-ish> [<URL>]";
 692        fatal($usage) if (!defined $ta || !defined $tb);
 693        my $svn_path;
 694        if (!defined $url) {
 695                my $gs = eval { Git::SVN->new };
 696                if (!$gs) {
 697                        fatal("Needed URL or usable git-svn --id in ",
 698                              "the command-line\n", $usage);
 699                }
 700                $url = $gs->{url};
 701                $svn_path = $gs->{path};
 702        }
 703        unless (defined $_revision) {
 704                fatal("-r|--revision is a required argument\n", $usage);
 705        }
 706        if (defined $_message && defined $_file) {
 707                fatal("Both --message/-m and --file/-F specified ",
 708                      "for the commit message.\n",
 709                      "I have no idea what you mean");
 710        }
 711        if (defined $_file) {
 712                $_message = file_to_s($_file);
 713        } else {
 714                $_message ||= get_commit_entry($tb)->{log};
 715        }
 716        my $ra ||= Git::SVN::Ra->new($url);
 717        $svn_path ||= $ra->{svn_path};
 718        my $r = $_revision;
 719        if ($r eq 'HEAD') {
 720                $r = $ra->get_latest_revnum;
 721        } elsif ($r !~ /^\d+$/) {
 722                die "revision argument: $r not understood by git-svn\n";
 723        }
 724        my %ed_opts = ( r => $r,
 725                        log => $_message,
 726                        ra => $ra,
 727                        tree_a => $ta,
 728                        tree_b => $tb,
 729                        editor_cb => sub { print "Committed r$_[0]\n" },
 730                        svn_path => $svn_path );
 731        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
 732                print "No changes\n$ta == $tb\n";
 733        }
 734}
 735
 736########################### utility functions #########################
 737
 738sub rebase_cmd {
 739        my @cmd = qw/rebase/;
 740        push @cmd, '-v' if $_verbose;
 741        push @cmd, qw/--merge/ if $_merge;
 742        push @cmd, "--strategy=$_strategy" if $_strategy;
 743        @cmd;
 744}
 745
 746sub post_fetch_checkout {
 747        return if $_no_checkout;
 748        my $gs = $Git::SVN::_head or return;
 749        return if verify_ref('refs/heads/master^0');
 750
 751        my $valid_head = verify_ref('HEAD^0');
 752        command_noisy(qw(update-ref refs/heads/master), $gs->refname);
 753        return if ($valid_head || !verify_ref('HEAD^0'));
 754
 755        return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
 756        my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
 757        return if -f $index;
 758
 759        return if command_oneline(qw/rev-parse --is-inside-work-tree/) eq 'false';
 760        return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
 761        command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
 762        print STDERR "Checked out HEAD:\n  ",
 763                     $gs->full_url, " r", $gs->last_rev, "\n";
 764}
 765
 766sub complete_svn_url {
 767        my ($url, $path) = @_;
 768        $path =~ s#/+$##;
 769        if ($path !~ m#^[a-z\+]+://#) {
 770                if (!defined $url || $url !~ m#^[a-z\+]+://#) {
 771                        fatal("E: '$path' is not a complete URL ",
 772                              "and a separate URL is not specified");
 773                }
 774                return ($url, $path);
 775        }
 776        return ($path, '');
 777}
 778
 779sub complete_url_ls_init {
 780        my ($ra, $repo_path, $switch, $pfx) = @_;
 781        unless ($repo_path) {
 782                print STDERR "W: $switch not specified\n";
 783                return;
 784        }
 785        $repo_path =~ s#/+$##;
 786        if ($repo_path =~ m#^[a-z\+]+://#) {
 787                $ra = Git::SVN::Ra->new($repo_path);
 788                $repo_path = '';
 789        } else {
 790                $repo_path =~ s#^/+##;
 791                unless ($ra) {
 792                        fatal("E: '$repo_path' is not a complete URL ",
 793                              "and a separate URL is not specified");
 794                }
 795        }
 796        my $url = $ra->{url};
 797        my $gs = Git::SVN->init($url, undef, undef, undef, 1);
 798        my $k = "svn-remote.$gs->{repo_id}.url";
 799        my $orig_url = eval { command_oneline(qw/config --get/, $k) };
 800        if ($orig_url && ($orig_url ne $gs->{url})) {
 801                die "$k already set: $orig_url\n",
 802                    "wanted to set to: $gs->{url}\n";
 803        }
 804        command_oneline('config', $k, $gs->{url}) unless $orig_url;
 805        my $remote_path = "$ra->{svn_path}/$repo_path/*";
 806        $remote_path =~ s#/+#/#g;
 807        $remote_path =~ s#^/##g;
 808        my ($n) = ($switch =~ /^--(\w+)/);
 809        if (length $pfx && $pfx !~ m#/$#) {
 810                die "--prefix='$pfx' must have a trailing slash '/'\n";
 811        }
 812        command_noisy('config', "svn-remote.$gs->{repo_id}.$n",
 813                                "$remote_path:refs/remotes/$pfx*");
 814}
 815
 816sub verify_ref {
 817        my ($ref) = @_;
 818        eval { command_oneline([ 'rev-parse', '--verify', $ref ],
 819                               { STDERR => 0 }); };
 820}
 821
 822sub get_tree_from_treeish {
 823        my ($treeish) = @_;
 824        # $treeish can be a symbolic ref, too:
 825        my $type = command_oneline(qw/cat-file -t/, $treeish);
 826        my $expected;
 827        while ($type eq 'tag') {
 828                ($treeish, $type) = command(qw/cat-file tag/, $treeish);
 829        }
 830        if ($type eq 'commit') {
 831                $expected = (grep /^tree /, command(qw/cat-file commit/,
 832                                                    $treeish))[0];
 833                ($expected) = ($expected =~ /^tree ($sha1)$/o);
 834                die "Unable to get tree from $treeish\n" unless $expected;
 835        } elsif ($type eq 'tree') {
 836                $expected = $treeish;
 837        } else {
 838                die "$treeish is a $type, expected tree, tag or commit\n";
 839        }
 840        return $expected;
 841}
 842
 843sub get_commit_entry {
 844        my ($treeish) = shift;
 845        my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
 846        my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
 847        my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
 848        open my $log_fh, '>', $commit_editmsg or croak $!;
 849
 850        my $type = command_oneline(qw/cat-file -t/, $treeish);
 851        if ($type eq 'commit' || $type eq 'tag') {
 852                my ($msg_fh, $ctx) = command_output_pipe('cat-file',
 853                                                         $type, $treeish);
 854                my $in_msg = 0;
 855                while (<$msg_fh>) {
 856                        if (!$in_msg) {
 857                                $in_msg = 1 if (/^\s*$/);
 858                        } elsif (/^git-svn-id: /) {
 859                                # skip this for now, we regenerate the
 860                                # correct one on re-fetch anyways
 861                                # TODO: set *:merge properties or like...
 862                        } else {
 863                                print $log_fh $_ or croak $!;
 864                        }
 865                }
 866                command_close_pipe($msg_fh, $ctx);
 867        }
 868        close $log_fh or croak $!;
 869
 870        if ($_edit || ($type eq 'tree')) {
 871                my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
 872                # TODO: strip out spaces, comments, like git-commit.sh
 873                system($editor, $commit_editmsg);
 874        }
 875        rename $commit_editmsg, $commit_msg or croak $!;
 876        open $log_fh, '<', $commit_msg or croak $!;
 877        { local $/; chomp($log_entry{log} = <$log_fh>); }
 878        close $log_fh or croak $!;
 879        unlink $commit_msg;
 880        \%log_entry;
 881}
 882
 883sub s_to_file {
 884        my ($str, $file, $mode) = @_;
 885        open my $fd,'>',$file or croak $!;
 886        print $fd $str,"\n" or croak $!;
 887        close $fd or croak $!;
 888        chmod ($mode &~ umask, $file) if (defined $mode);
 889}
 890
 891sub file_to_s {
 892        my $file = shift;
 893        open my $fd,'<',$file or croak "$!: file: $file\n";
 894        local $/;
 895        my $ret = <$fd>;
 896        close $fd or croak $!;
 897        $ret =~ s/\s*$//s;
 898        return $ret;
 899}
 900
 901# '<svn username> = real-name <email address>' mapping based on git-svnimport:
 902sub load_authors {
 903        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
 904        my $log = $cmd eq 'log';
 905        while (<$authors>) {
 906                chomp;
 907                next unless /^(.+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
 908                my ($user, $name, $email) = ($1, $2, $3);
 909                if ($log) {
 910                        $Git::SVN::Log::rusers{"$name <$email>"} = $user;
 911                } else {
 912                        $users{$user} = [$name, $email];
 913                }
 914        }
 915        close $authors or croak $!;
 916}
 917
 918# convert GetOpt::Long specs for use by git-config
 919sub read_repo_config {
 920        return unless -d $ENV{GIT_DIR};
 921        my $opts = shift;
 922        my @config_only;
 923        foreach my $o (keys %$opts) {
 924                # if we have mixedCase and a long option-only, then
 925                # it's a config-only variable that we don't need for
 926                # the command-line.
 927                push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
 928                my $v = $opts->{$o};
 929                my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
 930                $key =~ s/-//g;
 931                my $arg = 'git-config';
 932                $arg .= ' --int' if ($o =~ /[:=]i$/);
 933                $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
 934                if (ref $v eq 'ARRAY') {
 935                        chomp(my @tmp = `$arg --get-all svn.$key`);
 936                        @$v = @tmp if @tmp;
 937                } else {
 938                        chomp(my $tmp = `$arg --get svn.$key`);
 939                        if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
 940                                $$v = $tmp;
 941                        }
 942                }
 943        }
 944        delete @$opts{@config_only} if @config_only;
 945}
 946
 947sub extract_metadata {
 948        my $id = shift or return (undef, undef, undef);
 949        my ($url, $rev, $uuid) = ($id =~ /^\s*git-svn-id:\s+(.*)\@(\d+)
 950                                                        \s([a-f\d\-]+)$/x);
 951        if (!defined $rev || !$uuid || !$url) {
 952                # some of the original repositories I made had
 953                # identifiers like this:
 954                ($rev, $uuid) = ($id =~/^\s*git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
 955        }
 956        return ($url, $rev, $uuid);
 957}
 958
 959sub cmt_metadata {
 960        return extract_metadata((grep(/^git-svn-id: /,
 961                command(qw/cat-file commit/, shift)))[-1]);
 962}
 963
 964sub working_head_info {
 965        my ($head, $refs) = @_;
 966        my @args = ('log', '--no-color', '--first-parent');
 967        my ($fh, $ctx) = command_output_pipe(@args, $head);
 968        my $hash;
 969        my %max;
 970        while (<$fh>) {
 971                if ( m{^commit ($::sha1)$} ) {
 972                        unshift @$refs, $hash if $hash and $refs;
 973                        $hash = $1;
 974                        next;
 975                }
 976                next unless s{^\s*(git-svn-id:)}{$1};
 977                my ($url, $rev, $uuid) = extract_metadata($_);
 978                if (defined $url && defined $rev) {
 979                        next if $max{$url} and $max{$url} < $rev;
 980                        if (my $gs = Git::SVN->find_by_url($url)) {
 981                                my $c = $gs->rev_db_get($rev);
 982                                if ($c && $c eq $hash) {
 983                                        close $fh; # break the pipe
 984                                        return ($url, $rev, $uuid, $gs);
 985                                } else {
 986                                        $max{$url} ||= $gs->rev_db_max;
 987                                }
 988                        }
 989                }
 990        }
 991        command_close_pipe($fh, $ctx);
 992        (undef, undef, undef, undef);
 993}
 994
 995sub read_commit_parents {
 996        my ($parents, $c) = @_;
 997        chomp(my $p = command_oneline(qw/rev-list --parents -1/, $c));
 998        $p =~ s/^($c)\s*// or die "rev-list --parents -1 $c failed!\n";
 999        @{$parents->{$c}} = split(/ /, $p);
1000}
1001
1002sub linearize_history {
1003        my ($gs, $refs) = @_;
1004        my %parents;
1005        foreach my $c (@$refs) {
1006                read_commit_parents(\%parents, $c);
1007        }
1008
1009        my @linear_refs;
1010        my %skip = ();
1011        my $last_svn_commit = $gs->last_commit;
1012        foreach my $c (reverse @$refs) {
1013                next if $c eq $last_svn_commit;
1014                last if $skip{$c};
1015
1016                unshift @linear_refs, $c;
1017                $skip{$c} = 1;
1018
1019                # we only want the first parent to diff against for linear
1020                # history, we save the rest to inject when we finalize the
1021                # svn commit
1022                my $fp_a = verify_ref("$c~1");
1023                my $fp_b = shift @{$parents{$c}} if $parents{$c};
1024                if (!$fp_a || !$fp_b) {
1025                        die "Commit $c\n",
1026                            "has no parent commit, and therefore ",
1027                            "nothing to diff against.\n",
1028                            "You should be working from a repository ",
1029                            "originally created by git-svn\n";
1030                }
1031                if ($fp_a ne $fp_b) {
1032                        die "$c~1 = $fp_a, however parsing commit $c ",
1033                            "revealed that:\n$c~1 = $fp_b\nBUG!\n";
1034                }
1035
1036                foreach my $p (@{$parents{$c}}) {
1037                        $skip{$p} = 1;
1038                }
1039        }
1040        (\@linear_refs, \%parents);
1041}
1042
1043package Git::SVN;
1044use strict;
1045use warnings;
1046use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
1047            $_repack $_repack_flags $_use_svm_props $_head
1048            $_use_svnsync_props $no_reuse_existing $_minimize_url/;
1049use Carp qw/croak/;
1050use File::Path qw/mkpath/;
1051use File::Copy qw/copy/;
1052use IPC::Open3;
1053
1054my $_repack_nr;
1055# properties that we do not log:
1056my %SKIP_PROP;
1057BEGIN {
1058        %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
1059                                        svn:special svn:executable
1060                                        svn:entry:committed-rev
1061                                        svn:entry:last-author
1062                                        svn:entry:uuid
1063                                        svn:entry:committed-date/;
1064
1065        # some options are read globally, but can be overridden locally
1066        # per [svn-remote "..."] section.  Command-line options will *NOT*
1067        # override options set in an [svn-remote "..."] section
1068        no strict 'refs';
1069        for my $option (qw/follow_parent no_metadata use_svm_props
1070                           use_svnsync_props/) {
1071                my $key = $option;
1072                $key =~ tr/_//d;
1073                my $prop = "-$option";
1074                *$option = sub {
1075                        my ($self) = @_;
1076                        return $self->{$prop} if exists $self->{$prop};
1077                        my $k = "svn-remote.$self->{repo_id}.$key";
1078                        eval { command_oneline(qw/config --get/, $k) };
1079                        if ($@) {
1080                                $self->{$prop} = ${"Git::SVN::_$option"};
1081                        } else {
1082                                my $v = command_oneline(qw/config --bool/,$k);
1083                                $self->{$prop} = $v eq 'false' ? 0 : 1;
1084                        }
1085                        return $self->{$prop};
1086                }
1087        }
1088}
1089
1090my %LOCKFILES;
1091END { unlink keys %LOCKFILES if %LOCKFILES }
1092
1093sub resolve_local_globs {
1094        my ($url, $fetch, $glob_spec) = @_;
1095        return unless defined $glob_spec;
1096        my $ref = $glob_spec->{ref};
1097        my $path = $glob_spec->{path};
1098        foreach (command(qw#for-each-ref --format=%(refname) refs/remotes#)) {
1099                next unless m#^refs/remotes/$ref->{regex}$#;
1100                my $p = $1;
1101                my $pathname = desanitize_refname($path->full_path($p));
1102                my $refname = desanitize_refname($ref->full_path($p));
1103                if (my $existing = $fetch->{$pathname}) {
1104                        if ($existing ne $refname) {
1105                                die "Refspec conflict:\n",
1106                                    "existing: refs/remotes/$existing\n",
1107                                    " globbed: refs/remotes/$refname\n";
1108                        }
1109                        my $u = (::cmt_metadata("refs/remotes/$refname"))[0];
1110                        $u =~ s!^\Q$url\E(/|$)!! or die
1111                          "refs/remotes/$refname: '$url' not found in '$u'\n";
1112                        if ($pathname ne $u) {
1113                                warn "W: Refspec glob conflict ",
1114                                     "(ref: refs/remotes/$refname):\n",
1115                                     "expected path: $pathname\n",
1116                                     "    real path: $u\n",
1117                                     "Continuing ahead with $u\n";
1118                                next;
1119                        }
1120                } else {
1121                        $fetch->{$pathname} = $refname;
1122                }
1123        }
1124}
1125
1126sub parse_revision_argument {
1127        my ($base, $head) = @_;
1128        if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
1129                return ($base, $head);
1130        }
1131        return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
1132        return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
1133        return ($head, $head) if ($::_revision eq 'HEAD');
1134        return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
1135        return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
1136        die "revision argument: $::_revision not understood by git-svn\n";
1137}
1138
1139sub fetch_all {
1140        my ($repo_id, $remotes) = @_;
1141        if (ref $repo_id) {
1142                my $gs = $repo_id;
1143                $repo_id = undef;
1144                $repo_id = $gs->{repo_id};
1145        }
1146        $remotes ||= read_all_remotes();
1147        my $remote = $remotes->{$repo_id} or
1148                     die "[svn-remote \"$repo_id\"] unknown\n";
1149        my $fetch = $remote->{fetch};
1150        my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
1151        my (@gs, @globs);
1152        my $ra = Git::SVN::Ra->new($url);
1153        my $uuid = $ra->get_uuid;
1154        my $head = $ra->get_latest_revnum;
1155        my $base = defined $fetch ? $head : 0;
1156
1157        # read the max revs for wildcard expansion (branches/*, tags/*)
1158        foreach my $t (qw/branches tags/) {
1159                defined $remote->{$t} or next;
1160                push @globs, $remote->{$t};
1161                my $max_rev = eval { tmp_config(qw/--int --get/,
1162                                         "svn-remote.$repo_id.${t}-maxRev") };
1163                if (defined $max_rev && ($max_rev < $base)) {
1164                        $base = $max_rev;
1165                } elsif (!defined $max_rev) {
1166                        $base = 0;
1167                }
1168        }
1169
1170        if ($fetch) {
1171                foreach my $p (sort keys %$fetch) {
1172                        my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
1173                        my $lr = $gs->rev_db_max;
1174                        if (defined $lr) {
1175                                $base = $lr if ($lr < $base);
1176                        }
1177                        push @gs, $gs;
1178                }
1179        }
1180
1181        ($base, $head) = parse_revision_argument($base, $head);
1182        $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
1183}
1184
1185sub read_all_remotes {
1186        my $r = {};
1187        foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
1188                if (m!^(.+)\.fetch=\s*(.*)\s*:\s*refs/remotes/(.+)\s*$!) {
1189                        my ($remote, $local_ref, $remote_ref) = ($1, $2, $3);
1190                        $local_ref =~ s{^/}{};
1191                        $r->{$remote}->{fetch}->{$local_ref} = $remote_ref;
1192                } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
1193                        $r->{$1}->{url} = $2;
1194                } elsif (m!^(.+)\.(branches|tags)=
1195                           (.*):refs/remotes/(.+)\s*$/!x) {
1196                        my ($p, $g) = ($3, $4);
1197                        my $rs = $r->{$1}->{$2} = {
1198                                          t => $2,
1199                                          remote => $1,
1200                                          path => Git::SVN::GlobSpec->new($p),
1201                                          ref => Git::SVN::GlobSpec->new($g) };
1202                        if (length($rs->{ref}->{right}) != 0) {
1203                                die "The '*' glob character must be the last ",
1204                                    "character of '$g'\n";
1205                        }
1206                }
1207        }
1208        $r;
1209}
1210
1211sub init_vars {
1212        if (defined $_repack) {
1213                $_repack = 1000 if ($_repack <= 0);
1214                $_repack_nr = $_repack;
1215                $_repack_flags ||= '-d';
1216        }
1217}
1218
1219sub verify_remotes_sanity {
1220        return unless -d $ENV{GIT_DIR};
1221        my %seen;
1222        foreach (command(qw/config -l/)) {
1223                if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
1224                        if ($seen{$1}) {
1225                                die "Remote ref refs/remote/$1 is tracked by",
1226                                    "\n  \"$_\"\nand\n  \"$seen{$1}\"\n",
1227                                    "Please resolve this ambiguity in ",
1228                                    "your git configuration file before ",
1229                                    "continuing\n";
1230                        }
1231                        $seen{$1} = $_;
1232                }
1233        }
1234}
1235
1236# we allow more chars than remotes2config.sh...
1237sub sanitize_remote_name {
1238        my ($name) = @_;
1239        $name =~ tr{A-Za-z0-9:,/+-}{.}c;
1240        $name;
1241}
1242
1243sub find_existing_remote {
1244        my ($url, $remotes) = @_;
1245        return undef if $no_reuse_existing;
1246        my $existing;
1247        foreach my $repo_id (keys %$remotes) {
1248                my $u = $remotes->{$repo_id}->{url} or next;
1249                next if $u ne $url;
1250                $existing = $repo_id;
1251                last;
1252        }
1253        $existing;
1254}
1255
1256sub init_remote_config {
1257        my ($self, $url, $no_write) = @_;
1258        $url =~ s!/+$!!; # strip trailing slash
1259        my $r = read_all_remotes();
1260        my $existing = find_existing_remote($url, $r);
1261        if ($existing) {
1262                unless ($no_write) {
1263                        print STDERR "Using existing ",
1264                                     "[svn-remote \"$existing\"]\n";
1265                }
1266                $self->{repo_id} = $existing;
1267        } elsif ($_minimize_url) {
1268                my $min_url = Git::SVN::Ra->new($url)->minimize_url;
1269                $existing = find_existing_remote($min_url, $r);
1270                if ($existing) {
1271                        unless ($no_write) {
1272                                print STDERR "Using existing ",
1273                                             "[svn-remote \"$existing\"]\n";
1274                        }
1275                        $self->{repo_id} = $existing;
1276                }
1277                if ($min_url ne $url) {
1278                        unless ($no_write) {
1279                                print STDERR "Using higher level of URL: ",
1280                                             "$url => $min_url\n";
1281                        }
1282                        my $old_path = $self->{path};
1283                        $self->{path} = $url;
1284                        $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
1285                        if (length $old_path) {
1286                                $self->{path} .= "/$old_path";
1287                        }
1288                        $url = $min_url;
1289                }
1290        }
1291        my $orig_url;
1292        if (!$existing) {
1293                # verify that we aren't overwriting anything:
1294                $orig_url = eval {
1295                        command_oneline('config', '--get',
1296                                        "svn-remote.$self->{repo_id}.url")
1297                };
1298                if ($orig_url && ($orig_url ne $url)) {
1299                        die "svn-remote.$self->{repo_id}.url already set: ",
1300                            "$orig_url\nwanted to set to: $url\n";
1301                }
1302        }
1303        my ($xrepo_id, $xpath) = find_ref($self->refname);
1304        if (defined $xpath) {
1305                die "svn-remote.$xrepo_id.fetch already set to track ",
1306                    "$xpath:refs/remotes/", $self->refname, "\n";
1307        }
1308        unless ($no_write) {
1309                command_noisy('config',
1310                              "svn-remote.$self->{repo_id}.url", $url);
1311                $self->{path} =~ s{^/}{};
1312                command_noisy('config', '--add',
1313                              "svn-remote.$self->{repo_id}.fetch",
1314                              "$self->{path}:".$self->refname);
1315        }
1316        $self->{url} = $url;
1317}
1318
1319sub find_by_url { # repos_root and, path are optional
1320        my ($class, $full_url, $repos_root, $path) = @_;
1321
1322        return undef unless defined $full_url;
1323        remove_username($full_url);
1324        remove_username($repos_root) if defined $repos_root;
1325        my $remotes = read_all_remotes();
1326        if (defined $full_url && defined $repos_root && !defined $path) {
1327                $path = $full_url;
1328                $path =~ s#^\Q$repos_root\E(?:/|$)##;
1329        }
1330        foreach my $repo_id (keys %$remotes) {
1331                my $u = $remotes->{$repo_id}->{url} or next;
1332                remove_username($u);
1333                next if defined $repos_root && $repos_root ne $u;
1334
1335                my $fetch = $remotes->{$repo_id}->{fetch} || {};
1336                foreach (qw/branches tags/) {
1337                        resolve_local_globs($u, $fetch,
1338                                            $remotes->{$repo_id}->{$_});
1339                }
1340                my $p = $path;
1341                unless (defined $p) {
1342                        $p = $full_url;
1343                        $p =~ s#^\Q$u\E(?:/|$)## or next;
1344                }
1345                foreach my $f (keys %$fetch) {
1346                        next if $f ne $p;
1347                        return Git::SVN->new($fetch->{$f}, $repo_id, $f);
1348                }
1349        }
1350        undef;
1351}
1352
1353sub init {
1354        my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
1355        my $self = _new($class, $repo_id, $ref_id, $path);
1356        if (defined $url) {
1357                $self->init_remote_config($url, $no_write);
1358        }
1359        $self;
1360}
1361
1362sub find_ref {
1363        my ($ref_id) = @_;
1364        foreach (command(qw/config -l/)) {
1365                next unless m!^svn-remote\.(.+)\.fetch=
1366                              \s*(.*)\s*:\s*refs/remotes/(.+)\s*$!x;
1367                my ($repo_id, $path, $ref) = ($1, $2, $3);
1368                if ($ref eq $ref_id) {
1369                        $path = '' if ($path =~ m#^\./?#);
1370                        return ($repo_id, $path);
1371                }
1372        }
1373        (undef, undef, undef);
1374}
1375
1376sub new {
1377        my ($class, $ref_id, $repo_id, $path) = @_;
1378        if (defined $ref_id && !defined $repo_id && !defined $path) {
1379                ($repo_id, $path) = find_ref($ref_id);
1380                if (!defined $repo_id) {
1381                        die "Could not find a \"svn-remote.*.fetch\" key ",
1382                            "in the repository configuration matching: ",
1383                            "refs/remotes/$ref_id\n";
1384                }
1385        }
1386        my $self = _new($class, $repo_id, $ref_id, $path);
1387        if (!defined $self->{path} || !length $self->{path}) {
1388                my $fetch = command_oneline('config', '--get',
1389                                            "svn-remote.$repo_id.fetch",
1390                                            ":refs/remotes/$ref_id\$") or
1391                     die "Failed to read \"svn-remote.$repo_id.fetch\" ",
1392                         "\":refs/remotes/$ref_id\$\" in config\n";
1393                ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
1394        }
1395        $self->{url} = command_oneline('config', '--get',
1396                                       "svn-remote.$repo_id.url") or
1397                  die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
1398        $self->rebuild;
1399        $self;
1400}
1401
1402sub refname {
1403        my ($refname) = "refs/remotes/$_[0]->{ref_id}" ;
1404
1405        # It cannot end with a slash /, we'll throw up on this because
1406        # SVN can't have directories with a slash in their name, either:
1407        if ($refname =~ m{/$}) {
1408                die "ref: '$refname' ends with a trailing slash, this is ",
1409                    "not permitted by git nor Subversion\n";
1410        }
1411
1412        # It cannot have ASCII control character space, tilde ~, caret ^,
1413        # colon :, question-mark ?, asterisk *, space, or open bracket [
1414        # anywhere.
1415        #
1416        # Additionally, % must be escaped because it is used for escaping
1417        # and we want our escaped refname to be reversible
1418        $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
1419
1420        # no slash-separated component can begin with a dot .
1421        # /.* becomes /%2E*
1422        $refname =~ s{/\.}{/%2E}g;
1423
1424        # It cannot have two consecutive dots .. anywhere
1425        # .. becomes %2E%2E
1426        $refname =~ s{\.\.}{%2E%2E}g;
1427
1428        return $refname;
1429}
1430
1431sub desanitize_refname {
1432        my ($refname) = @_;
1433        $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
1434        return $refname;
1435}
1436
1437sub svm_uuid {
1438        my ($self) = @_;
1439        return $self->{svm}->{uuid} if $self->svm;
1440        $self->ra;
1441        unless ($self->{svm}) {
1442                die "SVM UUID not cached, and reading remotely failed\n";
1443        }
1444        $self->{svm}->{uuid};
1445}
1446
1447sub svm {
1448        my ($self) = @_;
1449        return $self->{svm} if $self->{svm};
1450        my $svm;
1451        # see if we have it in our config, first:
1452        eval {
1453                my $section = "svn-remote.$self->{repo_id}";
1454                $svm = {
1455                  source => tmp_config('--get', "$section.svm-source"),
1456                  uuid => tmp_config('--get', "$section.svm-uuid"),
1457                  replace => tmp_config('--get', "$section.svm-replace"),
1458                }
1459        };
1460        if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
1461                $self->{svm} = $svm;
1462        }
1463        $self->{svm};
1464}
1465
1466sub _set_svm_vars {
1467        my ($self, $ra) = @_;
1468        return $ra if $self->svm;
1469
1470        my @err = ( "useSvmProps set, but failed to read SVM properties\n",
1471                    "(svm:source, svm:uuid) ",
1472                    "from the following URLs:\n" );
1473        sub read_svm_props {
1474                my ($self, $ra, $path, $r) = @_;
1475                my $props = ($ra->get_dir($path, $r))[2];
1476                my $src = $props->{'svm:source'};
1477                my $uuid = $props->{'svm:uuid'};
1478                return undef if (!$src || !$uuid);
1479
1480                chomp($src, $uuid);
1481
1482                $uuid =~ m{^[0-9a-f\-]{30,}$}
1483                    or die "doesn't look right - svm:uuid is '$uuid'\n";
1484
1485                # the '!' is used to mark the repos_root!/relative/path
1486                $src =~ s{/?!/?}{/};
1487                $src =~ s{/+$}{}; # no trailing slashes please
1488                # username is of no interest
1489                $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
1490
1491                my $replace = $ra->{url};
1492                $replace .= "/$path" if length $path;
1493
1494                my $section = "svn-remote.$self->{repo_id}";
1495                tmp_config("$section.svm-source", $src);
1496                tmp_config("$section.svm-replace", $replace);
1497                tmp_config("$section.svm-uuid", $uuid);
1498                $self->{svm} = {
1499                        source => $src,
1500                        uuid => $uuid,
1501                        replace => $replace
1502                };
1503        }
1504
1505        my $r = $ra->get_latest_revnum;
1506        my $path = $self->{path};
1507        my %tried;
1508        while (length $path) {
1509                unless ($tried{"$self->{url}/$path"}) {
1510                        return $ra if $self->read_svm_props($ra, $path, $r);
1511                        $tried{"$self->{url}/$path"} = 1;
1512                }
1513                $path =~ s#/?[^/]+$##;
1514        }
1515        die "Path: '$path' should be ''\n" if $path ne '';
1516        return $ra if $self->read_svm_props($ra, $path, $r);
1517        $tried{"$self->{url}/$path"} = 1;
1518
1519        if ($ra->{repos_root} eq $self->{url}) {
1520                die @err, (map { "  $_\n" } keys %tried), "\n";
1521        }
1522
1523        # nope, make sure we're connected to the repository root:
1524        my $ok;
1525        my @tried_b;
1526        $path = $ra->{svn_path};
1527        $ra = Git::SVN::Ra->new($ra->{repos_root});
1528        while (length $path) {
1529                unless ($tried{"$ra->{url}/$path"}) {
1530                        $ok = $self->read_svm_props($ra, $path, $r);
1531                        last if $ok;
1532                        $tried{"$ra->{url}/$path"} = 1;
1533                }
1534                $path =~ s#/?[^/]+$##;
1535        }
1536        die "Path: '$path' should be ''\n" if $path ne '';
1537        $ok ||= $self->read_svm_props($ra, $path, $r);
1538        $tried{"$ra->{url}/$path"} = 1;
1539        if (!$ok) {
1540                die @err, (map { "  $_\n" } keys %tried), "\n";
1541        }
1542        Git::SVN::Ra->new($self->{url});
1543}
1544
1545sub svnsync {
1546        my ($self) = @_;
1547        return $self->{svnsync} if $self->{svnsync};
1548
1549        if ($self->no_metadata) {
1550                die "Can't have both 'noMetadata' and ",
1551                    "'useSvnsyncProps' options set!\n";
1552        }
1553        if ($self->rewrite_root) {
1554                die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
1555                    "options set!\n";
1556        }
1557
1558        my $svnsync;
1559        # see if we have it in our config, first:
1560        eval {
1561                my $section = "svn-remote.$self->{repo_id}";
1562                $svnsync = {
1563                  url => tmp_config('--get', "$section.svnsync-url"),
1564                  uuid => tmp_config('--get', "$section.svnsync-uuid"),
1565                }
1566        };
1567        if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
1568                return $self->{svnsync} = $svnsync;
1569        }
1570
1571        my $err = "useSvnsyncProps set, but failed to read " .
1572                  "svnsync property: svn:sync-from-";
1573        my $rp = $self->ra->rev_proplist(0);
1574
1575        my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
1576        $url =~ m{^[a-z\+]+://} or
1577                   die "doesn't look right - svn:sync-from-url is '$url'\n";
1578
1579        my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
1580        $uuid =~ m{^[0-9a-f\-]{30,}$} or
1581                   die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
1582
1583        my $section = "svn-remote.$self->{repo_id}";
1584        tmp_config('--add', "$section.svnsync-uuid", $uuid);
1585        tmp_config('--add', "$section.svnsync-url", $url);
1586        return $self->{svnsync} = { url => $url, uuid => $uuid };
1587}
1588
1589# this allows us to memoize our SVN::Ra UUID locally and avoid a
1590# remote lookup (useful for 'git svn log').
1591sub ra_uuid {
1592        my ($self) = @_;
1593        unless ($self->{ra_uuid}) {
1594                my $key = "svn-remote.$self->{repo_id}.uuid";
1595                my $uuid = eval { tmp_config('--get', $key) };
1596                if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/) {
1597                        $self->{ra_uuid} = $uuid;
1598                } else {
1599                        die "ra_uuid called without URL\n" unless $self->{url};
1600                        $self->{ra_uuid} = $self->ra->get_uuid;
1601                        tmp_config('--add', $key, $self->{ra_uuid});
1602                }
1603        }
1604        $self->{ra_uuid};
1605}
1606
1607sub ra {
1608        my ($self) = shift;
1609        my $ra = Git::SVN::Ra->new($self->{url});
1610        if ($self->use_svm_props && !$self->{svm}) {
1611                if ($self->no_metadata) {
1612                        die "Can't have both 'noMetadata' and ",
1613                            "'useSvmProps' options set!\n";
1614                } elsif ($self->use_svnsync_props) {
1615                        die "Can't have both 'useSvnsyncProps' and ",
1616                            "'useSvmProps' options set!\n";
1617                }
1618                $ra = $self->_set_svm_vars($ra);
1619                $self->{-want_revprops} = 1;
1620        }
1621        $ra;
1622}
1623
1624sub rel_path {
1625        my ($self) = @_;
1626        my $repos_root = $self->ra->{repos_root};
1627        return $self->{path} if ($self->{url} eq $repos_root);
1628        my $url = $self->{url} .
1629                  (length $self->{path} ? "/$self->{path}" : $self->{path});
1630        $url =~ s!^\Q$repos_root\E(?:/+|$)!!g;
1631        $url;
1632}
1633
1634# prop_walk(PATH, REV, SUB)
1635# -------------------------
1636# Recursively traverse PATH at revision REV and invoke SUB for each
1637# directory that contains a SVN property.  SUB will be invoked as
1638# follows:  &SUB(gs, path, props);  where `gs' is this instance of
1639# Git::SVN, `path' the path to the directory where the properties
1640# `props' were found.  The `path' will be relative to point of checkout,
1641# that is, if url://repo/trunk is the current Git branch, and that
1642# directory contains a sub-directory `d', SUB will be invoked with `/d/'
1643# as `path' (note the trailing `/').
1644sub prop_walk {
1645        my ($self, $path, $rev, $sub) = @_;
1646
1647        my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev);
1648        $path =~ s#^/*#/#g;
1649        my $p = $path;
1650        # Strip the irrelevant part of the path.
1651        $p =~ s#^/+\Q$self->{path}\E(/|$)#/#;
1652        # Ensure the path is terminated by a `/'.
1653        $p =~ s#/*$#/#;
1654
1655        # The properties contain all the internal SVN stuff nobody
1656        # (usually) cares about.
1657        my $interesting_props = 0;
1658        foreach (keys %{$props}) {
1659                # If it doesn't start with `svn:', it must be a
1660                # user-defined property.
1661                ++$interesting_props and next if $_ !~ /^svn:/;
1662                # FIXME: Fragile, if SVN adds new public properties,
1663                # this needs to be updated.
1664                ++$interesting_props if /^svn:(?:ignore|keywords|executable
1665                                                 |eol-style|mime-type
1666                                                 |externals|needs-lock)$/x;
1667        }
1668        &$sub($self, $p, $props) if $interesting_props;
1669
1670        foreach (sort keys %$dirent) {
1671                next if $dirent->{$_}->{kind} != $SVN::Node::dir;
1672                $self->prop_walk($path . '/' . $_, $rev, $sub);
1673        }
1674}
1675
1676sub last_rev { ($_[0]->last_rev_commit)[0] }
1677sub last_commit { ($_[0]->last_rev_commit)[1] }
1678
1679# returns the newest SVN revision number and newest commit SHA1
1680sub last_rev_commit {
1681        my ($self) = @_;
1682        if (defined $self->{last_rev} && defined $self->{last_commit}) {
1683                return ($self->{last_rev}, $self->{last_commit});
1684        }
1685        my $c = ::verify_ref($self->refname.'^0');
1686        if ($c && !$self->use_svm_props && !$self->no_metadata) {
1687                my $rev = (::cmt_metadata($c))[1];
1688                if (defined $rev) {
1689                        ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
1690                        return ($rev, $c);
1691                }
1692        }
1693        my $db_path = $self->db_path;
1694        unless (-e $db_path) {
1695                ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
1696                return (undef, undef);
1697        }
1698        my $offset = -41; # from tail
1699        my $rl;
1700        open my $fh, '<', $db_path or croak "$db_path not readable: $!\n";
1701        sysseek($fh, $offset, 2); # don't care for errors
1702        sysread($fh, $rl, 41) == 41 or return (undef, undef);
1703        chomp $rl;
1704        while (('0' x40) eq $rl && sysseek($fh, 0, 1) != 0) {
1705                $offset -= 41;
1706                sysseek($fh, $offset, 2); # don't care for errors
1707                sysread($fh, $rl, 41) == 41 or return (undef, undef);
1708                chomp $rl;
1709        }
1710        if ($c && $c ne $rl) {
1711                die "$db_path and ", $self->refname,
1712                    " inconsistent!:\n$c != $rl\n";
1713        }
1714        my $rev = sysseek($fh, 0, 1) or croak $!;
1715        $rev =  ($rev - 41) / 41;
1716        close $fh or croak $!;
1717        ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
1718        return ($rev, $c);
1719}
1720
1721sub get_fetch_range {
1722        my ($self, $min, $max) = @_;
1723        $max ||= $self->ra->get_latest_revnum;
1724        $min ||= $self->rev_db_max;
1725        (++$min, $max);
1726}
1727
1728sub tmp_config {
1729        my (@args) = @_;
1730        my $old_def_config = "$ENV{GIT_DIR}/svn/config";
1731        my $config = "$ENV{GIT_DIR}/svn/.metadata";
1732        if (! -f $config && -f $old_def_config) {
1733                rename $old_def_config, $config or
1734                       die "Failed rename $old_def_config => $config: $!\n";
1735        }
1736        my $old_config = $ENV{GIT_CONFIG};
1737        $ENV{GIT_CONFIG} = $config;
1738        $@ = undef;
1739        my @ret = eval {
1740                unless (-f $config) {
1741                        mkfile($config);
1742                        open my $fh, '>', $config or
1743                            die "Can't open $config: $!\n";
1744                        print $fh "; This file is used internally by ",
1745                                  "git-svn\n" or die
1746                                  "Couldn't write to $config: $!\n";
1747                        print $fh "; You should not have to edit it\n" or
1748                              die "Couldn't write to $config: $!\n";
1749                        close $fh or die "Couldn't close $config: $!\n";
1750                }
1751                command('config', @args);
1752        };
1753        my $err = $@;
1754        if (defined $old_config) {
1755                $ENV{GIT_CONFIG} = $old_config;
1756        } else {
1757                delete $ENV{GIT_CONFIG};
1758        }
1759        die $err if $err;
1760        wantarray ? @ret : $ret[0];
1761}
1762
1763sub tmp_index_do {
1764        my ($self, $sub) = @_;
1765        my $old_index = $ENV{GIT_INDEX_FILE};
1766        $ENV{GIT_INDEX_FILE} = $self->{index};
1767        $@ = undef;
1768        my @ret = eval {
1769                my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
1770                mkpath([$dir]) unless -d $dir;
1771                &$sub;
1772        };
1773        my $err = $@;
1774        if (defined $old_index) {
1775                $ENV{GIT_INDEX_FILE} = $old_index;
1776        } else {
1777                delete $ENV{GIT_INDEX_FILE};
1778        }
1779        die $err if $err;
1780        wantarray ? @ret : $ret[0];
1781}
1782
1783sub assert_index_clean {
1784        my ($self, $treeish) = @_;
1785
1786        $self->tmp_index_do(sub {
1787                command_noisy('read-tree', $treeish) unless -e $self->{index};
1788                my $x = command_oneline('write-tree');
1789                my ($y) = (command(qw/cat-file commit/, $treeish) =~
1790                           /^tree ($::sha1)/mo);
1791                return if $y eq $x;
1792
1793                warn "Index mismatch: $y != $x\nrereading $treeish\n";
1794                unlink $self->{index} or die "unlink $self->{index}: $!\n";
1795                command_noisy('read-tree', $treeish);
1796                $x = command_oneline('write-tree');
1797                if ($y ne $x) {
1798                        ::fatal "trees ($treeish) $y != $x\n",
1799                                "Something is seriously wrong...";
1800                }
1801        });
1802}
1803
1804sub get_commit_parents {
1805        my ($self, $log_entry) = @_;
1806        my (%seen, @ret, @tmp);
1807        # legacy support for 'set-tree'; this is only used by set_tree_cb:
1808        if (my $ip = $self->{inject_parents}) {
1809                if (my $commit = delete $ip->{$log_entry->{revision}}) {
1810                        push @tmp, $commit;
1811                }
1812        }
1813        if (my $cur = ::verify_ref($self->refname.'^0')) {
1814                push @tmp, $cur;
1815        }
1816        if (my $ipd = $self->{inject_parents_dcommit}) {
1817                if (my $commit = delete $ipd->{$log_entry->{revision}}) {
1818                        push @tmp, @$commit;
1819                }
1820        }
1821        push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
1822        while (my $p = shift @tmp) {
1823                next if $seen{$p};
1824                $seen{$p} = 1;
1825                push @ret, $p;
1826                # MAXPARENT is defined to 16 in commit-tree.c:
1827                last if @ret >= 16;
1828        }
1829        if (@tmp) {
1830                die "r$log_entry->{revision}: No room for parents:\n\t",
1831                    join("\n\t", @tmp), "\n";
1832        }
1833        @ret;
1834}
1835
1836sub rewrite_root {
1837        my ($self) = @_;
1838        return $self->{-rewrite_root} if exists $self->{-rewrite_root};
1839        my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
1840        my $rwr = eval { command_oneline(qw/config --get/, $k) };
1841        if ($rwr) {
1842                $rwr =~ s#/+$##;
1843                if ($rwr !~ m#^[a-z\+]+://#) {
1844                        die "$rwr is not a valid URL (key: $k)\n";
1845                }
1846        }
1847        $self->{-rewrite_root} = $rwr;
1848}
1849
1850sub metadata_url {
1851        my ($self) = @_;
1852        ($self->rewrite_root || $self->{url}) .
1853           (length $self->{path} ? '/' . $self->{path} : '');
1854}
1855
1856sub full_url {
1857        my ($self) = @_;
1858        $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
1859}
1860
1861sub do_git_commit {
1862        my ($self, $log_entry) = @_;
1863        my $lr = $self->last_rev;
1864        if (defined $lr && $lr >= $log_entry->{revision}) {
1865                die "Last fetched revision of ", $self->refname,
1866                    " was r$lr, but we are about to fetch: ",
1867                    "r$log_entry->{revision}!\n";
1868        }
1869        if (my $c = $self->rev_db_get($log_entry->{revision})) {
1870                croak "$log_entry->{revision} = $c already exists! ",
1871                      "Why are we refetching it?\n";
1872        }
1873        $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} = $log_entry->{name};
1874        $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} =
1875                                                          $log_entry->{email};
1876        $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
1877
1878        my $tree = $log_entry->{tree};
1879        if (!defined $tree) {
1880                $tree = $self->tmp_index_do(sub {
1881                                            command_oneline('write-tree') });
1882        }
1883        die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
1884
1885        my @exec = ('git-commit-tree', $tree);
1886        foreach ($self->get_commit_parents($log_entry)) {
1887                push @exec, '-p', $_;
1888        }
1889        defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
1890                                                                   or croak $!;
1891        print $msg_fh $log_entry->{log} or croak $!;
1892        unless ($self->no_metadata) {
1893                print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
1894                              or croak $!;
1895        }
1896        $msg_fh->flush == 0 or croak $!;
1897        close $msg_fh or croak $!;
1898        chomp(my $commit = do { local $/; <$out_fh> });
1899        close $out_fh or croak $!;
1900        waitpid $pid, 0;
1901        croak $? if $?;
1902        if ($commit !~ /^$::sha1$/o) {
1903                die "Failed to commit, invalid sha1: $commit\n";
1904        }
1905
1906        $self->rev_db_set($log_entry->{revision}, $commit, 1);
1907
1908        $self->{last_rev} = $log_entry->{revision};
1909        $self->{last_commit} = $commit;
1910        print "r$log_entry->{revision}";
1911        if (defined $log_entry->{svm_revision}) {
1912                 print " (\@$log_entry->{svm_revision})";
1913                 $self->rev_db_set($log_entry->{svm_revision}, $commit,
1914                                   0, $self->svm_uuid);
1915        }
1916        print " = $commit ($self->{ref_id})\n";
1917        if (defined $_repack && (--$_repack_nr == 0)) {
1918                $_repack_nr = $_repack;
1919                # repack doesn't use any arguments with spaces in them, does it?
1920                print "Running git repack $_repack_flags ...\n";
1921                command_noisy('repack', split(/\s+/, $_repack_flags));
1922                print "Done repacking\n";
1923        }
1924        return $commit;
1925}
1926
1927sub match_paths {
1928        my ($self, $paths, $r) = @_;
1929        return 1 if $self->{path} eq '';
1930        if (my $path = $paths->{"/$self->{path}"}) {
1931                return ($path->{action} eq 'D') ? 0 : 1;
1932        }
1933        $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
1934        if (grep /$self->{path_regex}/, keys %$paths) {
1935                return 1;
1936        }
1937        my $c = '';
1938        foreach (split m#/#, $self->{path}) {
1939                $c .= "/$_";
1940                next unless ($paths->{$c} &&
1941                             ($paths->{$c}->{action} =~ /^[AR]$/));
1942                if ($self->ra->check_path($self->{path}, $r) ==
1943                    $SVN::Node::dir) {
1944                        return 1;
1945                }
1946        }
1947        return 0;
1948}
1949
1950sub find_parent_branch {
1951        my ($self, $paths, $rev) = @_;
1952        return undef unless $self->follow_parent;
1953        unless (defined $paths) {
1954                my $err_handler = $SVN::Error::handler;
1955                $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
1956                $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1, sub {
1957                                   $paths =
1958                                      Git::SVN::Ra::dup_changed_paths($_[0]) });
1959                $SVN::Error::handler = $err_handler;
1960        }
1961        return undef unless defined $paths;
1962
1963        # look for a parent from another branch:
1964        my @b_path_components = split m#/#, $self->rel_path;
1965        my @a_path_components;
1966        my $i;
1967        while (@b_path_components) {
1968                $i = $paths->{'/'.join('/', @b_path_components)};
1969                last if $i && defined $i->{copyfrom_path};
1970                unshift(@a_path_components, pop(@b_path_components));
1971        }
1972        return undef unless defined $i && defined $i->{copyfrom_path};
1973        my $branch_from = $i->{copyfrom_path};
1974        if (@a_path_components) {
1975                print STDERR "branch_from: $branch_from => ";
1976                $branch_from .= '/'.join('/', @a_path_components);
1977                print STDERR $branch_from, "\n";
1978        }
1979        my $r = $i->{copyfrom_rev};
1980        my $repos_root = $self->ra->{repos_root};
1981        my $url = $self->ra->{url};
1982        my $new_url = $repos_root . $branch_from;
1983        print STDERR  "Found possible branch point: ",
1984                      "$new_url => ", $self->full_url, ", $r\n";
1985        $branch_from =~ s#^/##;
1986        my $gs = Git::SVN->find_by_url($new_url, $repos_root, $branch_from);
1987        unless ($gs) {
1988                my $ref_id = $self->{ref_id};
1989                $ref_id =~ s/\@\d+$//;
1990                $ref_id .= "\@$r";
1991                # just grow a tail if we're not unique enough :x
1992                $ref_id .= '-' while find_ref($ref_id);
1993                print STDERR "Initializing parent: $ref_id\n";
1994                $gs = Git::SVN->init($new_url, '', $ref_id, $ref_id, 1);
1995        }
1996        my ($r0, $parent) = $gs->find_rev_before($r, 1);
1997        if (!defined $r0 || !defined $parent) {
1998                my ($base, $head) = parse_revision_argument(0, $r);
1999                if ($base <= $r) {
2000                        $gs->fetch($base, $r);
2001                }
2002                ($r0, $parent) = $gs->last_rev_commit;
2003        }
2004        if (defined $r0 && defined $parent) {
2005                print STDERR "Found branch parent: ($self->{ref_id}) $parent\n";
2006                my $ed;
2007                if ($self->ra->can_do_switch) {
2008                        $self->assert_index_clean($parent);
2009                        print STDERR "Following parent with do_switch\n";
2010                        # do_switch works with svn/trunk >= r22312, but that
2011                        # is not included with SVN 1.4.3 (the latest version
2012                        # at the moment), so we can't rely on it
2013                        $self->{last_commit} = $parent;
2014                        $ed = SVN::Git::Fetcher->new($self);
2015                        $gs->ra->gs_do_switch($r0, $rev, $gs,
2016                                              $self->full_url, $ed)
2017                          or die "SVN connection failed somewhere...\n";
2018                } elsif ($self->ra->trees_match($new_url, $r0,
2019                                                $self->full_url, $rev)) {
2020                        print STDERR "Trees match:\n",
2021                                     "  $new_url\@$r0\n",
2022                                     "  ${\$self->full_url}\@$rev\n",
2023                                     "Following parent with no changes\n";
2024                        $self->tmp_index_do(sub {
2025                            command_noisy('read-tree', $parent);
2026                        });
2027                        $self->{last_commit} = $parent;
2028                } else {
2029                        print STDERR "Following parent with do_update\n";
2030                        $ed = SVN::Git::Fetcher->new($self);
2031                        $self->ra->gs_do_update($rev, $rev, $self, $ed)
2032                          or die "SVN connection failed somewhere...\n";
2033                }
2034                print STDERR "Successfully followed parent\n";
2035                return $self->make_log_entry($rev, [$parent], $ed);
2036        }
2037        return undef;
2038}
2039
2040sub do_fetch {
2041        my ($self, $paths, $rev) = @_;
2042        my $ed;
2043        my ($last_rev, @parents);
2044        if (my $lc = $self->last_commit) {
2045                # we can have a branch that was deleted, then re-added
2046                # under the same name but copied from another path, in
2047                # which case we'll have multiple parents (we don't
2048                # want to break the original ref, nor lose copypath info):
2049                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2050                        push @{$log_entry->{parents}}, $lc;
2051                        return $log_entry;
2052                }
2053                $ed = SVN::Git::Fetcher->new($self);
2054                $last_rev = $self->{last_rev};
2055                $ed->{c} = $lc;
2056                @parents = ($lc);
2057        } else {
2058                $last_rev = $rev;
2059                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2060                        return $log_entry;
2061                }
2062                $ed = SVN::Git::Fetcher->new($self);
2063        }
2064        unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
2065                die "SVN connection failed somewhere...\n";
2066        }
2067        $self->make_log_entry($rev, \@parents, $ed);
2068}
2069
2070sub get_untracked {
2071        my ($self, $ed) = @_;
2072        my @out;
2073        my $h = $ed->{empty};
2074        foreach (sort keys %$h) {
2075                my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
2076                push @out, "  $act: " . uri_encode($_);
2077                warn "W: $act: $_\n";
2078        }
2079        foreach my $t (qw/dir_prop file_prop/) {
2080                $h = $ed->{$t} or next;
2081                foreach my $path (sort keys %$h) {
2082                        my $ppath = $path eq '' ? '.' : $path;
2083                        foreach my $prop (sort keys %{$h->{$path}}) {
2084                                next if $SKIP_PROP{$prop};
2085                                my $v = $h->{$path}->{$prop};
2086                                my $t_ppath_prop = "$t: " .
2087                                                    uri_encode($ppath) . ' ' .
2088                                                    uri_encode($prop);
2089                                if (defined $v) {
2090                                        push @out, "  +$t_ppath_prop " .
2091                                                   uri_encode($v);
2092                                } else {
2093                                        push @out, "  -$t_ppath_prop";
2094                                }
2095                        }
2096                }
2097        }
2098        foreach my $t (qw/absent_file absent_directory/) {
2099                $h = $ed->{$t} or next;
2100                foreach my $parent (sort keys %$h) {
2101                        foreach my $path (sort @{$h->{$parent}}) {
2102                                push @out, "  $t: " .
2103                                           uri_encode("$parent/$path");
2104                                warn "W: $t: $parent/$path ",
2105                                     "Insufficient permissions?\n";
2106                        }
2107                }
2108        }
2109        \@out;
2110}
2111
2112sub parse_svn_date {
2113        my $date = shift || return '+0000 1970-01-01 00:00:00';
2114        my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2115                                            (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x) or
2116                                         croak "Unable to parse date: $date\n";
2117        "+0000 $Y-$m-$d $H:$M:$S";
2118}
2119
2120sub check_author {
2121        my ($author) = @_;
2122        if (!defined $author || length $author == 0) {
2123                $author = '(no author)';
2124        }
2125        if (defined $::_authors && ! defined $::users{$author}) {
2126                die "Author: $author not defined in $::_authors file\n";
2127        }
2128        $author;
2129}
2130
2131sub make_log_entry {
2132        my ($self, $rev, $parents, $ed) = @_;
2133        my $untracked = $self->get_untracked($ed);
2134
2135        open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
2136        print $un "r$rev\n" or croak $!;
2137        print $un $_, "\n" foreach @$untracked;
2138        my %log_entry = ( parents => $parents || [], revision => $rev,
2139                          log => '');
2140
2141        my $headrev;
2142        my $logged = delete $self->{logged_rev_props};
2143        if (!$logged || $self->{-want_revprops}) {
2144                my $rp = $self->ra->rev_proplist($rev);
2145                foreach (sort keys %$rp) {
2146                        my $v = $rp->{$_};
2147                        if (/^svn:(author|date|log)$/) {
2148                                $log_entry{$1} = $v;
2149                        } elsif ($_ eq 'svm:headrev') {
2150                                $headrev = $v;
2151                        } else {
2152                                print $un "  rev_prop: ", uri_encode($_), ' ',
2153                                          uri_encode($v), "\n";
2154                        }
2155                }
2156        } else {
2157                map { $log_entry{$_} = $logged->{$_} } keys %$logged;
2158        }
2159        close $un or croak $!;
2160
2161        $log_entry{date} = parse_svn_date($log_entry{date});
2162        $log_entry{log} .= "\n";
2163        my $author = $log_entry{author} = check_author($log_entry{author});
2164        my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
2165                                                       : ($author, undef);
2166        if (defined $headrev && $self->use_svm_props) {
2167                if ($self->rewrite_root) {
2168                        die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
2169                            "options set!\n";
2170                }
2171                my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$};
2172                # we don't want "SVM: initializing mirror for junk" ...
2173                return undef if $r == 0;
2174                my $svm = $self->svm;
2175                if ($uuid ne $svm->{uuid}) {
2176                        die "UUID mismatch on SVM path:\n",
2177                            "expected: $svm->{uuid}\n",
2178                            "     got: $uuid\n";
2179                }
2180                my $full_url = $self->full_url;
2181                $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
2182                             die "Failed to replace '$svm->{replace}' with ",
2183                                 "'$svm->{source}' in $full_url\n";
2184                # throw away username for storing in records
2185                remove_username($full_url);
2186                $log_entry{metadata} = "$full_url\@$r $uuid";
2187                $log_entry{svm_revision} = $r;
2188                $email ||= "$author\@$uuid"
2189        } elsif ($self->use_svnsync_props) {
2190                my $full_url = $self->svnsync->{url};
2191                $full_url .= "/$self->{path}" if length $self->{path};
2192                remove_username($full_url);
2193                my $uuid = $self->svnsync->{uuid};
2194                $log_entry{metadata} = "$full_url\@$rev $uuid";
2195                $email ||= "$author\@$uuid"
2196        } else {
2197                my $url = $self->metadata_url;
2198                remove_username($url);
2199                $log_entry{metadata} = "$url\@$rev " .
2200                                       $self->ra->get_uuid;
2201                $email ||= "$author\@" . $self->ra->get_uuid;
2202        }
2203        $log_entry{name} = $name;
2204        $log_entry{email} = $email;
2205        \%log_entry;
2206}
2207
2208sub fetch {
2209        my ($self, $min_rev, $max_rev, @parents) = @_;
2210        my ($last_rev, $last_commit) = $self->last_rev_commit;
2211        my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
2212        $self->ra->gs_fetch_loop_common($base, $head, [$self]);
2213}
2214
2215sub set_tree_cb {
2216        my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
2217        $self->{inject_parents} = { $rev => $tree };
2218        $self->fetch(undef, undef);
2219}
2220
2221sub set_tree {
2222        my ($self, $tree) = (shift, shift);
2223        my $log_entry = ::get_commit_entry($tree);
2224        unless ($self->{last_rev}) {
2225                fatal("Must have an existing revision to commit");
2226        }
2227        my %ed_opts = ( r => $self->{last_rev},
2228                        log => $log_entry->{log},
2229                        ra => $self->ra,
2230                        tree_a => $self->{last_commit},
2231                        tree_b => $tree,
2232                        editor_cb => sub {
2233                               $self->set_tree_cb($log_entry, $tree, @_) },
2234                        svn_path => $self->{path} );
2235        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
2236                print "No changes\nr$self->{last_rev} = $tree\n";
2237        }
2238}
2239
2240sub rebuild {
2241        my ($self) = @_;
2242        my $db_path = $self->db_path;
2243        return if (-e $db_path && ! -z $db_path);
2244        return unless ::verify_ref($self->refname.'^0');
2245        if (-f $self->{db_root}) {
2246                rename $self->{db_root}, $db_path or die
2247                     "rename $self->{db_root} => $db_path failed: $!\n";
2248                my ($dir, $base) = ($db_path =~ m#^(.*?)/?([^/]+)$#);
2249                symlink $base, $self->{db_root} or die
2250                     "symlink $base => $self->{db_root} failed: $!\n";
2251                return;
2252        }
2253        print "Rebuilding $db_path ...\n";
2254        my ($log, $ctx) = command_output_pipe("log", '--no-color', $self->refname);
2255        my $latest;
2256        my $full_url = $self->full_url;
2257        remove_username($full_url);
2258        my $svn_uuid;
2259        my $c;
2260        while (<$log>) {
2261                if ( m{^commit ($::sha1)$} ) {
2262                        $c = $1;
2263                        next;
2264                }
2265                next unless s{^\s*(git-svn-id:)}{$1};
2266                my ($url, $rev, $uuid) = ::extract_metadata($_);
2267                remove_username($url);
2268
2269                # ignore merges (from set-tree)
2270                next if (!defined $rev || !$uuid);
2271
2272                # if we merged or otherwise started elsewhere, this is
2273                # how we break out of it
2274                if ((defined $svn_uuid && ($uuid ne $svn_uuid)) ||
2275                    ($full_url && $url && ($url ne $full_url))) {
2276                        next;
2277                }
2278                $latest ||= $rev;
2279                $svn_uuid ||= $uuid;
2280
2281                $self->rev_db_set($rev, $c);
2282                print "r$rev = $c\n";
2283        }
2284        command_close_pipe($log, $ctx);
2285        print "Done rebuilding $db_path\n";
2286}
2287
2288# rev_db:
2289# Tie::File seems to be prone to offset errors if revisions get sparse,
2290# it's not that fast, either.  Tie::File is also not in Perl 5.6.  So
2291# one of my favorite modules is out :<  Next up would be one of the DBM
2292# modules, but I'm not sure which is most portable...  So I'll just
2293# go with something that's plain-text, but still capable of
2294# being randomly accessed.  So here's my ultra-simple fixed-width
2295# database.  All records are 40 characters + "\n", so it's easy to seek
2296# to a revision: (41 * rev) is the byte offset.
2297# A record of 40 0s denotes an empty revision.
2298# And yes, it's still pretty fast (faster than Tie::File).
2299# These files are disposable unless noMetadata or useSvmProps is set
2300
2301sub _rev_db_set {
2302        my ($fh, $rev, $commit) = @_;
2303        my $offset = $rev * 41;
2304        # assume that append is the common case:
2305        seek $fh, 0, 2 or croak $!;
2306        my $pos = tell $fh;
2307        if ($pos < $offset) {
2308                for (1 .. (($offset - $pos) / 41)) {
2309                        print $fh (('0' x 40),"\n") or croak $!;
2310                }
2311        }
2312        seek $fh, $offset, 0 or croak $!;
2313        print $fh $commit,"\n" or croak $!;
2314}
2315
2316sub mkfile {
2317        my ($path) = @_;
2318        unless (-e $path) {
2319                my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
2320                mkpath([$dir]) unless -d $dir;
2321                open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
2322                close $fh or die "Couldn't close (create) $path: $!\n";
2323        }
2324}
2325
2326sub rev_db_set {
2327        my ($self, $rev, $commit, $update_ref, $uuid) = @_;
2328        length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
2329        my $db = $self->db_path($uuid);
2330        my $db_lock = "$db.lock";
2331        my $sig;
2332        if ($update_ref) {
2333                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
2334                            $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
2335        }
2336        mkfile($db);
2337
2338        $LOCKFILES{$db_lock} = 1;
2339        my $sync;
2340        # both of these options make our .rev_db file very, very important
2341        # and we can't afford to lose it because rebuild() won't work
2342        if ($self->use_svm_props || $self->no_metadata) {
2343                $sync = 1;
2344                copy($db, $db_lock) or die "rev_db_set(@_): ",
2345                                           "Failed to copy: ",
2346                                           "$db => $db_lock ($!)\n";
2347        } else {
2348                rename $db, $db_lock or die "rev_db_set(@_): ",
2349                                            "Failed to rename: ",
2350                                            "$db => $db_lock ($!)\n";
2351        }
2352        open my $fh, '+<', $db_lock or die "Couldn't open $db_lock: $!\n";
2353        _rev_db_set($fh, $rev, $commit);
2354        if ($sync) {
2355                $fh->flush or die "Couldn't flush $db_lock: $!\n";
2356                $fh->sync or die "Couldn't sync $db_lock: $!\n";
2357        }
2358        close $fh or croak $!;
2359        if ($update_ref) {
2360                $_head = $self;
2361                command_noisy('update-ref', '-m', "r$rev",
2362                              $self->refname, $commit);
2363        }
2364        rename $db_lock, $db or die "rev_db_set(@_): ", "Failed to rename: ",
2365                                    "$db_lock => $db ($!)\n";
2366        delete $LOCKFILES{$db_lock};
2367        if ($update_ref) {
2368                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
2369                            $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
2370                kill $sig, $$ if defined $sig;
2371        }
2372}
2373
2374sub rev_db_max {
2375        my ($self) = @_;
2376        $self->rebuild;
2377        my $db_path = $self->db_path;
2378        my @stat = stat $db_path or return 0;
2379        ($stat[7] % 41) == 0 or die "$db_path inconsistent size: $stat[7]\n";
2380        my $max = $stat[7] / 41;
2381        (($max > 0) ? $max - 1 : 0);
2382}
2383
2384sub rev_db_get {
2385        my ($self, $rev, $uuid) = @_;
2386        my $ret;
2387        my $offset = $rev * 41;
2388        my $db_path = $self->db_path($uuid);
2389        return undef unless -e $db_path;
2390        open my $fh, '<', $db_path or croak $!;
2391        if (sysseek($fh, $offset, 0) == $offset) {
2392                my $read = sysread($fh, $ret, 40);
2393                $ret = undef if ($read != 40 || $ret eq ('0'x40));
2394        }
2395        close $fh or croak $!;
2396        $ret;
2397}
2398
2399sub find_rev_before {
2400        my ($self, $rev, $eq_ok) = @_;
2401        --$rev unless $eq_ok;
2402        while ($rev > 0) {
2403                if (my $c = $self->rev_db_get($rev)) {
2404                        return ($rev, $c);
2405                }
2406                --$rev;
2407        }
2408        return (undef, undef);
2409}
2410
2411sub _new {
2412        my ($class, $repo_id, $ref_id, $path) = @_;
2413        unless (defined $repo_id && length $repo_id) {
2414                $repo_id = $Git::SVN::default_repo_id;
2415        }
2416        unless (defined $ref_id && length $ref_id) {
2417                $_[2] = $ref_id = $Git::SVN::default_ref_id;
2418        }
2419        $_[1] = $repo_id = sanitize_remote_name($repo_id);
2420        my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
2421        $_[3] = $path = '' unless (defined $path);
2422        mkpath(["$ENV{GIT_DIR}/svn"]);
2423        bless {
2424                ref_id => $ref_id, dir => $dir, index => "$dir/index",
2425                path => $path, config => "$ENV{GIT_DIR}/svn/config",
2426                db_root => "$dir/.rev_db", repo_id => $repo_id }, $class;
2427}
2428
2429sub db_path {
2430        my ($self, $uuid) = @_;
2431        $uuid ||= $self->ra_uuid;
2432        "$self->{db_root}.$uuid";
2433}
2434
2435sub uri_encode {
2436        my ($f) = @_;
2437        $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
2438        $f
2439}
2440
2441sub remove_username {
2442        $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
2443}
2444
2445package Git::SVN::Prompt;
2446use strict;
2447use warnings;
2448require SVN::Core;
2449use vars qw/$_no_auth_cache $_username/;
2450
2451sub simple {
2452        my ($cred, $realm, $default_username, $may_save, $pool) = @_;
2453        $may_save = undef if $_no_auth_cache;
2454        $default_username = $_username if defined $_username;
2455        if (defined $default_username && length $default_username) {
2456                if (defined $realm && length $realm) {
2457                        print STDERR "Authentication realm: $realm\n";
2458                        STDERR->flush;
2459                }
2460                $cred->username($default_username);
2461        } else {
2462                username($cred, $realm, $may_save, $pool);
2463        }
2464        $cred->password(_read_password("Password for '" .
2465                                       $cred->username . "': ", $realm));
2466        $cred->may_save($may_save);
2467        $SVN::_Core::SVN_NO_ERROR;
2468}
2469
2470sub ssl_server_trust {
2471        my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
2472        $may_save = undef if $_no_auth_cache;
2473        print STDERR "Error validating server certificate for '$realm':\n";
2474        {
2475                no warnings 'once';
2476                # All variables SVN::Auth::SSL::* are used only once,
2477                # so we're shutting up Perl warnings about this.
2478                if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
2479                        print STDERR " - The certificate is not issued ",
2480                            "by a trusted authority. Use the\n",
2481                            "   fingerprint to validate ",
2482                            "the certificate manually!\n";
2483                }
2484                if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
2485                        print STDERR " - The certificate hostname ",
2486                            "does not match.\n";
2487                }
2488                if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
2489                        print STDERR " - The certificate is not yet valid.\n";
2490                }
2491                if ($failures & $SVN::Auth::SSL::EXPIRED) {
2492                        print STDERR " - The certificate has expired.\n";
2493                }
2494                if ($failures & $SVN::Auth::SSL::OTHER) {
2495                        print STDERR " - The certificate has ",
2496                            "an unknown error.\n";
2497                }
2498        } # no warnings 'once'
2499        printf STDERR
2500                "Certificate information:\n".
2501                " - Hostname: %s\n".
2502                " - Valid: from %s until %s\n".
2503                " - Issuer: %s\n".
2504                " - Fingerprint: %s\n",
2505                map $cert_info->$_, qw(hostname valid_from valid_until
2506                                       issuer_dname fingerprint);
2507        my $choice;
2508prompt:
2509        print STDERR $may_save ?
2510              "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
2511              "(R)eject or accept (t)emporarily? ";
2512        STDERR->flush;
2513        $choice = lc(substr(<STDIN> || 'R', 0, 1));
2514        if ($choice =~ /^t$/i) {
2515                $cred->may_save(undef);
2516        } elsif ($choice =~ /^r$/i) {
2517                return -1;
2518        } elsif ($may_save && $choice =~ /^p$/i) {
2519                $cred->may_save($may_save);
2520        } else {
2521                goto prompt;
2522        }
2523        $cred->accepted_failures($failures);
2524        $SVN::_Core::SVN_NO_ERROR;
2525}
2526
2527sub ssl_client_cert {
2528        my ($cred, $realm, $may_save, $pool) = @_;
2529        $may_save = undef if $_no_auth_cache;
2530        print STDERR "Client certificate filename: ";
2531        STDERR->flush;
2532        chomp(my $filename = <STDIN>);
2533        $cred->cert_file($filename);
2534        $cred->may_save($may_save);
2535        $SVN::_Core::SVN_NO_ERROR;
2536}
2537
2538sub ssl_client_cert_pw {
2539        my ($cred, $realm, $may_save, $pool) = @_;
2540        $may_save = undef if $_no_auth_cache;
2541        $cred->password(_read_password("Password: ", $realm));
2542        $cred->may_save($may_save);
2543        $SVN::_Core::SVN_NO_ERROR;
2544}
2545
2546sub username {
2547        my ($cred, $realm, $may_save, $pool) = @_;
2548        $may_save = undef if $_no_auth_cache;
2549        if (defined $realm && length $realm) {
2550                print STDERR "Authentication realm: $realm\n";
2551        }
2552        my $username;
2553        if (defined $_username) {
2554                $username = $_username;
2555        } else {
2556                print STDERR "Username: ";
2557                STDERR->flush;
2558                chomp($username = <STDIN>);
2559        }
2560        $cred->username($username);
2561        $cred->may_save($may_save);
2562        $SVN::_Core::SVN_NO_ERROR;
2563}
2564
2565sub _read_password {
2566        my ($prompt, $realm) = @_;
2567        print STDERR $prompt;
2568        STDERR->flush;
2569        require Term::ReadKey;
2570        Term::ReadKey::ReadMode('noecho');
2571        my $password = '';
2572        while (defined(my $key = Term::ReadKey::ReadKey(0))) {
2573                last if $key =~ /[\012\015]/; # \n\r
2574                $password .= $key;
2575        }
2576        Term::ReadKey::ReadMode('restore');
2577        print STDERR "\n";
2578        STDERR->flush;
2579        $password;
2580}
2581
2582package SVN::Git::Fetcher;
2583use vars qw/@ISA/;
2584use strict;
2585use warnings;
2586use Carp qw/croak/;
2587use IO::File qw//;
2588use Digest::MD5;
2589
2590# file baton members: path, mode_a, mode_b, pool, fh, blob, base
2591sub new {
2592        my ($class, $git_svn) = @_;
2593        my $self = SVN::Delta::Editor->new;
2594        bless $self, $class;
2595        $self->{c} = $git_svn->{last_commit} if exists $git_svn->{last_commit};
2596        $self->{empty} = {};
2597        $self->{dir_prop} = {};
2598        $self->{file_prop} = {};
2599        $self->{absent_dir} = {};
2600        $self->{absent_file} = {};
2601        $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
2602        $self;
2603}
2604
2605sub set_path_strip {
2606        my ($self, $path) = @_;
2607        $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
2608}
2609
2610sub open_root {
2611        { path => '' };
2612}
2613
2614sub open_directory {
2615        my ($self, $path, $pb, $rev) = @_;
2616        { path => $path };
2617}
2618
2619sub git_path {
2620        my ($self, $path) = @_;
2621        if ($self->{path_strip}) {
2622                $path =~ s!$self->{path_strip}!! or
2623                  die "Failed to strip path '$path' ($self->{path_strip})\n";
2624        }
2625        $path;
2626}
2627
2628sub delete_entry {
2629        my ($self, $path, $rev, $pb) = @_;
2630
2631        my $gpath = $self->git_path($path);
2632        return undef if ($gpath eq '');
2633
2634        # remove entire directories.
2635        if (command('ls-tree', $self->{c}, '--', $gpath) =~ /^040000 tree/) {
2636                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
2637                                                     -r --name-only -z/,
2638                                                     $self->{c}, '--', $gpath);
2639                local $/ = "\0";
2640                while (<$ls>) {
2641                        chomp;
2642                        $self->{gii}->remove($_);
2643                        print "\tD\t$_\n" unless $::_q;
2644                }
2645                print "\tD\t$gpath/\n" unless $::_q;
2646                command_close_pipe($ls, $ctx);
2647                $self->{empty}->{$path} = 0
2648        } else {
2649                $self->{gii}->remove($gpath);
2650                print "\tD\t$gpath\n" unless $::_q;
2651        }
2652        undef;
2653}
2654
2655sub open_file {
2656        my ($self, $path, $pb, $rev) = @_;
2657        my $gpath = $self->git_path($path);
2658        my ($mode, $blob) = (command('ls-tree', $self->{c}, '--', $gpath)
2659                             =~ /^(\d{6}) blob ([a-f\d]{40})\t/);
2660        unless (defined $mode && defined $blob) {
2661                die "$path was not found in commit $self->{c} (r$rev)\n";
2662        }
2663        { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
2664          pool => SVN::Pool->new, action => 'M' };
2665}
2666
2667sub add_file {
2668        my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
2669        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
2670        delete $self->{empty}->{$dir};
2671        { path => $path, mode_a => 100644, mode_b => 100644,
2672          pool => SVN::Pool->new, action => 'A' };
2673}
2674
2675sub add_directory {
2676        my ($self, $path, $cp_path, $cp_rev) = @_;
2677        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
2678        delete $self->{empty}->{$dir};
2679        $self->{empty}->{$path} = 1;
2680        { path => $path };
2681}
2682
2683sub change_dir_prop {
2684        my ($self, $db, $prop, $value) = @_;
2685        $self->{dir_prop}->{$db->{path}} ||= {};
2686        $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
2687        undef;
2688}
2689
2690sub absent_directory {
2691        my ($self, $path, $pb) = @_;
2692        $self->{absent_dir}->{$pb->{path}} ||= [];
2693        push @{$self->{absent_dir}->{$pb->{path}}}, $path;
2694        undef;
2695}
2696
2697sub absent_file {
2698        my ($self, $path, $pb) = @_;
2699        $self->{absent_file}->{$pb->{path}} ||= [];
2700        push @{$self->{absent_file}->{$pb->{path}}}, $path;
2701        undef;
2702}
2703
2704sub change_file_prop {
2705        my ($self, $fb, $prop, $value) = @_;
2706        if ($prop eq 'svn:executable') {
2707                if ($fb->{mode_b} != 120000) {
2708                        $fb->{mode_b} = defined $value ? 100755 : 100644;
2709                }
2710        } elsif ($prop eq 'svn:special') {
2711                $fb->{mode_b} = defined $value ? 120000 : 100644;
2712        } else {
2713                $self->{file_prop}->{$fb->{path}} ||= {};
2714                $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
2715        }
2716        undef;
2717}
2718
2719sub apply_textdelta {
2720        my ($self, $fb, $exp) = @_;
2721        my $fh = IO::File->new_tmpfile;
2722        $fh->autoflush(1);
2723        # $fh gets auto-closed() by SVN::TxDelta::apply(),
2724        # (but $base does not,) so dup() it for reading in close_file
2725        open my $dup, '<&', $fh or croak $!;
2726        my $base = IO::File->new_tmpfile;
2727        $base->autoflush(1);
2728        if ($fb->{blob}) {
2729                defined (my $pid = fork) or croak $!;
2730                if (!$pid) {
2731                        open STDOUT, '>&', $base or croak $!;
2732                        print STDOUT 'link ' if ($fb->{mode_a} == 120000);
2733                        exec qw/git-cat-file blob/, $fb->{blob} or croak $!;
2734                }
2735                waitpid $pid, 0;
2736                croak $? if $?;
2737
2738                if (defined $exp) {
2739                        seek $base, 0, 0 or croak $!;
2740                        my $md5 = Digest::MD5->new;
2741                        $md5->addfile($base);
2742                        my $got = $md5->hexdigest;
2743                        die "Checksum mismatch: $fb->{path} $fb->{blob}\n",
2744                            "expected: $exp\n",
2745                            "     got: $got\n" if ($got ne $exp);
2746                }
2747        }
2748        seek $base, 0, 0 or croak $!;
2749        $fb->{fh} = $dup;
2750        $fb->{base} = $base;
2751        [ SVN::TxDelta::apply($base, $fh, undef, $fb->{path}, $fb->{pool}) ];
2752}
2753
2754sub close_file {
2755        my ($self, $fb, $exp) = @_;
2756        my $hash;
2757        my $path = $self->git_path($fb->{path});
2758        if (my $fh = $fb->{fh}) {
2759                if (defined $exp) {
2760                        seek($fh, 0, 0) or croak $!;
2761                        my $md5 = Digest::MD5->new;
2762                        $md5->addfile($fh);
2763                        my $got = $md5->hexdigest;
2764                        if ($got ne $exp) {
2765                                die "Checksum mismatch: $path\n",
2766                                    "expected: $exp\n    got: $got\n";
2767                        }
2768                }
2769                sysseek($fh, 0, 0) or croak $!;
2770                if ($fb->{mode_b} == 120000) {
2771                        sysread($fh, my $buf, 5) == 5 or croak $!;
2772                        $buf eq 'link ' or die "$path has mode 120000",
2773                                               "but is not a link\n";
2774                }
2775                defined(my $pid = open my $out,'-|') or die "Can't fork: $!\n";
2776                if (!$pid) {
2777                        open STDIN, '<&', $fh or croak $!;
2778                        exec qw/git-hash-object -w --stdin/ or croak $!;
2779                }
2780                chomp($hash = do { local $/; <$out> });
2781                close $out or croak $!;
2782                close $fh or croak $!;
2783                $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
2784                close $fb->{base} or croak $!;
2785        } else {
2786                $hash = $fb->{blob} or die "no blob information\n";
2787        }
2788        $fb->{pool}->clear;
2789        $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
2790        print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
2791        undef;
2792}
2793
2794sub abort_edit {
2795        my $self = shift;
2796        $self->{nr} = $self->{gii}->{nr};
2797        delete $self->{gii};
2798        $self->SUPER::abort_edit(@_);
2799}
2800
2801sub close_edit {
2802        my $self = shift;
2803        $self->{git_commit_ok} = 1;
2804        $self->{nr} = $self->{gii}->{nr};
2805        delete $self->{gii};
2806        $self->SUPER::close_edit(@_);
2807}
2808
2809package SVN::Git::Editor;
2810use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
2811use strict;
2812use warnings;
2813use Carp qw/croak/;
2814use IO::File;
2815use Digest::MD5;
2816
2817sub new {
2818        my ($class, $opts) = @_;
2819        foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
2820                die "$_ required!\n" unless (defined $opts->{$_});
2821        }
2822
2823        my $pool = SVN::Pool->new;
2824        my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
2825        my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
2826                                     $opts->{r}, $mods);
2827
2828        # $opts->{ra} functions should not be used after this:
2829        my @ce  = $opts->{ra}->get_commit_editor($opts->{log},
2830                                                $opts->{editor_cb}, $pool);
2831        my $self = SVN::Delta::Editor->new(@ce, $pool);
2832        bless $self, $class;
2833        foreach (qw/svn_path r tree_a tree_b/) {
2834                $self->{$_} = $opts->{$_};
2835        }
2836        $self->{url} = $opts->{ra}->{url};
2837        $self->{mods} = $mods;
2838        $self->{types} = $types;
2839        $self->{pool} = $pool;
2840        $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
2841        $self->{rm} = { };
2842        $self->{path_prefix} = length $self->{svn_path} ?
2843                               "$self->{svn_path}/" : '';
2844        return $self;
2845}
2846
2847sub generate_diff {
2848        my ($tree_a, $tree_b) = @_;
2849        my @diff_tree = qw(diff-tree -z -r);
2850        if ($_cp_similarity) {
2851                push @diff_tree, "-C$_cp_similarity";
2852        } else {
2853                push @diff_tree, '-C';
2854        }
2855        push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
2856        push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
2857        push @diff_tree, $tree_a, $tree_b;
2858        my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
2859        local $/ = "\0";
2860        my $state = 'meta';
2861        my @mods;
2862        while (<$diff_fh>) {
2863                chomp $_; # this gets rid of the trailing "\0"
2864                if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
2865                                        $::sha1\s($::sha1)\s
2866                                        ([MTCRAD])\d*$/xo) {
2867                        push @mods, {   mode_a => $1, mode_b => $2,
2868                                        sha1_b => $3, chg => $4 };
2869                        if ($4 =~ /^(?:C|R)$/) {
2870                                $state = 'file_a';
2871                        } else {
2872                                $state = 'file_b';
2873                        }
2874                } elsif ($state eq 'file_a') {
2875                        my $x = $mods[$#mods] or croak "Empty array\n";
2876                        if ($x->{chg} !~ /^(?:C|R)$/) {
2877                                croak "Error parsing $_, $x->{chg}\n";
2878                        }
2879                        $x->{file_a} = $_;
2880                        $state = 'file_b';
2881                } elsif ($state eq 'file_b') {
2882                        my $x = $mods[$#mods] or croak "Empty array\n";
2883                        if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
2884                                croak "Error parsing $_, $x->{chg}\n";
2885                        }
2886                        if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
2887                                croak "Error parsing $_, $x->{chg}\n";
2888                        }
2889                        $x->{file_b} = $_;
2890                        $state = 'meta';
2891                } else {
2892                        croak "Error parsing $_\n";
2893                }
2894        }
2895        command_close_pipe($diff_fh, $ctx);
2896        \@mods;
2897}
2898
2899sub check_diff_paths {
2900        my ($ra, $pfx, $rev, $mods) = @_;
2901        my %types;
2902        $pfx .= '/' if length $pfx;
2903
2904        sub type_diff_paths {
2905                my ($ra, $types, $path, $rev) = @_;
2906                my @p = split m#/+#, $path;
2907                my $c = shift @p;
2908                unless (defined $types->{$c}) {
2909                        $types->{$c} = $ra->check_path($c, $rev);
2910                }
2911                while (@p) {
2912                        $c .= '/' . shift @p;
2913                        next if defined $types->{$c};
2914                        $types->{$c} = $ra->check_path($c, $rev);
2915                }
2916        }
2917
2918        foreach my $m (@$mods) {
2919                foreach my $f (qw/file_a file_b/) {
2920                        next unless defined $m->{$f};
2921                        my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
2922                        if (length $pfx.$dir && ! defined $types{$dir}) {
2923                                type_diff_paths($ra, \%types, $pfx.$dir, $rev);
2924                        }
2925                }
2926        }
2927        \%types;
2928}
2929
2930sub split_path {
2931        return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
2932}
2933
2934sub repo_path {
2935        my ($self, $path) = @_;
2936        $self->{path_prefix}.(defined $path ? $path : '');
2937}
2938
2939sub url_path {
2940        my ($self, $path) = @_;
2941        if ($self->{url} =~ m#^https?://#) {
2942                $path =~ s/([^a-zA-Z0-9_.-])/uc sprintf("%%%02x",ord($1))/eg;
2943        }
2944        $self->{url} . '/' . $self->repo_path($path);
2945}
2946
2947sub rmdirs {
2948        my ($self) = @_;
2949        my $rm = $self->{rm};
2950        delete $rm->{''}; # we never delete the url we're tracking
2951        return unless %$rm;
2952
2953        foreach (keys %$rm) {
2954                my @d = split m#/#, $_;
2955                my $c = shift @d;
2956                $rm->{$c} = 1;
2957                while (@d) {
2958                        $c .= '/' . shift @d;
2959                        $rm->{$c} = 1;
2960                }
2961        }
2962        delete $rm->{$self->{svn_path}};
2963        delete $rm->{''}; # we never delete the url we're tracking
2964        return unless %$rm;
2965
2966        my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
2967                                             $self->{tree_b});
2968        local $/ = "\0";
2969        while (<$fh>) {
2970                chomp;
2971                my @dn = split m#/#, $_;
2972                while (pop @dn) {
2973                        delete $rm->{join '/', @dn};
2974                }
2975                unless (%$rm) {
2976                        close $fh;
2977                        return;
2978                }
2979        }
2980        command_close_pipe($fh, $ctx);
2981
2982        my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
2983        foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
2984                $self->close_directory($bat->{$d}, $p);
2985                my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
2986                print "\tD+\t$d/\n" unless $::_q;
2987                $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
2988                delete $bat->{$d};
2989        }
2990}
2991
2992sub open_or_add_dir {
2993        my ($self, $full_path, $baton) = @_;
2994        my $t = $self->{types}->{$full_path};
2995        if (!defined $t) {
2996                die "$full_path not known in r$self->{r} or we have a bug!\n";
2997        }
2998        {
2999                no warnings 'once';
3000                # SVN::Node::none and SVN::Node::file are used only once,
3001                # so we're shutting up Perl's warnings about them.
3002                if ($t == $SVN::Node::none) {
3003                        return $self->add_directory($full_path, $baton,
3004                            undef, -1, $self->{pool});
3005                } elsif ($t == $SVN::Node::dir) {
3006                        return $self->open_directory($full_path, $baton,
3007                            $self->{r}, $self->{pool});
3008                } # no warnings 'once'
3009                print STDERR "$full_path already exists in repository at ",
3010                    "r$self->{r} and it is not a directory (",
3011                    ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
3012        } # no warnings 'once'
3013        exit 1;
3014}
3015
3016sub ensure_path {
3017        my ($self, $path) = @_;
3018        my $bat = $self->{bat};
3019        my $repo_path = $self->repo_path($path);
3020        return $bat->{''} unless (length $repo_path);
3021        my @p = split m#/+#, $repo_path;
3022        my $c = shift @p;
3023        $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
3024        while (@p) {
3025                my $c0 = $c;
3026                $c .= '/' . shift @p;
3027                $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
3028        }
3029        return $bat->{$c};
3030}
3031
3032sub A {
3033        my ($self, $m) = @_;
3034        my ($dir, $file) = split_path($m->{file_b});
3035        my $pbat = $self->ensure_path($dir);
3036        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3037                                        undef, -1);
3038        print "\tA\t$m->{file_b}\n" unless $::_q;
3039        $self->chg_file($fbat, $m);
3040        $self->close_file($fbat,undef,$self->{pool});
3041}
3042
3043sub C {
3044        my ($self, $m) = @_;
3045        my ($dir, $file) = split_path($m->{file_b});
3046        my $pbat = $self->ensure_path($dir);
3047        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3048                                $self->url_path($m->{file_a}), $self->{r});
3049        print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
3050        $self->chg_file($fbat, $m);
3051        $self->close_file($fbat,undef,$self->{pool});
3052}
3053
3054sub delete_entry {
3055        my ($self, $path, $pbat) = @_;
3056        my $rpath = $self->repo_path($path);
3057        my ($dir, $file) = split_path($rpath);
3058        $self->{rm}->{$dir} = 1;
3059        $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
3060}
3061
3062sub R {
3063        my ($self, $m) = @_;
3064        my ($dir, $file) = split_path($m->{file_b});
3065        my $pbat = $self->ensure_path($dir);
3066        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3067                                $self->url_path($m->{file_a}), $self->{r});
3068        print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
3069        $self->chg_file($fbat, $m);
3070        $self->close_file($fbat,undef,$self->{pool});
3071
3072        ($dir, $file) = split_path($m->{file_a});
3073        $pbat = $self->ensure_path($dir);
3074        $self->delete_entry($m->{file_a}, $pbat);
3075}
3076
3077sub M {
3078        my ($self, $m) = @_;
3079        my ($dir, $file) = split_path($m->{file_b});
3080        my $pbat = $self->ensure_path($dir);
3081        my $fbat = $self->open_file($self->repo_path($m->{file_b}),
3082                                $pbat,$self->{r},$self->{pool});
3083        print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
3084        $self->chg_file($fbat, $m);
3085        $self->close_file($fbat,undef,$self->{pool});
3086}
3087
3088sub T { shift->M(@_) }
3089
3090sub change_file_prop {
3091        my ($self, $fbat, $pname, $pval) = @_;
3092        $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
3093}
3094
3095sub chg_file {
3096        my ($self, $fbat, $m) = @_;
3097        if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
3098                $self->change_file_prop($fbat,'svn:executable','*');
3099        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
3100                $self->change_file_prop($fbat,'svn:executable',undef);
3101        }
3102        my $fh = IO::File->new_tmpfile or croak $!;
3103        if ($m->{mode_b} =~ /^120/) {
3104                print $fh 'link ' or croak $!;
3105                $self->change_file_prop($fbat,'svn:special','*');
3106        } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
3107                $self->change_file_prop($fbat,'svn:special',undef);
3108        }
3109        defined(my $pid = fork) or croak $!;
3110        if (!$pid) {
3111                open STDOUT, '>&', $fh or croak $!;
3112                exec qw/git-cat-file blob/, $m->{sha1_b} or croak $!;
3113        }
3114        waitpid $pid, 0;
3115        croak $? if $?;
3116        $fh->flush == 0 or croak $!;
3117        seek $fh, 0, 0 or croak $!;
3118
3119        my $md5 = Digest::MD5->new;
3120        $md5->addfile($fh) or croak $!;
3121        seek $fh, 0, 0 or croak $!;
3122
3123        my $exp = $md5->hexdigest;
3124        my $pool = SVN::Pool->new;
3125        my $atd = $self->apply_textdelta($fbat, undef, $pool);
3126        my $got = SVN::TxDelta::send_stream($fh, @$atd, $pool);
3127        die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
3128        $pool->clear;
3129
3130        close $fh or croak $!;
3131}
3132
3133sub D {
3134        my ($self, $m) = @_;
3135        my ($dir, $file) = split_path($m->{file_b});
3136        my $pbat = $self->ensure_path($dir);
3137        print "\tD\t$m->{file_b}\n" unless $::_q;
3138        $self->delete_entry($m->{file_b}, $pbat);
3139}
3140
3141sub close_edit {
3142        my ($self) = @_;
3143        my ($p,$bat) = ($self->{pool}, $self->{bat});
3144        foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
3145                next if $_ eq '';
3146                $self->close_directory($bat->{$_}, $p);
3147        }
3148        $self->close_directory($bat->{''}, $p);
3149        $self->SUPER::close_edit($p);
3150        $p->clear;
3151}
3152
3153sub abort_edit {
3154        my ($self) = @_;
3155        $self->SUPER::abort_edit($self->{pool});
3156}
3157
3158sub DESTROY {
3159        my $self = shift;
3160        $self->SUPER::DESTROY(@_);
3161        $self->{pool}->clear;
3162}
3163
3164# this drives the editor
3165sub apply_diff {
3166        my ($self) = @_;
3167        my $mods = $self->{mods};
3168        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
3169        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
3170                my $f = $m->{chg};
3171                if (defined $o{$f}) {
3172                        $self->$f($m);
3173                } else {
3174                        fatal("Invalid change type: $f");
3175                }
3176        }
3177        $self->rmdirs if $_rmdir;
3178        if (@$mods == 0) {
3179                $self->abort_edit;
3180        } else {
3181                $self->close_edit;
3182        }
3183        return scalar @$mods;
3184}
3185
3186package Git::SVN::Ra;
3187use vars qw/@ISA $config_dir $_log_window_size/;
3188use strict;
3189use warnings;
3190my ($ra_invalid, $can_do_switch, %ignored_err, $RA);
3191
3192BEGIN {
3193        # enforce temporary pool usage for some simple functions
3194        no strict 'refs';
3195        for my $f (qw/rev_proplist get_latest_revnum get_uuid get_repos_root/) {
3196                my $SUPER = "SUPER::$f";
3197                *$f = sub {
3198                        my $self = shift;
3199                        my $pool = SVN::Pool->new;
3200                        my @ret = $self->$SUPER(@_,$pool);
3201                        $pool->clear;
3202                        wantarray ? @ret : $ret[0];
3203                };
3204        }
3205}
3206
3207sub _auth_providers () {
3208        [
3209          SVN::Client::get_simple_provider(),
3210          SVN::Client::get_ssl_server_trust_file_provider(),
3211          SVN::Client::get_simple_prompt_provider(
3212            \&Git::SVN::Prompt::simple, 2),
3213          SVN::Client::get_ssl_client_cert_file_provider(),
3214          SVN::Client::get_ssl_client_cert_prompt_provider(
3215            \&Git::SVN::Prompt::ssl_client_cert, 2),
3216          SVN::Client::get_ssl_client_cert_pw_prompt_provider(
3217            \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
3218          SVN::Client::get_username_provider(),
3219          SVN::Client::get_ssl_server_trust_prompt_provider(
3220            \&Git::SVN::Prompt::ssl_server_trust),
3221          SVN::Client::get_username_prompt_provider(
3222            \&Git::SVN::Prompt::username, 2)
3223        ]
3224}
3225
3226sub escape_uri_only {
3227        my ($uri) = @_;
3228        my @tmp;
3229        foreach (split m{/}, $uri) {
3230                s/([^\w.-])/sprintf("%%%02X",ord($1))/eg;
3231                push @tmp, $_;
3232        }
3233        join('/', @tmp);
3234}
3235
3236sub escape_url {
3237        my ($url) = @_;
3238        if ($url =~ m#^(https?)://([^/]+)(.*)$#) {
3239                my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
3240                $url = "$scheme://$domain$uri";
3241        }
3242        $url;
3243}
3244
3245sub new {
3246        my ($class, $url) = @_;
3247        $url =~ s!/+$!!;
3248        return $RA if ($RA && $RA->{url} eq $url);
3249
3250        SVN::_Core::svn_config_ensure($config_dir, undef);
3251        my ($baton, $callbacks) = SVN::Core::auth_open_helper(_auth_providers);
3252        my $config = SVN::Core::config_get_config($config_dir);
3253        $RA = undef;
3254        my $dont_store_passwords = 1;
3255        my $conf_t = ${$config}{'config'};
3256        {
3257                no warnings 'once';
3258                # The usage of $SVN::_Core::SVN_CONFIG_* variables
3259                # produces warnings that variables are used only once.
3260                # I had not found the better way to shut them up, so
3261                # the warnings of type 'once' are disabled in this block.
3262                if (SVN::_Core::svn_config_get_bool($conf_t,
3263                    $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
3264                    $SVN::_Core::SVN_CONFIG_OPTION_STORE_PASSWORDS,
3265                    1) == 0) {
3266                        SVN::_Core::svn_auth_set_parameter($baton,
3267                            $SVN::_Core::SVN_AUTH_PARAM_DONT_STORE_PASSWORDS,
3268                            bless (\$dont_store_passwords, "_p_void"));
3269                }
3270                if (SVN::_Core::svn_config_get_bool($conf_t,
3271                    $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
3272                    $SVN::_Core::SVN_CONFIG_OPTION_STORE_AUTH_CREDS,
3273                    1) == 0) {
3274                        $Git::SVN::Prompt::_no_auth_cache = 1;
3275                }
3276        } # no warnings 'once'
3277        my $self = SVN::Ra->new(url => escape_url($url), auth => $baton,
3278                              config => $config,
3279                              pool => SVN::Pool->new,
3280                              auth_provider_callbacks => $callbacks);
3281        $self->{url} = $url;
3282        $self->{svn_path} = $url;
3283        $self->{repos_root} = $self->get_repos_root;
3284        $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
3285        $self->{cache} = { check_path => { r => 0, data => {} },
3286                           get_dir => { r => 0, data => {} } };
3287        $RA = bless $self, $class;
3288}
3289
3290sub check_path {
3291        my ($self, $path, $r) = @_;
3292        my $cache = $self->{cache}->{check_path};
3293        if ($r == $cache->{r} && exists $cache->{data}->{$path}) {
3294                return $cache->{data}->{$path};
3295        }
3296        my $pool = SVN::Pool->new;
3297        my $t = $self->SUPER::check_path($path, $r, $pool);
3298        $pool->clear;
3299        if ($r != $cache->{r}) {
3300                %{$cache->{data}} = ();
3301                $cache->{r} = $r;
3302        }
3303        $cache->{data}->{$path} = $t;
3304}
3305
3306sub get_dir {
3307        my ($self, $dir, $r) = @_;
3308        my $cache = $self->{cache}->{get_dir};
3309        if ($r == $cache->{r}) {
3310                if (my $x = $cache->{data}->{$dir}) {
3311                        return wantarray ? @$x : $x->[0];
3312                }
3313        }
3314        my $pool = SVN::Pool->new;
3315        my ($d, undef, $props) = $self->SUPER::get_dir($dir, $r, $pool);
3316        my %dirents = map { $_ => { kind => $d->{$_}->kind } } keys %$d;
3317        $pool->clear;
3318        if ($r != $cache->{r}) {
3319                %{$cache->{data}} = ();
3320                $cache->{r} = $r;
3321        }
3322        $cache->{data}->{$dir} = [ \%dirents, $r, $props ];
3323        wantarray ? (\%dirents, $r, $props) : \%dirents;
3324}
3325
3326sub DESTROY {
3327        # do not call the real DESTROY since we store ourselves in $RA
3328}
3329
3330sub get_log {
3331        my ($self, @args) = @_;
3332        my $pool = SVN::Pool->new;
3333        splice(@args, 3, 1) if ($SVN::Core::VERSION le '1.2.0');
3334        my $ret = $self->SUPER::get_log(@args, $pool);
3335        $pool->clear;
3336        $ret;
3337}
3338
3339sub trees_match {
3340        my ($self, $url1, $rev1, $url2, $rev2) = @_;
3341        my $ctx = SVN::Client->new(auth => _auth_providers);
3342        my $out = IO::File->new_tmpfile;
3343
3344        # older SVN (1.1.x) doesn't take $pool as the last parameter for
3345        # $ctx->diff(), so we'll create a default one
3346        my $pool = SVN::Pool->new_default_sub;
3347
3348        $ra_invalid = 1; # this will open a new SVN::Ra connection to $url1
3349        $ctx->diff([], $url1, $rev1, $url2, $rev2, 1, 1, 0, $out, $out);
3350        $out->flush;
3351        my $ret = (($out->stat)[7] == 0);
3352        close $out or croak $!;
3353
3354        $ret;
3355}
3356
3357sub get_commit_editor {
3358        my ($self, $log, $cb, $pool) = @_;
3359        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
3360        $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
3361}
3362
3363sub gs_do_update {
3364        my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
3365        my $new = ($rev_a == $rev_b);
3366        my $path = $gs->{path};
3367
3368        if ($new && -e $gs->{index}) {
3369                unlink $gs->{index} or die
3370                  "Couldn't unlink index: $gs->{index}: $!\n";
3371        }
3372        my $pool = SVN::Pool->new;
3373        $editor->set_path_strip($path);
3374        my (@pc) = split m#/#, $path;
3375        my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
3376                                        1, $editor, $pool);
3377        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
3378
3379        # Since we can't rely on svn_ra_reparent being available, we'll
3380        # just have to do some magic with set_path to make it so
3381        # we only want a partial path.
3382        my $sp = '';
3383        my $final = join('/', @pc);
3384        while (@pc) {
3385                $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
3386                $sp .= '/' if length $sp;
3387                $sp .= shift @pc;
3388        }
3389        die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
3390
3391        $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
3392
3393        $reporter->finish_report($pool);
3394        $pool->clear;
3395        $editor->{git_commit_ok};
3396}
3397
3398# this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
3399# svn_ra_reparent didn't work before 1.4)
3400sub gs_do_switch {
3401        my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
3402        my $path = $gs->{path};
3403        my $pool = SVN::Pool->new;
3404
3405        my $full_url = $self->{url};
3406        my $old_url = $full_url;
3407        $full_url .= '/' . escape_uri_only($path) if length $path;
3408        my ($ra, $reparented);
3409        if ($old_url ne $full_url) {
3410                if ($old_url !~ m#^svn(\+ssh)?://#) {
3411                        SVN::_Ra::svn_ra_reparent($self->{session}, $full_url,
3412                                                  $pool);
3413                        $self->{url} = $full_url;
3414                        $reparented = 1;
3415                } else {
3416                        $_[0] = undef;
3417                        $self = undef;
3418                        $RA = undef;
3419                        $ra = Git::SVN::Ra->new($full_url);
3420                        $ra_invalid = 1;
3421                }
3422        }
3423        $ra ||= $self;
3424        my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
3425        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
3426        $reporter->set_path('', $rev_a, 0, @lock, $pool);
3427        $reporter->finish_report($pool);
3428
3429        if ($reparented) {
3430                SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
3431                $self->{url} = $old_url;
3432        }
3433
3434        $pool->clear;
3435        $editor->{git_commit_ok};
3436}
3437
3438sub longest_common_path {
3439        my ($gsv, $globs) = @_;
3440        my %common;
3441        my $common_max = scalar @$gsv;
3442
3443        foreach my $gs (@$gsv) {
3444                my @tmp = split m#/#, $gs->{path};
3445                my $p = '';
3446                foreach (@tmp) {
3447                        $p .= length($p) ? "/$_" : $_;
3448                        $common{$p} ||= 0;
3449                        $common{$p}++;
3450                }
3451        }
3452        $globs ||= [];
3453        $common_max += scalar @$globs;
3454        foreach my $glob (@$globs) {
3455                my @tmp = split m#/#, $glob->{path}->{left};
3456                my $p = '';
3457                foreach (@tmp) {
3458                        $p .= length($p) ? "/$_" : $_;
3459                        $common{$p} ||= 0;
3460                        $common{$p}++;
3461                }
3462        }
3463
3464        my $longest_path = '';
3465        foreach (sort {length $b <=> length $a} keys %common) {
3466                if ($common{$_} == $common_max) {
3467                        $longest_path = $_;
3468                        last;
3469                }
3470        }
3471        $longest_path;
3472}
3473
3474sub gs_fetch_loop_common {
3475        my ($self, $base, $head, $gsv, $globs) = @_;
3476        return if ($base > $head);
3477        my $inc = $_log_window_size;
3478        my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
3479        my $longest_path = longest_common_path($gsv, $globs);
3480        my $ra_url = $self->{url};
3481        while (1) {
3482                my %revs;
3483                my $err;
3484                my $err_handler = $SVN::Error::handler;
3485                $SVN::Error::handler = sub {
3486                        ($err) = @_;
3487                        skip_unknown_revs($err);
3488                };
3489                sub _cb {
3490                        my ($paths, $r, $author, $date, $log) = @_;
3491                        [ dup_changed_paths($paths),
3492                          { author => $author, date => $date, log => $log } ];
3493                }
3494                $self->get_log([$longest_path], $min, $max, 0, 1, 1,
3495                               sub { $revs{$_[1]} = _cb(@_) });
3496                if ($err && $max >= $head) {
3497                        print STDERR "Path '$longest_path' ",
3498                                     "was probably deleted:\n",
3499                                     $err->expanded_message,
3500                                     "\nWill attempt to follow ",
3501                                     "revisions r$min .. r$max ",
3502                                     "committed before the deletion\n";
3503                        my $hi = $max;
3504                        while (--$hi >= $min) {
3505                                my $ok;
3506                                $self->get_log([$longest_path], $min, $hi,
3507                                               0, 1, 1, sub {
3508                                               $ok ||= $_[1];
3509                                               $revs{$_[1]} = _cb(@_) });
3510                                if ($ok) {
3511                                        print STDERR "r$min .. r$ok OK\n";
3512                                        last;
3513                                }
3514                        }
3515                }
3516                $SVN::Error::handler = $err_handler;
3517
3518                my %exists = map { $_->{path} => $_ } @$gsv;
3519                foreach my $r (sort {$a <=> $b} keys %revs) {
3520                        my ($paths, $logged) = @{$revs{$r}};
3521
3522                        foreach my $gs ($self->match_globs(\%exists, $paths,
3523                                                           $globs, $r)) {
3524                                if ($gs->rev_db_max >= $r) {
3525                                        next;
3526                                }
3527                                next unless $gs->match_paths($paths, $r);
3528                                $gs->{logged_rev_props} = $logged;
3529                                if (my $last_commit = $gs->last_commit) {
3530                                        $gs->assert_index_clean($last_commit);
3531                                }
3532                                my $log_entry = $gs->do_fetch($paths, $r);
3533                                if ($log_entry) {
3534                                        $gs->do_git_commit($log_entry);
3535                                }
3536                        }
3537                        foreach my $g (@$globs) {
3538                                my $k = "svn-remote.$g->{remote}." .
3539                                        "$g->{t}-maxRev";
3540                                Git::SVN::tmp_config($k, $r);
3541                        }
3542                        if ($ra_invalid) {
3543                                $_[0] = undef;
3544                                $self = undef;
3545                                $RA = undef;
3546                                $self = Git::SVN::Ra->new($ra_url);
3547                                $ra_invalid = undef;
3548                        }
3549                }
3550                # pre-fill the .rev_db since it'll eventually get filled in
3551                # with '0' x40 if something new gets committed
3552                foreach my $gs (@$gsv) {
3553                        next if defined $gs->rev_db_get($max);
3554                        $gs->rev_db_set($max, 0 x40);
3555                }
3556                foreach my $g (@$globs) {
3557                        my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
3558                        Git::SVN::tmp_config($k, $max);
3559                }
3560                last if $max >= $head;
3561                $min = $max + 1;
3562                $max += $inc;
3563                $max = $head if ($max > $head);
3564        }
3565}
3566
3567sub match_globs {
3568        my ($self, $exists, $paths, $globs, $r) = @_;
3569
3570        sub get_dir_check {
3571                my ($self, $exists, $g, $r) = @_;
3572                my @x = eval { $self->get_dir($g->{path}->{left}, $r) };
3573                return unless scalar @x == 3;
3574                my $dirents = $x[0];
3575                foreach my $de (keys %$dirents) {
3576                        next if $dirents->{$de}->{kind} != $SVN::Node::dir;
3577                        my $p = $g->{path}->full_path($de);
3578                        next if $exists->{$p};
3579                        next if (length $g->{path}->{right} &&
3580                                 ($self->check_path($p, $r) !=
3581                                  $SVN::Node::dir));
3582                        $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
3583                                         $g->{ref}->full_path($de), 1);
3584                }
3585        }
3586        foreach my $g (@$globs) {
3587                if (my $path = $paths->{"/$g->{path}->{left}"}) {
3588                        if ($path->{action} =~ /^[AR]$/) {
3589                                get_dir_check($self, $exists, $g, $r);
3590                        }
3591                }
3592                foreach (keys %$paths) {
3593                        if (/$g->{path}->{left_regex}/ &&
3594                            !/$g->{path}->{regex}/) {
3595                                next if $paths->{$_}->{action} !~ /^[AR]$/;
3596                                get_dir_check($self, $exists, $g, $r);
3597                        }
3598                        next unless /$g->{path}->{regex}/;
3599                        my $p = $1;
3600                        my $pathname = $g->{path}->full_path($p);
3601                        next if $exists->{$pathname};
3602                        next if ($self->check_path($pathname, $r) !=
3603                                 $SVN::Node::dir);
3604                        $exists->{$pathname} = Git::SVN->init(
3605                                              $self->{url}, $pathname, undef,
3606                                              $g->{ref}->full_path($p), 1);
3607                }
3608                my $c = '';
3609                foreach (split m#/#, $g->{path}->{left}) {
3610                        $c .= "/$_";
3611                        next unless ($paths->{$c} &&
3612                                     ($paths->{$c}->{action} =~ /^[AR]$/));
3613                        get_dir_check($self, $exists, $g, $r);
3614                }
3615        }
3616        values %$exists;
3617}
3618
3619sub minimize_url {
3620        my ($self) = @_;
3621        return $self->{url} if ($self->{url} eq $self->{repos_root});
3622        my $url = $self->{repos_root};
3623        my @components = split(m!/!, $self->{svn_path});
3624        my $c = '';
3625        do {
3626                $url .= "/$c" if length $c;
3627                eval { (ref $self)->new($url)->get_latest_revnum };
3628        } while ($@ && ($c = shift @components));
3629        $url;
3630}
3631
3632sub can_do_switch {
3633        my $self = shift;
3634        unless (defined $can_do_switch) {
3635                my $pool = SVN::Pool->new;
3636                my $rep = eval {
3637                        $self->do_switch(1, '', 0, $self->{url},
3638                                         SVN::Delta::Editor->new, $pool);
3639                };
3640                if ($@) {
3641                        $can_do_switch = 0;
3642                } else {
3643                        $rep->abort_report($pool);
3644                        $can_do_switch = 1;
3645                }
3646                $pool->clear;
3647        }
3648        $can_do_switch;
3649}
3650
3651sub skip_unknown_revs {
3652        my ($err) = @_;
3653        my $errno = $err->apr_err();
3654        # Maybe the branch we're tracking didn't
3655        # exist when the repo started, so it's
3656        # not an error if it doesn't, just continue
3657        #
3658        # Wonderfully consistent library, eh?
3659        # 160013 - svn:// and file://
3660        # 175002 - http(s)://
3661        # 175007 - http(s):// (this repo required authorization, too...)
3662        #   More codes may be discovered later...
3663        if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
3664                my $err_key = $err->expanded_message;
3665                # revision numbers change every time, filter them out
3666                $err_key =~ s/\d+/\0/g;
3667                $err_key = "$errno\0$err_key";
3668                unless ($ignored_err{$err_key}) {
3669                        warn "W: Ignoring error from SVN, path probably ",
3670                             "does not exist: ($errno): ",
3671                             $err->expanded_message,"\n";
3672                        $ignored_err{$err_key} = 1;
3673                }
3674                return;
3675        }
3676        die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
3677}
3678
3679# svn_log_changed_path_t objects passed to get_log are likely to be
3680# overwritten even if only the refs are copied to an external variable,
3681# so we should dup the structures in their entirety.  Using an externally
3682# passed pool (instead of our temporary and quickly cleared pool in
3683# Git::SVN::Ra) does not help matters at all...
3684sub dup_changed_paths {
3685        my ($paths) = @_;
3686        return undef unless $paths;
3687        my %ret;
3688        foreach my $p (keys %$paths) {
3689                my $i = $paths->{$p};
3690                my %s = map { $_ => $i->$_ }
3691                              qw/copyfrom_path copyfrom_rev action/;
3692                $ret{$p} = \%s;
3693        }
3694        \%ret;
3695}
3696
3697package Git::SVN::Log;
3698use strict;
3699use warnings;
3700use POSIX qw/strftime/;
3701use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
3702            %rusers $show_commit $incremental/;
3703my $l_fmt;
3704
3705sub cmt_showable {
3706        my ($c) = @_;
3707        return 1 if defined $c->{r};
3708
3709        # big commit message got truncated by the 16k pretty buffer in rev-list
3710        if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
3711                                $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
3712                @{$c->{l}} = ();
3713                my @log = command(qw/cat-file commit/, $c->{c});
3714
3715                # shift off the headers
3716                shift @log while ($log[0] ne '');
3717                shift @log;
3718
3719                # TODO: make $c->{l} not have a trailing newline in the future
3720                @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
3721
3722                (undef, $c->{r}, undef) = ::extract_metadata(
3723                                (grep(/^git-svn-id: /, @log))[-1]);
3724        }
3725        return defined $c->{r};
3726}
3727
3728sub log_use_color {
3729        return 1 if $color;
3730        my ($dc, $dcvar);
3731        $dcvar = 'color.diff';
3732        $dc = `git-config --get $dcvar`;
3733        if ($dc eq '') {
3734                # nothing at all; fallback to "diff.color"
3735                $dcvar = 'diff.color';
3736                $dc = `git-config --get $dcvar`;
3737        }
3738        chomp($dc);
3739        if ($dc eq 'auto') {
3740                my $pc;
3741                $pc = `git-config --get color.pager`;
3742                if ($pc eq '') {
3743                        # does not have it -- fallback to pager.color
3744                        $pc = `git-config --bool --get pager.color`;
3745                }
3746                else {
3747                        $pc = `git-config --bool --get color.pager`;
3748                        if ($?) {
3749                                $pc = 'false';
3750                        }
3751                }
3752                chomp($pc);
3753                if (-t *STDOUT || (defined $pager && $pc eq 'true')) {
3754                        return ($ENV{TERM} && $ENV{TERM} ne 'dumb');
3755                }
3756                return 0;
3757        }
3758        return 0 if $dc eq 'never';
3759        return 1 if $dc eq 'always';
3760        chomp($dc = `git-config --bool --get $dcvar`);
3761        return ($dc eq 'true');
3762}
3763
3764sub git_svn_log_cmd {
3765        my ($r_min, $r_max, @args) = @_;
3766        my $head = 'HEAD';
3767        my (@files, @log_opts);
3768        foreach my $x (@args) {
3769                if ($x eq '--' || @files) {
3770                        push @files, $x;
3771                } else {
3772                        if (::verify_ref("$x^0")) {
3773                                $head = $x;
3774                        } else {
3775                                push @log_opts, $x;
3776                        }
3777                }
3778        }
3779
3780        my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
3781        $gs ||= Git::SVN->_new;
3782        my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
3783                   $gs->refname);
3784        push @cmd, '-r' unless $non_recursive;
3785        push @cmd, qw/--raw --name-status/ if $verbose;
3786        push @cmd, '--color' if log_use_color();
3787        push @cmd, @log_opts;
3788        if (defined $r_max && $r_max == $r_min) {
3789                push @cmd, '--max-count=1';
3790                if (my $c = $gs->rev_db_get($r_max)) {
3791                        push @cmd, $c;
3792                }
3793        } elsif (defined $r_max) {
3794                my ($c_min, $c_max);
3795                $c_max = $gs->rev_db_get($r_max);
3796                $c_min = $gs->rev_db_get($r_min);
3797                if (defined $c_min && defined $c_max) {
3798                        if ($r_max > $r_max) {
3799                                push @cmd, "$c_min..$c_max";
3800                        } else {
3801                                push @cmd, "$c_max..$c_min";
3802                        }
3803                } elsif ($r_max > $r_min) {
3804                        push @cmd, $c_max;
3805                } else {
3806                        push @cmd, $c_min;
3807                }
3808        }
3809        return (@cmd, @files);
3810}
3811
3812# adapted from pager.c
3813sub config_pager {
3814        $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
3815        if (!defined $pager) {
3816                $pager = 'less';
3817        } elsif (length $pager == 0 || $pager eq 'cat') {
3818                $pager = undef;
3819        }
3820}
3821
3822sub run_pager {
3823        return unless -t *STDOUT && defined $pager;
3824        pipe my $rfd, my $wfd or return;
3825        defined(my $pid = fork) or ::fatal "Can't fork: $!";
3826        if (!$pid) {
3827                open STDOUT, '>&', $wfd or
3828                                     ::fatal "Can't redirect to stdout: $!";
3829                return;
3830        }
3831        open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!";
3832        $ENV{LESS} ||= 'FRSX';
3833        exec $pager or ::fatal "Can't run pager: $! ($pager)";
3834}
3835
3836sub tz_to_s_offset {
3837        my ($tz) = @_;
3838        $tz =~ s/(\d\d)$//;
3839        return ($1 * 60) + ($tz * 3600);
3840}
3841
3842sub get_author_info {
3843        my ($dest, $author, $t, $tz) = @_;
3844        $author =~ s/(?:^\s*|\s*$)//g;
3845        $dest->{a_raw} = $author;
3846        my $au;
3847        if ($::_authors) {
3848                $au = $rusers{$author} || undef;
3849        }
3850        if (!$au) {
3851                ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
3852        }
3853        $dest->{t} = $t;
3854        $dest->{tz} = $tz;
3855        $dest->{a} = $au;
3856        # Date::Parse isn't in the standard Perl distro :(
3857        if ($tz =~ s/^\+//) {
3858                $t += tz_to_s_offset($tz);
3859        } elsif ($tz =~ s/^\-//) {
3860                $t -= tz_to_s_offset($tz);
3861        }
3862        $dest->{t_utc} = $t;
3863}
3864
3865sub process_commit {
3866        my ($c, $r_min, $r_max, $defer) = @_;
3867        if (defined $r_min && defined $r_max) {
3868                if ($r_min == $c->{r} && $r_min == $r_max) {
3869                        show_commit($c);
3870                        return 0;
3871                }
3872                return 1 if $r_min == $r_max;
3873                if ($r_min < $r_max) {
3874                        # we need to reverse the print order
3875                        return 0 if (defined $limit && --$limit < 0);
3876                        push @$defer, $c;
3877                        return 1;
3878                }
3879                if ($r_min != $r_max) {
3880                        return 1 if ($r_min < $c->{r});
3881                        return 1 if ($r_max > $c->{r});
3882                }
3883        }
3884        return 0 if (defined $limit && --$limit < 0);
3885        show_commit($c);
3886        return 1;
3887}
3888
3889sub show_commit {
3890        my $c = shift;
3891        if ($oneline) {
3892                my $x = "\n";
3893                if (my $l = $c->{l}) {
3894                        while ($l->[0] =~ /^\s*$/) { shift @$l }
3895                        $x = $l->[0];
3896                }
3897                $l_fmt ||= 'A' . length($c->{r});
3898                print 'r',pack($l_fmt, $c->{r}),' | ';
3899                print "$c->{c} | " if $show_commit;
3900                print $x;
3901        } else {
3902                show_commit_normal($c);
3903        }
3904}
3905
3906sub show_commit_changed_paths {
3907        my ($c) = @_;
3908        return unless $c->{changed};
3909        print "Changed paths:\n", @{$c->{changed}};
3910}
3911
3912sub show_commit_normal {
3913        my ($c) = @_;
3914        print '-' x72, "\nr$c->{r} | ";
3915        print "$c->{c} | " if $show_commit;
3916        print "$c->{a} | ", strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)",
3917                                 localtime($c->{t_utc})), ' | ';
3918        my $nr_line = 0;
3919
3920        if (my $l = $c->{l}) {
3921                while ($l->[$#$l] eq "\n" && $#$l > 0
3922                                          && $l->[($#$l - 1)] eq "\n") {
3923                        pop @$l;
3924                }
3925                $nr_line = scalar @$l;
3926                if (!$nr_line) {
3927                        print "1 line\n\n\n";
3928                } else {
3929                        if ($nr_line == 1) {
3930                                $nr_line = '1 line';
3931                        } else {
3932                                $nr_line .= ' lines';
3933                        }
3934                        print $nr_line, "\n";
3935                        show_commit_changed_paths($c);
3936                        print "\n";
3937                        print $_ foreach @$l;
3938                }
3939        } else {
3940                print "1 line\n";
3941                show_commit_changed_paths($c);
3942                print "\n";
3943
3944        }
3945        foreach my $x (qw/raw stat diff/) {
3946                if ($c->{$x}) {
3947                        print "\n";
3948                        print $_ foreach @{$c->{$x}}
3949                }
3950        }
3951}
3952
3953sub cmd_show_log {
3954        my (@args) = @_;
3955        my ($r_min, $r_max);
3956        my $r_last = -1; # prevent dupes
3957        if (defined $TZ) {
3958                $ENV{TZ} = $TZ;
3959        } else {
3960                delete $ENV{TZ};
3961        }
3962        if (defined $::_revision) {
3963                if ($::_revision =~ /^(\d+):(\d+)$/) {
3964                        ($r_min, $r_max) = ($1, $2);
3965                } elsif ($::_revision =~ /^\d+$/) {
3966                        $r_min = $r_max = $::_revision;
3967                } else {
3968                        ::fatal "-r$::_revision is not supported, use ",
3969                                "standard 'git log' arguments instead";
3970                }
3971        }
3972
3973        config_pager();
3974        @args = git_svn_log_cmd($r_min, $r_max, @args);
3975        my $log = command_output_pipe(@args);
3976        run_pager();
3977        my (@k, $c, $d, $stat);
3978        my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
3979        while (<$log>) {
3980                if (/^${esc_color}commit ($::sha1_short)/o) {
3981                        my $cmt = $1;
3982                        if ($c && cmt_showable($c) && $c->{r} != $r_last) {
3983                                $r_last = $c->{r};
3984                                process_commit($c, $r_min, $r_max, \@k) or
3985                                                                goto out;
3986                        }
3987                        $d = undef;
3988                        $c = { c => $cmt };
3989                } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
3990                        get_author_info($c, $1, $2, $3);
3991                } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
3992                        # ignore
3993                } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
3994                        push @{$c->{raw}}, $_;
3995                } elsif (/^${esc_color}[ACRMDT]\t/) {
3996                        # we could add $SVN->{svn_path} here, but that requires
3997                        # remote access at the moment (repo_path_split)...
3998                        s#^(${esc_color})([ACRMDT])\t#$1   $2 #o;
3999                        push @{$c->{changed}}, $_;
4000                } elsif (/^${esc_color}diff /o) {
4001                        $d = 1;
4002                        push @{$c->{diff}}, $_;
4003                } elsif ($d) {
4004                        push @{$c->{diff}}, $_;
4005                } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
4006                          $esc_color*[\+\-]*$esc_color$/x) {
4007                        $stat = 1;
4008                        push @{$c->{stat}}, $_;
4009                } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
4010                        push @{$c->{stat}}, $_;
4011                        $stat = undef;
4012                } elsif (/^${esc_color}    (git-svn-id:.+)$/o) {
4013                        ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
4014                } elsif (s/^${esc_color}    //o) {
4015                        push @{$c->{l}}, $_;
4016                }
4017        }
4018        if ($c && defined $c->{r} && $c->{r} != $r_last) {
4019                $r_last = $c->{r};
4020                process_commit($c, $r_min, $r_max, \@k);
4021        }
4022        if (@k) {
4023                my $swap = $r_max;
4024                $r_max = $r_min;
4025                $r_min = $swap;
4026                process_commit($_, $r_min, $r_max) foreach reverse @k;
4027        }
4028out:
4029        close $log;
4030        print '-' x72,"\n" unless $incremental || $oneline;
4031}
4032
4033package Git::SVN::Migration;
4034# these version numbers do NOT correspond to actual version numbers
4035# of git nor git-svn.  They are just relative.
4036#
4037# v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
4038#
4039# v1 layout: .git/$id/info/url, refs/remotes/$id
4040#
4041# v2 layout: .git/svn/$id/info/url, refs/remotes/$id
4042#
4043# v3 layout: .git/svn/$id, refs/remotes/$id
4044#            - info/url may remain for backwards compatibility
4045#            - this is what we migrate up to this layout automatically,
4046#            - this will be used by git svn init on single branches
4047# v3.1 layout (auto migrated):
4048#            - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
4049#              for backwards compatibility
4050#
4051# v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
4052#            - this is only created for newly multi-init-ed
4053#              repositories.  Similar in spirit to the
4054#              --use-separate-remotes option in git-clone (now default)
4055#            - we do not automatically migrate to this (following
4056#              the example set by core git)
4057use strict;
4058use warnings;
4059use Carp qw/croak/;
4060use File::Path qw/mkpath/;
4061use File::Basename qw/dirname basename/;
4062use vars qw/$_minimize/;
4063
4064sub migrate_from_v0 {
4065        my $git_dir = $ENV{GIT_DIR};
4066        return undef unless -d $git_dir;
4067        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
4068        my $migrated = 0;
4069        while (<$fh>) {
4070                chomp;
4071                my ($id, $orig_ref) = ($_, $_);
4072                next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
4073                next unless -f "$git_dir/$id/info/url";
4074                my $new_ref = "refs/remotes/$id";
4075                if (::verify_ref("$new_ref^0")) {
4076                        print STDERR "W: $orig_ref is probably an old ",
4077                                     "branch used by an ancient version of ",
4078                                     "git-svn.\n",
4079                                     "However, $new_ref also exists.\n",
4080                                     "We will not be able ",
4081                                     "to use this branch until this ",
4082                                     "ambiguity is resolved.\n";
4083                        next;
4084                }
4085                print STDERR "Migrating from v0 layout...\n" if !$migrated;
4086                print STDERR "Renaming ref: $orig_ref => $new_ref\n";
4087                command_noisy('update-ref', $new_ref, $orig_ref);
4088                command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
4089                $migrated++;
4090        }
4091        command_close_pipe($fh, $ctx);
4092        print STDERR "Done migrating from v0 layout...\n" if $migrated;
4093        $migrated;
4094}
4095
4096sub migrate_from_v1 {
4097        my $git_dir = $ENV{GIT_DIR};
4098        my $migrated = 0;
4099        return $migrated unless -d $git_dir;
4100        my $svn_dir = "$git_dir/svn";
4101
4102        # just in case somebody used 'svn' as their $id at some point...
4103        return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
4104
4105        print STDERR "Migrating from a git-svn v1 layout...\n";
4106        mkpath([$svn_dir]);
4107        print STDERR "Data from a previous version of git-svn exists, but\n\t",
4108                     "$svn_dir\n\t(required for this version ",
4109                     "($::VERSION) of git-svn) does not. exist\n";
4110        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
4111        while (<$fh>) {
4112                my $x = $_;
4113                next unless $x =~ s#^refs/remotes/##;
4114                chomp $x;
4115                next unless -f "$git_dir/$x/info/url";
4116                my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
4117                next unless $u;
4118                my $dn = dirname("$git_dir/svn/$x");
4119                mkpath([$dn]) unless -d $dn;
4120                if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
4121                        mkpath(["$git_dir/svn/svn"]);
4122                        print STDERR " - $git_dir/$x/info => ",
4123                                        "$git_dir/svn/$x/info\n";
4124                        rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
4125                               croak "$!: $x";
4126                        # don't worry too much about these, they probably
4127                        # don't exist with repos this old (save for index,
4128                        # and we can easily regenerate that)
4129                        foreach my $f (qw/unhandled.log index .rev_db/) {
4130                                rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
4131                        }
4132                } else {
4133                        print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
4134                        rename "$git_dir/$x", "$git_dir/svn/$x" or
4135                               croak "$!: $x";
4136                }
4137                $migrated++;
4138        }
4139        command_close_pipe($fh, $ctx);
4140        print STDERR "Done migrating from a git-svn v1 layout\n";
4141        $migrated;
4142}
4143
4144sub read_old_urls {
4145        my ($l_map, $pfx, $path) = @_;
4146        my @dir;
4147        foreach (<$path/*>) {
4148                if (-r "$_/info/url") {
4149                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
4150                        my $ref_id = $pfx . basename $_;
4151                        my $url = ::file_to_s("$_/info/url");
4152                        $l_map->{$ref_id} = $url;
4153                } elsif (-d $_) {
4154                        push @dir, $_;
4155                }
4156        }
4157        foreach (@dir) {
4158                my $x = $_;
4159                $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
4160                read_old_urls($l_map, $x, $_);
4161        }
4162}
4163
4164sub migrate_from_v2 {
4165        my @cfg = command(qw/config -l/);
4166        return if grep /^svn-remote\..+\.url=/, @cfg;
4167        my %l_map;
4168        read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
4169        my $migrated = 0;
4170
4171        foreach my $ref_id (sort keys %l_map) {
4172                eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
4173                if ($@) {
4174                        Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
4175                }
4176                $migrated++;
4177        }
4178        $migrated;
4179}
4180
4181sub minimize_connections {
4182        my $r = Git::SVN::read_all_remotes();
4183        my $new_urls = {};
4184        my $root_repos = {};
4185        foreach my $repo_id (keys %$r) {
4186                my $url = $r->{$repo_id}->{url} or next;
4187                my $fetch = $r->{$repo_id}->{fetch} or next;
4188                my $ra = Git::SVN::Ra->new($url);
4189
4190                # skip existing cases where we already connect to the root
4191                if (($ra->{url} eq $ra->{repos_root}) ||
4192                    (Git::SVN::sanitize_remote_name($ra->{repos_root}) eq
4193                     $repo_id)) {
4194                        $root_repos->{$ra->{url}} = $repo_id;
4195                        next;
4196                }
4197
4198                my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
4199                my $root_path = $ra->{url};
4200                $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
4201                foreach my $path (keys %$fetch) {
4202                        my $ref_id = $fetch->{$path};
4203                        my $gs = Git::SVN->new($ref_id, $repo_id, $path);
4204
4205                        # make sure we can read when connecting to
4206                        # a higher level of a repository
4207                        my ($last_rev, undef) = $gs->last_rev_commit;
4208                        if (!defined $last_rev) {
4209                                $last_rev = eval {
4210                                        $root_ra->get_latest_revnum;
4211                                };
4212                                next if $@;
4213                        }
4214                        my $new = $root_path;
4215                        $new .= length $path ? "/$path" : '';
4216                        eval {
4217                                $root_ra->get_log([$new], $last_rev, $last_rev,
4218                                                  0, 0, 1, sub { });
4219                        };
4220                        next if $@;
4221                        $new_urls->{$ra->{repos_root}}->{$new} =
4222                                { ref_id => $ref_id,
4223                                  old_repo_id => $repo_id,
4224                                  old_path => $path };
4225                }
4226        }
4227
4228        my @emptied;
4229        foreach my $url (keys %$new_urls) {
4230                # see if we can re-use an existing [svn-remote "repo_id"]
4231                # instead of creating a(n ugly) new section:
4232                my $repo_id = $root_repos->{$url} ||
4233                              Git::SVN::sanitize_remote_name($url);
4234
4235                my $fetch = $new_urls->{$url};
4236                foreach my $path (keys %$fetch) {
4237                        my $x = $fetch->{$path};
4238                        Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
4239                        my $pfx = "svn-remote.$x->{old_repo_id}";
4240
4241                        my $old_fetch = quotemeta("$x->{old_path}:".
4242                                                  "refs/remotes/$x->{ref_id}");
4243                        command_noisy(qw/config --unset/,
4244                                      "$pfx.fetch", '^'. $old_fetch . '$');
4245                        delete $r->{$x->{old_repo_id}}->
4246                               {fetch}->{$x->{old_path}};
4247                        if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
4248                                command_noisy(qw/config --unset/,
4249                                              "$pfx.url");
4250                                push @emptied, $x->{old_repo_id}
4251                        }
4252                }
4253        }
4254        if (@emptied) {
4255                my $file = $ENV{GIT_CONFIG} || $ENV{GIT_CONFIG_LOCAL} ||
4256                           "$ENV{GIT_DIR}/config";
4257                print STDERR <<EOF;
4258The following [svn-remote] sections in your config file ($file) are empty
4259and can be safely removed:
4260EOF
4261                print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
4262        }
4263}
4264
4265sub migration_check {
4266        migrate_from_v0();
4267        migrate_from_v1();
4268        migrate_from_v2();
4269        minimize_connections() if $_minimize;
4270}
4271
4272package Git::IndexInfo;
4273use strict;
4274use warnings;
4275use Git qw/command_input_pipe command_close_pipe/;
4276
4277sub new {
4278        my ($class) = @_;
4279        my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
4280        bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
4281}
4282
4283sub remove {
4284        my ($self, $path) = @_;
4285        if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
4286                return ++$self->{nr};
4287        }
4288        undef;
4289}
4290
4291sub update {
4292        my ($self, $mode, $hash, $path) = @_;
4293        if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
4294                return ++$self->{nr};
4295        }
4296        undef;
4297}
4298
4299sub DESTROY {
4300        my ($self) = @_;
4301        command_close_pipe($self->{gui}, $self->{ctx});
4302}
4303
4304package Git::SVN::GlobSpec;
4305use strict;
4306use warnings;
4307
4308sub new {
4309        my ($class, $glob) = @_;
4310        my $re = $glob;
4311        $re =~ s!/+$!!g; # no need for trailing slashes
4312        my $nr = ($re =~ s!^(.*)\*(.*)$!\(\[^/\]+\)!g);
4313        my ($left, $right) = ($1, $2);
4314        if ($nr > 1) {
4315                die "Only one '*' wildcard expansion ",
4316                    "is supported (got $nr): '$glob'\n";
4317        } elsif ($nr == 0) {
4318                die "One '*' is needed for glob: '$glob'\n";
4319        }
4320        $re = quotemeta($left) . $re . quotemeta($right);
4321        if (length $left && !($left =~ s!/+$!!g)) {
4322                die "Missing trailing '/' on left side of: '$glob' ($left)\n";
4323        }
4324        if (length $right && !($right =~ s!^/+!!g)) {
4325                die "Missing leading '/' on right side of: '$glob' ($right)\n";
4326        }
4327        my $left_re = qr/^\/\Q$left\E(\/|$)/;
4328        bless { left => $left, right => $right, left_regex => $left_re,
4329                regex => qr/$re/, glob => $glob }, $class;
4330}
4331
4332sub full_path {
4333        my ($self, $path) = @_;
4334        return (length $self->{left} ? "$self->{left}/" : '') .
4335               $path . (length $self->{right} ? "/$self->{right}" : '');
4336}
4337
4338__END__
4339
4340Data structures:
4341
4342
4343$remotes = { # returned by read_all_remotes()
4344        'svn' => {
4345                # svn-remote.svn.url=https://svn.musicpd.org
4346                url => 'https://svn.musicpd.org',
4347                # svn-remote.svn.fetch=mpd/trunk:trunk
4348                fetch => {
4349                        'mpd/trunk' => 'trunk',
4350                },
4351                # svn-remote.svn.tags=mpd/tags/*:tags/*
4352                tags => {
4353                        path => {
4354                                left => 'mpd/tags',
4355                                right => '',
4356                                regex => qr!mpd/tags/([^/]+)$!,
4357                                glob => 'tags/*',
4358                        },
4359                        ref => {
4360                                left => 'tags',
4361                                right => '',
4362                                regex => qr!tags/([^/]+)$!,
4363                                glob => 'tags/*',
4364                        },
4365                }
4366        }
4367};
4368
4369$log_entry hashref as returned by libsvn_log_entry()
4370{
4371        log => 'whitespace-formatted log entry
4372',                                              # trailing newline is preserved
4373        revision => '8',                        # integer
4374        date => '2004-02-24T17:01:44.108345Z',  # commit date
4375        author => 'committer name'
4376};
4377
4378
4379# this is generated by generate_diff();
4380@mods = array of diff-index line hashes, each element represents one line
4381        of diff-index output
4382
4383diff-index line ($m hash)
4384{
4385        mode_a => first column of diff-index output, no leading ':',
4386        mode_b => second column of diff-index output,
4387        sha1_b => sha1sum of the final blob,
4388        chg => change type [MCRADT],
4389        file_a => original file name of a file (iff chg is 'C' or 'R')
4390        file_b => new/current file name of a file (any chg)
4391}
4392;
4393
4394# retval of read_url_paths{,_all}();
4395$l_map = {
4396        # repository root url
4397        'https://svn.musicpd.org' => {
4398                # repository path               # GIT_SVN_ID
4399                'mpd/trunk'             =>      'trunk',
4400                'mpd/tags/0.11.5'       =>      'tags/0.11.5',
4401        },
4402}
4403
4404Notes:
4405        I don't trust the each() function on unless I created %hash myself
4406        because the internal iterator may not have started at base.