git-svn.perlon commit git.html: Update the links to stale versions (171d766)
   1#!/usr/bin/env perl
   2# Copyright (C) 2006, Eric Wong <normalperson@yhbt.net>
   3# License: GPL v2 or later
   4use warnings;
   5use strict;
   6use vars qw/    $AUTHOR $VERSION
   7                $sha1 $sha1_short $_revision $_repository
   8                $_q $_authors %users/;
   9$AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
  10$VERSION = '@@GIT_VERSION@@';
  11
  12# From which subdir have we been invoked?
  13my $cmd_dir_prefix = eval {
  14        command_oneline([qw/rev-parse --show-prefix/], STDERR => 0)
  15} || '';
  16
  17my $git_dir_user_set = 1 if defined $ENV{GIT_DIR};
  18$ENV{GIT_DIR} ||= '.git';
  19$Git::SVN::default_repo_id = 'svn';
  20$Git::SVN::default_ref_id = $ENV{GIT_SVN_ID} || 'git-svn';
  21$Git::SVN::Ra::_log_window_size = 100;
  22
  23$Git::SVN::Log::TZ = $ENV{TZ};
  24$ENV{TZ} = 'UTC';
  25$| = 1; # unbuffer STDOUT
  26
  27sub fatal (@) { print STDERR "@_\n"; exit 1 }
  28require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
  29require SVN::Ra;
  30require SVN::Delta;
  31if ($SVN::Core::VERSION lt '1.1.0') {
  32        fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)";
  33}
  34push @Git::SVN::Ra::ISA, 'SVN::Ra';
  35push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
  36push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
  37use Carp qw/croak/;
  38use Digest::MD5;
  39use IO::File qw//;
  40use File::Basename qw/dirname basename/;
  41use File::Path qw/mkpath/;
  42use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev/;
  43use IPC::Open3;
  44use Git;
  45
  46BEGIN {
  47        # import functions from Git into our packages, en masse
  48        no strict 'refs';
  49        foreach (qw/command command_oneline command_noisy command_output_pipe
  50                    command_input_pipe command_close_pipe/) {
  51                for my $package ( qw(SVN::Git::Editor SVN::Git::Fetcher
  52                        Git::SVN::Migration Git::SVN::Log Git::SVN),
  53                        __PACKAGE__) {
  54                        *{"${package}::$_"} = \&{"Git::$_"};
  55                }
  56        }
  57}
  58
  59my ($SVN);
  60
  61$sha1 = qr/[a-f\d]{40}/;
  62$sha1_short = qr/[a-f\d]{4,40}/;
  63my ($_stdin, $_help, $_edit,
  64        $_message, $_file,
  65        $_template, $_shared,
  66        $_version, $_fetch_all, $_no_rebase,
  67        $_merge, $_strategy, $_dry_run, $_local,
  68        $_prefix, $_no_checkout, $_url, $_verbose,
  69        $_git_format, $_commit_url, $_tag);
  70$Git::SVN::_follow_parent = 1;
  71my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
  72                    'config-dir=s' => \$Git::SVN::Ra::config_dir,
  73                    'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache );
  74my %fc_opts = ( 'follow-parent|follow!' => \$Git::SVN::_follow_parent,
  75                'authors-file|A=s' => \$_authors,
  76                'repack:i' => \$Git::SVN::_repack,
  77                'noMetadata' => \$Git::SVN::_no_metadata,
  78                'useSvmProps' => \$Git::SVN::_use_svm_props,
  79                'useSvnsyncProps' => \$Git::SVN::_use_svnsync_props,
  80                'log-window-size=i' => \$Git::SVN::Ra::_log_window_size,
  81                'no-checkout' => \$_no_checkout,
  82                'quiet|q' => \$_q,
  83                'repack-flags|repack-args|repack-opts=s' =>
  84                   \$Git::SVN::_repack_flags,
  85                'use-log-author' => \$Git::SVN::_use_log_author,
  86                'add-author-from' => \$Git::SVN::_add_author_from,
  87                %remote_opts );
  88
  89my ($_trunk, $_tags, $_branches, $_stdlayout);
  90my %icv;
  91my %init_opts = ( 'template=s' => \$_template, 'shared:s' => \$_shared,
  92                  'trunk|T=s' => \$_trunk, 'tags|t=s' => \$_tags,
  93                  'branches|b=s' => \$_branches, 'prefix=s' => \$_prefix,
  94                  'stdlayout|s' => \$_stdlayout,
  95                  'minimize-url|m' => \$Git::SVN::_minimize_url,
  96                  'no-metadata' => sub { $icv{noMetadata} = 1 },
  97                  'use-svm-props' => sub { $icv{useSvmProps} = 1 },
  98                  'use-svnsync-props' => sub { $icv{useSvnsyncProps} = 1 },
  99                  'rewrite-root=s' => sub { $icv{rewriteRoot} = $_[1] },
 100                  %remote_opts );
 101my %cmt_opts = ( 'edit|e' => \$_edit,
 102                'rmdir' => \$SVN::Git::Editor::_rmdir,
 103                'find-copies-harder' => \$SVN::Git::Editor::_find_copies_harder,
 104                'l=i' => \$SVN::Git::Editor::_rename_limit,
 105                'copy-similarity|C=i'=> \$SVN::Git::Editor::_cp_similarity
 106);
 107
 108my %cmd = (
 109        fetch => [ \&cmd_fetch, "Download new revisions from SVN",
 110                        { 'revision|r=s' => \$_revision,
 111                          'fetch-all|all' => \$_fetch_all,
 112                           %fc_opts } ],
 113        clone => [ \&cmd_clone, "Initialize and fetch revisions",
 114                        { 'revision|r=s' => \$_revision,
 115                           %fc_opts, %init_opts } ],
 116        init => [ \&cmd_init, "Initialize a repo for tracking" .
 117                          " (requires URL argument)",
 118                          \%init_opts ],
 119        'multi-init' => [ \&cmd_multi_init,
 120                          "Deprecated alias for ".
 121                          "'$0 init -T<trunk> -b<branches> -t<tags>'",
 122                          \%init_opts ],
 123        dcommit => [ \&cmd_dcommit,
 124                     'Commit several diffs to merge with upstream',
 125                        { 'merge|m|M' => \$_merge,
 126                          'strategy|s=s' => \$_strategy,
 127                          'verbose|v' => \$_verbose,
 128                          'dry-run|n' => \$_dry_run,
 129                          'fetch-all|all' => \$_fetch_all,
 130                          'commit-url=s' => \$_commit_url,
 131                          'revision|r=i' => \$_revision,
 132                          'no-rebase' => \$_no_rebase,
 133                        %cmt_opts, %fc_opts } ],
 134        branch => [ \&cmd_branch,
 135                    'Create a branch in the SVN repository',
 136                    { 'message|m=s' => \$_message,
 137                      'dry-run|n' => \$_dry_run,
 138                      'tag|t' => \$_tag } ],
 139        tag => [ sub { $_tag = 1; cmd_branch(@_) },
 140                 'Create a tag in the SVN repository',
 141                 { 'message|m=s' => \$_message,
 142                   'dry-run|n' => \$_dry_run } ],
 143        'set-tree' => [ \&cmd_set_tree,
 144                        "Set an SVN repository to a git tree-ish",
 145                        { 'stdin|' => \$_stdin, %cmt_opts, %fc_opts, } ],
 146        'create-ignore' => [ \&cmd_create_ignore,
 147                             'Create a .gitignore per svn:ignore',
 148                             { 'revision|r=i' => \$_revision
 149                             } ],
 150        'propget' => [ \&cmd_propget,
 151                       'Print the value of a property on a file or directory',
 152                       { 'revision|r=i' => \$_revision } ],
 153        'proplist' => [ \&cmd_proplist,
 154                       'List all properties of a file or directory',
 155                       { 'revision|r=i' => \$_revision } ],
 156        'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
 157                        { 'revision|r=i' => \$_revision
 158                        } ],
 159        'show-externals' => [ \&cmd_show_externals, "Show svn:externals listings",
 160                        { 'revision|r=i' => \$_revision
 161                        } ],
 162        'multi-fetch' => [ \&cmd_multi_fetch,
 163                           "Deprecated alias for $0 fetch --all",
 164                           { 'revision|r=s' => \$_revision, %fc_opts } ],
 165        'migrate' => [ sub { },
 166                       # no-op, we automatically run this anyways,
 167                       'Migrate configuration/metadata/layout from
 168                        previous versions of git-svn',
 169                       { 'minimize' => \$Git::SVN::Migration::_minimize,
 170                         %remote_opts } ],
 171        'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
 172                        { 'limit=i' => \$Git::SVN::Log::limit,
 173                          'revision|r=s' => \$_revision,
 174                          'verbose|v' => \$Git::SVN::Log::verbose,
 175                          'incremental' => \$Git::SVN::Log::incremental,
 176                          'oneline' => \$Git::SVN::Log::oneline,
 177                          'show-commit' => \$Git::SVN::Log::show_commit,
 178                          'non-recursive' => \$Git::SVN::Log::non_recursive,
 179                          'authors-file|A=s' => \$_authors,
 180                          'color' => \$Git::SVN::Log::color,
 181                          'pager=s' => \$Git::SVN::Log::pager
 182                        } ],
 183        'find-rev' => [ \&cmd_find_rev,
 184                        "Translate between SVN revision numbers and tree-ish",
 185                        {} ],
 186        'rebase' => [ \&cmd_rebase, "Fetch and rebase your working directory",
 187                        { 'merge|m|M' => \$_merge,
 188                          'verbose|v' => \$_verbose,
 189                          'strategy|s=s' => \$_strategy,
 190                          'local|l' => \$_local,
 191                          'fetch-all|all' => \$_fetch_all,
 192                          'dry-run|n' => \$_dry_run,
 193                          %fc_opts } ],
 194        'commit-diff' => [ \&cmd_commit_diff,
 195                           'Commit a diff between two trees',
 196                        { 'message|m=s' => \$_message,
 197                          'file|F=s' => \$_file,
 198                          'revision|r=s' => \$_revision,
 199                        %cmt_opts } ],
 200        'info' => [ \&cmd_info,
 201                    "Show info about the latest SVN revision
 202                     on the current branch",
 203                    { 'url' => \$_url, } ],
 204        'blame' => [ \&Git::SVN::Log::cmd_blame,
 205                    "Show what revision and author last modified each line of a file",
 206                    { 'git-format' => \$_git_format } ],
 207);
 208
 209my $cmd;
 210for (my $i = 0; $i < @ARGV; $i++) {
 211        if (defined $cmd{$ARGV[$i]}) {
 212                $cmd = $ARGV[$i];
 213                splice @ARGV, $i, 1;
 214                last;
 215        }
 216};
 217
 218# make sure we're always running at the top-level working directory
 219unless ($cmd && $cmd =~ /(?:clone|init|multi-init)$/) {
 220        unless (-d $ENV{GIT_DIR}) {
 221                if ($git_dir_user_set) {
 222                        die "GIT_DIR=$ENV{GIT_DIR} explicitly set, ",
 223                            "but it is not a directory\n";
 224                }
 225                my $git_dir = delete $ENV{GIT_DIR};
 226                chomp(my $cdup = command_oneline(qw/rev-parse --show-cdup/));
 227                unless (length $cdup) {
 228                        die "Already at toplevel, but $git_dir ",
 229                            "not found '$cdup'\n";
 230                }
 231                chdir $cdup or die "Unable to chdir up to '$cdup'\n";
 232                unless (-d $git_dir) {
 233                        die "$git_dir still not found after going to ",
 234                            "'$cdup'\n";
 235                }
 236                $ENV{GIT_DIR} = $git_dir;
 237        }
 238        $_repository = Git->repository(Repository => $ENV{GIT_DIR});
 239}
 240
 241my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
 242
 243read_repo_config(\%opts);
 244if ($cmd && ($cmd eq 'log' || $cmd eq 'blame')) {
 245        Getopt::Long::Configure('pass_through');
 246}
 247my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
 248                    'minimize-connections' => \$Git::SVN::Migration::_minimize,
 249                    'id|i=s' => \$Git::SVN::default_ref_id,
 250                    'svn-remote|remote|R=s' => sub {
 251                       $Git::SVN::no_reuse_existing = 1;
 252                       $Git::SVN::default_repo_id = $_[1] });
 253exit 1 if (!$rv && $cmd && $cmd ne 'log');
 254
 255usage(0) if $_help;
 256version() if $_version;
 257usage(1) unless defined $cmd;
 258load_authors() if $_authors;
 259
 260unless ($cmd =~ /^(?:clone|init|multi-init|commit-diff)$/) {
 261        Git::SVN::Migration::migration_check();
 262}
 263Git::SVN::init_vars();
 264eval {
 265        Git::SVN::verify_remotes_sanity();
 266        $cmd{$cmd}->[0]->(@ARGV);
 267};
 268fatal $@ if $@;
 269post_fetch_checkout();
 270exit 0;
 271
 272####################### primary functions ######################
 273sub usage {
 274        my $exit = shift || 0;
 275        my $fd = $exit ? \*STDERR : \*STDOUT;
 276        print $fd <<"";
 277git-svn - bidirectional operations between a single Subversion tree and git
 278Usage: git svn <command> [options] [arguments]\n
 279
 280        print $fd "Available commands:\n" unless $cmd;
 281
 282        foreach (sort keys %cmd) {
 283                next if $cmd && $cmd ne $_;
 284                next if /^multi-/; # don't show deprecated commands
 285                print $fd '  ',pack('A17',$_),$cmd{$_}->[1],"\n";
 286                foreach (sort keys %{$cmd{$_}->[2]}) {
 287                        # mixed-case options are for .git/config only
 288                        next if /[A-Z]/ && /^[a-z]+$/i;
 289                        # prints out arguments as they should be passed:
 290                        my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
 291                        print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
 292                                                        "--$_" : "-$_" }
 293                                                split /\|/,$_)," $x\n";
 294                }
 295        }
 296        print $fd <<"";
 297\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
 298arbitrary identifier if you're tracking multiple SVN branches/repositories in
 299one git repository and want to keep them separate.  See git-svn(1) for more
 300information.
 301
 302        exit $exit;
 303}
 304
 305sub version {
 306        print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
 307        exit 0;
 308}
 309
 310sub do_git_init_db {
 311        unless (-d $ENV{GIT_DIR}) {
 312                my @init_db = ('init');
 313                push @init_db, "--template=$_template" if defined $_template;
 314                if (defined $_shared) {
 315                        if ($_shared =~ /[a-z]/) {
 316                                push @init_db, "--shared=$_shared";
 317                        } else {
 318                                push @init_db, "--shared";
 319                        }
 320                }
 321                command_noisy(@init_db);
 322                $_repository = Git->repository(Repository => ".git");
 323        }
 324        my $set;
 325        my $pfx = "svn-remote.$Git::SVN::default_repo_id";
 326        foreach my $i (keys %icv) {
 327                die "'$set' and '$i' cannot both be set\n" if $set;
 328                next unless defined $icv{$i};
 329                command_noisy('config', "$pfx.$i", $icv{$i});
 330                $set = $i;
 331        }
 332}
 333
 334sub init_subdir {
 335        my $repo_path = shift or return;
 336        mkpath([$repo_path]) unless -d $repo_path;
 337        chdir $repo_path or die "Couldn't chdir to $repo_path: $!\n";
 338        $ENV{GIT_DIR} = '.git';
 339        $_repository = Git->repository(Repository => $ENV{GIT_DIR});
 340}
 341
 342sub cmd_clone {
 343        my ($url, $path) = @_;
 344        if (!defined $path &&
 345            (defined $_trunk || defined $_branches || defined $_tags ||
 346             defined $_stdlayout) &&
 347            $url !~ m#^[a-z\+]+://#) {
 348                $path = $url;
 349        }
 350        $path = basename($url) if !defined $path || !length $path;
 351        cmd_init($url, $path);
 352        Git::SVN::fetch_all($Git::SVN::default_repo_id);
 353}
 354
 355sub cmd_init {
 356        if (defined $_stdlayout) {
 357                $_trunk = 'trunk' if (!defined $_trunk);
 358                $_tags = 'tags' if (!defined $_tags);
 359                $_branches = 'branches' if (!defined $_branches);
 360        }
 361        if (defined $_trunk || defined $_branches || defined $_tags) {
 362                return cmd_multi_init(@_);
 363        }
 364        my $url = shift or die "SVN repository location required ",
 365                               "as a command-line argument\n";
 366        init_subdir(@_);
 367        do_git_init_db();
 368
 369        Git::SVN->init($url);
 370}
 371
 372sub cmd_fetch {
 373        if (grep /^\d+=./, @_) {
 374                die "'<rev>=<commit>' fetch arguments are ",
 375                    "no longer supported.\n";
 376        }
 377        my ($remote) = @_;
 378        if (@_ > 1) {
 379                die "Usage: $0 fetch [--all] [svn-remote]\n";
 380        }
 381        $remote ||= $Git::SVN::default_repo_id;
 382        if ($_fetch_all) {
 383                cmd_multi_fetch();
 384        } else {
 385                Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
 386        }
 387}
 388
 389sub cmd_set_tree {
 390        my (@commits) = @_;
 391        if ($_stdin || !@commits) {
 392                print "Reading from stdin...\n";
 393                @commits = ();
 394                while (<STDIN>) {
 395                        if (/\b($sha1_short)\b/o) {
 396                                unshift @commits, $1;
 397                        }
 398                }
 399        }
 400        my @revs;
 401        foreach my $c (@commits) {
 402                my @tmp = command('rev-parse',$c);
 403                if (scalar @tmp == 1) {
 404                        push @revs, $tmp[0];
 405                } elsif (scalar @tmp > 1) {
 406                        push @revs, reverse(command('rev-list',@tmp));
 407                } else {
 408                        fatal "Failed to rev-parse $c";
 409                }
 410        }
 411        my $gs = Git::SVN->new;
 412        my ($r_last, $cmt_last) = $gs->last_rev_commit;
 413        $gs->fetch;
 414        if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
 415                fatal "There are new revisions that were fetched ",
 416                      "and need to be merged (or acknowledged) ",
 417                      "before committing.\nlast rev: $r_last\n",
 418                      " current: $gs->{last_rev}";
 419        }
 420        $gs->set_tree($_) foreach @revs;
 421        print "Done committing ",scalar @revs," revisions to SVN\n";
 422        unlink $gs->{index};
 423}
 424
 425sub cmd_dcommit {
 426        my $head = shift;
 427        git_cmd_try { command_oneline(qw/diff-index --quiet HEAD/) }
 428                'Cannot dcommit with a dirty index.  Commit your changes first, '
 429                . "or stash them with `git stash'.\n";
 430        $head ||= 'HEAD';
 431        my @refs;
 432        my ($url, $rev, $uuid, $gs) = working_head_info($head, \@refs);
 433        unless ($gs) {
 434                die "Unable to determine upstream SVN information from ",
 435                    "$head history.\nPerhaps the repository is empty.";
 436        }
 437        $url = defined $_commit_url ? $_commit_url : $gs->full_url;
 438        my $last_rev = $_revision if defined $_revision;
 439        if ($url) {
 440                print "Committing to $url ...\n";
 441        }
 442        my ($linear_refs, $parents) = linearize_history($gs, \@refs);
 443        if ($_no_rebase && scalar(@$linear_refs) > 1) {
 444                warn "Attempting to commit more than one change while ",
 445                     "--no-rebase is enabled.\n",
 446                     "If these changes depend on each other, re-running ",
 447                     "without --no-rebase may be required."
 448        }
 449        my $expect_url = $url;
 450        Git::SVN::remove_username($expect_url);
 451        while (1) {
 452                my $d = shift @$linear_refs or last;
 453                unless (defined $last_rev) {
 454                        (undef, $last_rev, undef) = cmt_metadata("$d~1");
 455                        unless (defined $last_rev) {
 456                                fatal "Unable to extract revision information ",
 457                                      "from commit $d~1";
 458                        }
 459                }
 460                if ($_dry_run) {
 461                        print "diff-tree $d~1 $d\n";
 462                } else {
 463                        my $cmt_rev;
 464                        my %ed_opts = ( r => $last_rev,
 465                                        log => get_commit_entry($d)->{log},
 466                                        ra => Git::SVN::Ra->new($url),
 467                                        config => SVN::Core::config_get_config(
 468                                                $Git::SVN::Ra::config_dir
 469                                        ),
 470                                        tree_a => "$d~1",
 471                                        tree_b => $d,
 472                                        editor_cb => sub {
 473                                               print "Committed r$_[0]\n";
 474                                               $cmt_rev = $_[0];
 475                                        },
 476                                        svn_path => '');
 477                        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
 478                                print "No changes\n$d~1 == $d\n";
 479                        } elsif ($parents->{$d} && @{$parents->{$d}}) {
 480                                $gs->{inject_parents_dcommit}->{$cmt_rev} =
 481                                                               $parents->{$d};
 482                        }
 483                        $_fetch_all ? $gs->fetch_all : $gs->fetch;
 484                        $last_rev = $cmt_rev;
 485                        next if $_no_rebase;
 486
 487                        # we always want to rebase against the current HEAD,
 488                        # not any head that was passed to us
 489                        my @diff = command('diff-tree', $d,
 490                                           $gs->refname, '--');
 491                        my @finish;
 492                        if (@diff) {
 493                                @finish = rebase_cmd();
 494                                print STDERR "W: $d and ", $gs->refname,
 495                                             " differ, using @finish:\n",
 496                                             join("\n", @diff), "\n";
 497                        } else {
 498                                print "No changes between current HEAD and ",
 499                                      $gs->refname,
 500                                      "\nResetting to the latest ",
 501                                      $gs->refname, "\n";
 502                                @finish = qw/reset --mixed/;
 503                        }
 504                        command_noisy(@finish, $gs->refname);
 505                        if (@diff) {
 506                                @refs = ();
 507                                my ($url_, $rev_, $uuid_, $gs_) =
 508                                              working_head_info($head, \@refs);
 509                                my ($linear_refs_, $parents_) =
 510                                              linearize_history($gs_, \@refs);
 511                                if (scalar(@$linear_refs) !=
 512                                    scalar(@$linear_refs_)) {
 513                                        fatal "# of revisions changed ",
 514                                          "\nbefore:\n",
 515                                          join("\n", @$linear_refs),
 516                                          "\n\nafter:\n",
 517                                          join("\n", @$linear_refs_), "\n",
 518                                          'If you are attempting to commit ',
 519                                          "merges, try running:\n\t",
 520                                          'git rebase --interactive',
 521                                          '--preserve-merges ',
 522                                          $gs->refname,
 523                                          "\nBefore dcommitting";
 524                                }
 525                                if ($url_ ne $expect_url) {
 526                                        fatal "URL mismatch after rebase: ",
 527                                              "$url_ != $expect_url";
 528                                }
 529                                if ($uuid_ ne $uuid) {
 530                                        fatal "uuid mismatch after rebase: ",
 531                                              "$uuid_ != $uuid";
 532                                }
 533                                # remap parents
 534                                my (%p, @l, $i);
 535                                for ($i = 0; $i < scalar @$linear_refs; $i++) {
 536                                        my $new = $linear_refs_->[$i] or next;
 537                                        $p{$new} =
 538                                                $parents->{$linear_refs->[$i]};
 539                                        push @l, $new;
 540                                }
 541                                $parents = \%p;
 542                                $linear_refs = \@l;
 543                        }
 544                }
 545        }
 546        unlink $gs->{index};
 547}
 548
 549sub cmd_branch {
 550        my ($branch_name, $head) = @_;
 551
 552        unless (defined $branch_name && length $branch_name) {
 553                die(($_tag ? "tag" : "branch") . " name required\n");
 554        }
 555        $head ||= 'HEAD';
 556
 557        my ($src, $rev, undef, $gs) = working_head_info($head);
 558
 559        my $remote = Git::SVN::read_all_remotes()->{svn};
 560        my $glob = $remote->{ $_tag ? 'tags' : 'branches' };
 561        my ($lft, $rgt) = @{ $glob->{path} }{qw/left right/};
 562        my $dst = join '/', $remote->{url}, $lft, $branch_name, ($rgt || ());
 563
 564        my $ctx = SVN::Client->new(
 565                auth    => Git::SVN::Ra::_auth_providers(),
 566                log_msg => sub {
 567                        ${ $_[0] } = defined $_message
 568                                ? $_message
 569                                : 'Create ' . ($_tag ? 'tag ' : 'branch ' )
 570                                . $branch_name;
 571                },
 572        );
 573
 574        eval {
 575                $ctx->ls($dst, 'HEAD', 0);
 576        } and die "branch ${branch_name} already exists\n";
 577
 578        print "Copying ${src} at r${rev} to ${dst}...\n";
 579        $ctx->copy($src, $rev, $dst)
 580                unless $_dry_run;
 581
 582        $gs->fetch_all;
 583}
 584
 585sub cmd_find_rev {
 586        my $revision_or_hash = shift or die "SVN or git revision required ",
 587                                            "as a command-line argument\n";
 588        my $result;
 589        if ($revision_or_hash =~ /^r\d+$/) {
 590                my $head = shift;
 591                $head ||= 'HEAD';
 592                my @refs;
 593                my (undef, undef, $uuid, $gs) = working_head_info($head, \@refs);
 594                unless ($gs) {
 595                        die "Unable to determine upstream SVN information from ",
 596                            "$head history\n";
 597                }
 598                my $desired_revision = substr($revision_or_hash, 1);
 599                $result = $gs->rev_map_get($desired_revision, $uuid);
 600        } else {
 601                my (undef, $rev, undef) = cmt_metadata($revision_or_hash);
 602                $result = $rev;
 603        }
 604        print "$result\n" if $result;
 605}
 606
 607sub cmd_rebase {
 608        command_noisy(qw/update-index --refresh/);
 609        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 610        unless ($gs) {
 611                die "Unable to determine upstream SVN information from ",
 612                    "working tree history\n";
 613        }
 614        if ($_dry_run) {
 615                print "Remote Branch: " . $gs->refname . "\n";
 616                print "SVN URL: " . $url . "\n";
 617                return;
 618        }
 619        if (command(qw/diff-index HEAD --/)) {
 620                print STDERR "Cannot rebase with uncommited changes:\n";
 621                command_noisy('status');
 622                exit 1;
 623        }
 624        unless ($_local) {
 625                # rebase will checkout for us, so no need to do it explicitly
 626                $_no_checkout = 'true';
 627                $_fetch_all ? $gs->fetch_all : $gs->fetch;
 628        }
 629        command_noisy(rebase_cmd(), $gs->refname);
 630}
 631
 632sub cmd_show_ignore {
 633        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 634        $gs ||= Git::SVN->new;
 635        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 636        $gs->prop_walk($gs->{path}, $r, sub {
 637                my ($gs, $path, $props) = @_;
 638                print STDOUT "\n# $path\n";
 639                my $s = $props->{'svn:ignore'} or return;
 640                $s =~ s/[\r\n]+/\n/g;
 641                chomp $s;
 642                $s =~ s#^#$path#gm;
 643                print STDOUT "$s\n";
 644        });
 645}
 646
 647sub cmd_show_externals {
 648        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 649        $gs ||= Git::SVN->new;
 650        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 651        $gs->prop_walk($gs->{path}, $r, sub {
 652                my ($gs, $path, $props) = @_;
 653                print STDOUT "\n# $path\n";
 654                my $s = $props->{'svn:externals'} or return;
 655                $s =~ s/[\r\n]+/\n/g;
 656                chomp $s;
 657                $s =~ s#^#$path#gm;
 658                print STDOUT "$s\n";
 659        });
 660}
 661
 662sub cmd_create_ignore {
 663        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 664        $gs ||= Git::SVN->new;
 665        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 666        $gs->prop_walk($gs->{path}, $r, sub {
 667                my ($gs, $path, $props) = @_;
 668                # $path is of the form /path/to/dir/
 669                my $ignore = '.' . $path . '.gitignore';
 670                my $s = $props->{'svn:ignore'} or return;
 671                open(GITIGNORE, '>', $ignore)
 672                  or fatal("Failed to open `$ignore' for writing: $!");
 673                $s =~ s/[\r\n]+/\n/g;
 674                chomp $s;
 675                # Prefix all patterns so that the ignore doesn't apply
 676                # to sub-directories.
 677                $s =~ s#^#/#gm;
 678                print GITIGNORE "$s\n";
 679                close(GITIGNORE)
 680                  or fatal("Failed to close `$ignore': $!");
 681                command_noisy('add', '-f', $ignore);
 682        });
 683}
 684
 685sub canonicalize_path {
 686        my ($path) = @_;
 687        my $dot_slash_added = 0;
 688        if (substr($path, 0, 1) ne "/") {
 689                $path = "./" . $path;
 690                $dot_slash_added = 1;
 691        }
 692        # File::Spec->canonpath doesn't collapse x/../y into y (for a
 693        # good reason), so let's do this manually.
 694        $path =~ s#/+#/#g;
 695        $path =~ s#/\.(?:/|$)#/#g;
 696        $path =~ s#/[^/]+/\.\.##g;
 697        $path =~ s#/$##g;
 698        $path =~ s#^\./## if $dot_slash_added;
 699        $path =~ s#^/##;
 700        $path =~ s#^\.$##;
 701        return $path;
 702}
 703
 704# get_svnprops(PATH)
 705# ------------------
 706# Helper for cmd_propget and cmd_proplist below.
 707sub get_svnprops {
 708        my $path = shift;
 709        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 710        $gs ||= Git::SVN->new;
 711
 712        # prefix THE PATH by the sub-directory from which the user
 713        # invoked us.
 714        $path = $cmd_dir_prefix . $path;
 715        fatal("No such file or directory: $path") unless -e $path;
 716        my $is_dir = -d $path ? 1 : 0;
 717        $path = $gs->{path} . '/' . $path;
 718
 719        # canonicalize the path (otherwise libsvn will abort or fail to
 720        # find the file)
 721        $path = canonicalize_path($path);
 722
 723        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 724        my $props;
 725        if ($is_dir) {
 726                (undef, undef, $props) = $gs->ra->get_dir($path, $r);
 727        }
 728        else {
 729                (undef, $props) = $gs->ra->get_file($path, $r, undef);
 730        }
 731        return $props;
 732}
 733
 734# cmd_propget (PROP, PATH)
 735# ------------------------
 736# Print the SVN property PROP for PATH.
 737sub cmd_propget {
 738        my ($prop, $path) = @_;
 739        $path = '.' if not defined $path;
 740        usage(1) if not defined $prop;
 741        my $props = get_svnprops($path);
 742        if (not defined $props->{$prop}) {
 743                fatal("`$path' does not have a `$prop' SVN property.");
 744        }
 745        print $props->{$prop} . "\n";
 746}
 747
 748# cmd_proplist (PATH)
 749# -------------------
 750# Print the list of SVN properties for PATH.
 751sub cmd_proplist {
 752        my $path = shift;
 753        $path = '.' if not defined $path;
 754        my $props = get_svnprops($path);
 755        print "Properties on '$path':\n";
 756        foreach (sort keys %{$props}) {
 757                print "  $_\n";
 758        }
 759}
 760
 761sub cmd_multi_init {
 762        my $url = shift;
 763        unless (defined $_trunk || defined $_branches || defined $_tags) {
 764                usage(1);
 765        }
 766
 767        # there are currently some bugs that prevent multi-init/multi-fetch
 768        # setups from working well without this.
 769        $Git::SVN::_minimize_url = 1;
 770
 771        $_prefix = '' unless defined $_prefix;
 772        if (defined $url) {
 773                $url =~ s#/+$##;
 774                init_subdir(@_);
 775        }
 776        do_git_init_db();
 777        if (defined $_trunk) {
 778                my $trunk_ref = $_prefix . 'trunk';
 779                # try both old-style and new-style lookups:
 780                my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
 781                unless ($gs_trunk) {
 782                        my ($trunk_url, $trunk_path) =
 783                                              complete_svn_url($url, $_trunk);
 784                        $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
 785                                                   undef, $trunk_ref);
 786                }
 787        }
 788        return unless defined $_branches || defined $_tags;
 789        my $ra = $url ? Git::SVN::Ra->new($url) : undef;
 790        complete_url_ls_init($ra, $_branches, '--branches/-b', $_prefix);
 791        complete_url_ls_init($ra, $_tags, '--tags/-t', $_prefix . 'tags/');
 792}
 793
 794sub cmd_multi_fetch {
 795        my $remotes = Git::SVN::read_all_remotes();
 796        foreach my $repo_id (sort keys %$remotes) {
 797                if ($remotes->{$repo_id}->{url}) {
 798                        Git::SVN::fetch_all($repo_id, $remotes);
 799                }
 800        }
 801}
 802
 803# this command is special because it requires no metadata
 804sub cmd_commit_diff {
 805        my ($ta, $tb, $url) = @_;
 806        my $usage = "Usage: $0 commit-diff -r<revision> ".
 807                    "<tree-ish> <tree-ish> [<URL>]";
 808        fatal($usage) if (!defined $ta || !defined $tb);
 809        my $svn_path = '';
 810        if (!defined $url) {
 811                my $gs = eval { Git::SVN->new };
 812                if (!$gs) {
 813                        fatal("Needed URL or usable git-svn --id in ",
 814                              "the command-line\n", $usage);
 815                }
 816                $url = $gs->{url};
 817                $svn_path = $gs->{path};
 818        }
 819        unless (defined $_revision) {
 820                fatal("-r|--revision is a required argument\n", $usage);
 821        }
 822        if (defined $_message && defined $_file) {
 823                fatal("Both --message/-m and --file/-F specified ",
 824                      "for the commit message.\n",
 825                      "I have no idea what you mean");
 826        }
 827        if (defined $_file) {
 828                $_message = file_to_s($_file);
 829        } else {
 830                $_message ||= get_commit_entry($tb)->{log};
 831        }
 832        my $ra ||= Git::SVN::Ra->new($url);
 833        my $r = $_revision;
 834        if ($r eq 'HEAD') {
 835                $r = $ra->get_latest_revnum;
 836        } elsif ($r !~ /^\d+$/) {
 837                die "revision argument: $r not understood by git-svn\n";
 838        }
 839        my %ed_opts = ( r => $r,
 840                        log => $_message,
 841                        ra => $ra,
 842                        tree_a => $ta,
 843                        tree_b => $tb,
 844                        editor_cb => sub { print "Committed r$_[0]\n" },
 845                        svn_path => $svn_path );
 846        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
 847                print "No changes\n$ta == $tb\n";
 848        }
 849}
 850
 851sub escape_uri_only {
 852        my ($uri) = @_;
 853        my @tmp;
 854        foreach (split m{/}, $uri) {
 855                s/([^\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
 856                push @tmp, $_;
 857        }
 858        join('/', @tmp);
 859}
 860
 861sub escape_url {
 862        my ($url) = @_;
 863        if ($url =~ m#^([^:]+)://([^/]*)(.*)$#) {
 864                my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
 865                $url = "$scheme://$domain$uri";
 866        }
 867        $url;
 868}
 869
 870sub cmd_info {
 871        my $path = canonicalize_path(defined($_[0]) ? $_[0] : ".");
 872        my $fullpath = canonicalize_path($cmd_dir_prefix . $path);
 873        if (exists $_[1]) {
 874                die "Too many arguments specified\n";
 875        }
 876
 877        my ($file_type, $diff_status) = find_file_type_and_diff_status($path);
 878
 879        if (!$file_type && !$diff_status) {
 880                print STDERR "svn: '$path' is not under version control\n";
 881                exit 1;
 882        }
 883
 884        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 885        unless ($gs) {
 886                die "Unable to determine upstream SVN information from ",
 887                    "working tree history\n";
 888        }
 889
 890        # canonicalize_path() will return "" to make libsvn 1.5.x happy,
 891        $path = "." if $path eq "";
 892
 893        my $full_url = $url . ($fullpath eq "" ? "" : "/$fullpath");
 894
 895        if ($_url) {
 896                print escape_url($full_url), "\n";
 897                return;
 898        }
 899
 900        my $result = "Path: $path\n";
 901        $result .= "Name: " . basename($path) . "\n" if $file_type ne "dir";
 902        $result .= "URL: " . escape_url($full_url) . "\n";
 903
 904        eval {
 905                my $repos_root = $gs->repos_root;
 906                Git::SVN::remove_username($repos_root);
 907                $result .= "Repository Root: " . escape_url($repos_root) . "\n";
 908        };
 909        if ($@) {
 910                $result .= "Repository Root: (offline)\n";
 911        }
 912        $result .= "Repository UUID: $uuid\n" unless $diff_status eq "A";
 913        $result .= "Revision: " . ($diff_status eq "A" ? 0 : $rev) . "\n";
 914
 915        $result .= "Node Kind: " .
 916                   ($file_type eq "dir" ? "directory" : "file") . "\n";
 917
 918        my $schedule = $diff_status eq "A"
 919                       ? "add"
 920                       : ($diff_status eq "D" ? "delete" : "normal");
 921        $result .= "Schedule: $schedule\n";
 922
 923        if ($diff_status eq "A") {
 924                print $result, "\n";
 925                return;
 926        }
 927
 928        my ($lc_author, $lc_rev, $lc_date_utc);
 929        my @args = Git::SVN::Log::git_svn_log_cmd($rev, $rev, "--", $fullpath);
 930        my $log = command_output_pipe(@args);
 931        my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
 932        while (<$log>) {
 933                if (/^${esc_color}author (.+) <[^>]+> (\d+) ([\-\+]?\d+)$/o) {
 934                        $lc_author = $1;
 935                        $lc_date_utc = Git::SVN::Log::parse_git_date($2, $3);
 936                } elsif (/^${esc_color}    (git-svn-id:.+)$/o) {
 937                        (undef, $lc_rev, undef) = ::extract_metadata($1);
 938                }
 939        }
 940        close $log;
 941
 942        Git::SVN::Log::set_local_timezone();
 943
 944        $result .= "Last Changed Author: $lc_author\n";
 945        $result .= "Last Changed Rev: $lc_rev\n";
 946        $result .= "Last Changed Date: " .
 947                   Git::SVN::Log::format_svn_date($lc_date_utc) . "\n";
 948
 949        if ($file_type ne "dir") {
 950                my $text_last_updated_date =
 951                    ($diff_status eq "D" ? $lc_date_utc : (stat $path)[9]);
 952                $result .=
 953                    "Text Last Updated: " .
 954                    Git::SVN::Log::format_svn_date($text_last_updated_date) .
 955                    "\n";
 956                my $checksum;
 957                if ($diff_status eq "D") {
 958                        my ($fh, $ctx) =
 959                            command_output_pipe(qw(cat-file blob), "HEAD:$path");
 960                        if ($file_type eq "link") {
 961                                my $file_name = <$fh>;
 962                                $checksum = md5sum("link $file_name");
 963                        } else {
 964                                $checksum = md5sum($fh);
 965                        }
 966                        command_close_pipe($fh, $ctx);
 967                } elsif ($file_type eq "link") {
 968                        my $file_name =
 969                            command(qw(cat-file blob), "HEAD:$path");
 970                        $checksum =
 971                            md5sum("link " . $file_name);
 972                } else {
 973                        open FILE, "<", $path or die $!;
 974                        $checksum = md5sum(\*FILE);
 975                        close FILE or die $!;
 976                }
 977                $result .= "Checksum: " . $checksum . "\n";
 978        }
 979
 980        print $result, "\n";
 981}
 982
 983########################### utility functions #########################
 984
 985sub rebase_cmd {
 986        my @cmd = qw/rebase/;
 987        push @cmd, '-v' if $_verbose;
 988        push @cmd, qw/--merge/ if $_merge;
 989        push @cmd, "--strategy=$_strategy" if $_strategy;
 990        @cmd;
 991}
 992
 993sub post_fetch_checkout {
 994        return if $_no_checkout;
 995        my $gs = $Git::SVN::_head or return;
 996        return if verify_ref('refs/heads/master^0');
 997
 998        my $valid_head = verify_ref('HEAD^0');
 999        command_noisy(qw(update-ref refs/heads/master), $gs->refname);
1000        return if ($valid_head || !verify_ref('HEAD^0'));
1001
1002        return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
1003        my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
1004        return if -f $index;
1005
1006        return if command_oneline(qw/rev-parse --is-inside-work-tree/) eq 'false';
1007        return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
1008        command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
1009        print STDERR "Checked out HEAD:\n  ",
1010                     $gs->full_url, " r", $gs->last_rev, "\n";
1011}
1012
1013sub complete_svn_url {
1014        my ($url, $path) = @_;
1015        $path =~ s#/+$##;
1016        if ($path !~ m#^[a-z\+]+://#) {
1017                if (!defined $url || $url !~ m#^[a-z\+]+://#) {
1018                        fatal("E: '$path' is not a complete URL ",
1019                              "and a separate URL is not specified");
1020                }
1021                return ($url, $path);
1022        }
1023        return ($path, '');
1024}
1025
1026sub complete_url_ls_init {
1027        my ($ra, $repo_path, $switch, $pfx) = @_;
1028        unless ($repo_path) {
1029                print STDERR "W: $switch not specified\n";
1030                return;
1031        }
1032        $repo_path =~ s#/+$##;
1033        if ($repo_path =~ m#^[a-z\+]+://#) {
1034                $ra = Git::SVN::Ra->new($repo_path);
1035                $repo_path = '';
1036        } else {
1037                $repo_path =~ s#^/+##;
1038                unless ($ra) {
1039                        fatal("E: '$repo_path' is not a complete URL ",
1040                              "and a separate URL is not specified");
1041                }
1042        }
1043        my $url = $ra->{url};
1044        my $gs = Git::SVN->init($url, undef, undef, undef, 1);
1045        my $k = "svn-remote.$gs->{repo_id}.url";
1046        my $orig_url = eval { command_oneline(qw/config --get/, $k) };
1047        if ($orig_url && ($orig_url ne $gs->{url})) {
1048                die "$k already set: $orig_url\n",
1049                    "wanted to set to: $gs->{url}\n";
1050        }
1051        command_oneline('config', $k, $gs->{url}) unless $orig_url;
1052        my $remote_path = "$ra->{svn_path}/$repo_path";
1053        $remote_path =~ s#/+#/#g;
1054        $remote_path =~ s#^/##g;
1055        $remote_path .= "/*" if $remote_path !~ /\*/;
1056        my ($n) = ($switch =~ /^--(\w+)/);
1057        if (length $pfx && $pfx !~ m#/$#) {
1058                die "--prefix='$pfx' must have a trailing slash '/'\n";
1059        }
1060        command_noisy('config',
1061                      "svn-remote.$gs->{repo_id}.$n",
1062                      "$remote_path:refs/remotes/$pfx*" .
1063                        ('/*' x (($remote_path =~ tr/*/*/) - 1)) );
1064}
1065
1066sub verify_ref {
1067        my ($ref) = @_;
1068        eval { command_oneline([ 'rev-parse', '--verify', $ref ],
1069                               { STDERR => 0 }); };
1070}
1071
1072sub get_tree_from_treeish {
1073        my ($treeish) = @_;
1074        # $treeish can be a symbolic ref, too:
1075        my $type = command_oneline(qw/cat-file -t/, $treeish);
1076        my $expected;
1077        while ($type eq 'tag') {
1078                ($treeish, $type) = command(qw/cat-file tag/, $treeish);
1079        }
1080        if ($type eq 'commit') {
1081                $expected = (grep /^tree /, command(qw/cat-file commit/,
1082                                                    $treeish))[0];
1083                ($expected) = ($expected =~ /^tree ($sha1)$/o);
1084                die "Unable to get tree from $treeish\n" unless $expected;
1085        } elsif ($type eq 'tree') {
1086                $expected = $treeish;
1087        } else {
1088                die "$treeish is a $type, expected tree, tag or commit\n";
1089        }
1090        return $expected;
1091}
1092
1093sub get_commit_entry {
1094        my ($treeish) = shift;
1095        my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
1096        my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
1097        my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
1098        open my $log_fh, '>', $commit_editmsg or croak $!;
1099
1100        my $type = command_oneline(qw/cat-file -t/, $treeish);
1101        if ($type eq 'commit' || $type eq 'tag') {
1102                my ($msg_fh, $ctx) = command_output_pipe('cat-file',
1103                                                         $type, $treeish);
1104                my $in_msg = 0;
1105                my $author;
1106                my $saw_from = 0;
1107                my $msgbuf = "";
1108                while (<$msg_fh>) {
1109                        if (!$in_msg) {
1110                                $in_msg = 1 if (/^\s*$/);
1111                                $author = $1 if (/^author (.*>)/);
1112                        } elsif (/^git-svn-id: /) {
1113                                # skip this for now, we regenerate the
1114                                # correct one on re-fetch anyways
1115                                # TODO: set *:merge properties or like...
1116                        } else {
1117                                if (/^From:/ || /^Signed-off-by:/) {
1118                                        $saw_from = 1;
1119                                }
1120                                $msgbuf .= $_;
1121                        }
1122                }
1123                $msgbuf =~ s/\s+$//s;
1124                if ($Git::SVN::_add_author_from && defined($author)
1125                    && !$saw_from) {
1126                        $msgbuf .= "\n\nFrom: $author";
1127                }
1128                print $log_fh $msgbuf or croak $!;
1129                command_close_pipe($msg_fh, $ctx);
1130        }
1131        close $log_fh or croak $!;
1132
1133        if ($_edit || ($type eq 'tree')) {
1134                my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
1135                # TODO: strip out spaces, comments, like git-commit.sh
1136                system($editor, $commit_editmsg);
1137        }
1138        rename $commit_editmsg, $commit_msg or croak $!;
1139        open $log_fh, '<', $commit_msg or croak $!;
1140        { local $/; chomp($log_entry{log} = <$log_fh>); }
1141        close $log_fh or croak $!;
1142        unlink $commit_msg;
1143        \%log_entry;
1144}
1145
1146sub s_to_file {
1147        my ($str, $file, $mode) = @_;
1148        open my $fd,'>',$file or croak $!;
1149        print $fd $str,"\n" or croak $!;
1150        close $fd or croak $!;
1151        chmod ($mode &~ umask, $file) if (defined $mode);
1152}
1153
1154sub file_to_s {
1155        my $file = shift;
1156        open my $fd,'<',$file or croak "$!: file: $file\n";
1157        local $/;
1158        my $ret = <$fd>;
1159        close $fd or croak $!;
1160        $ret =~ s/\s*$//s;
1161        return $ret;
1162}
1163
1164# '<svn username> = real-name <email address>' mapping based on git-svnimport:
1165sub load_authors {
1166        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1167        my $log = $cmd eq 'log';
1168        while (<$authors>) {
1169                chomp;
1170                next unless /^(.+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
1171                my ($user, $name, $email) = ($1, $2, $3);
1172                if ($log) {
1173                        $Git::SVN::Log::rusers{"$name <$email>"} = $user;
1174                } else {
1175                        $users{$user} = [$name, $email];
1176                }
1177        }
1178        close $authors or croak $!;
1179}
1180
1181# convert GetOpt::Long specs for use by git-config
1182sub read_repo_config {
1183        return unless -d $ENV{GIT_DIR};
1184        my $opts = shift;
1185        my @config_only;
1186        foreach my $o (keys %$opts) {
1187                # if we have mixedCase and a long option-only, then
1188                # it's a config-only variable that we don't need for
1189                # the command-line.
1190                push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
1191                my $v = $opts->{$o};
1192                my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
1193                $key =~ s/-//g;
1194                my $arg = 'git config';
1195                $arg .= ' --int' if ($o =~ /[:=]i$/);
1196                $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
1197                if (ref $v eq 'ARRAY') {
1198                        chomp(my @tmp = `$arg --get-all svn.$key`);
1199                        @$v = @tmp if @tmp;
1200                } else {
1201                        chomp(my $tmp = `$arg --get svn.$key`);
1202                        if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
1203                                $$v = $tmp;
1204                        }
1205                }
1206        }
1207        delete @$opts{@config_only} if @config_only;
1208}
1209
1210sub extract_metadata {
1211        my $id = shift or return (undef, undef, undef);
1212        my ($url, $rev, $uuid) = ($id =~ /^\s*git-svn-id:\s+(.*)\@(\d+)
1213                                                        \s([a-f\d\-]+)$/x);
1214        if (!defined $rev || !$uuid || !$url) {
1215                # some of the original repositories I made had
1216                # identifiers like this:
1217                ($rev, $uuid) = ($id =~/^\s*git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
1218        }
1219        return ($url, $rev, $uuid);
1220}
1221
1222sub cmt_metadata {
1223        return extract_metadata((grep(/^git-svn-id: /,
1224                command(qw/cat-file commit/, shift)))[-1]);
1225}
1226
1227sub working_head_info {
1228        my ($head, $refs) = @_;
1229        my @args = ('log', '--no-color', '--first-parent', '--pretty=medium');
1230        my ($fh, $ctx) = command_output_pipe(@args, $head);
1231        my $hash;
1232        my %max;
1233        while (<$fh>) {
1234                if ( m{^commit ($::sha1)$} ) {
1235                        unshift @$refs, $hash if $hash and $refs;
1236                        $hash = $1;
1237                        next;
1238                }
1239                next unless s{^\s*(git-svn-id:)}{$1};
1240                my ($url, $rev, $uuid) = extract_metadata($_);
1241                if (defined $url && defined $rev) {
1242                        next if $max{$url} and $max{$url} < $rev;
1243                        if (my $gs = Git::SVN->find_by_url($url)) {
1244                                my $c = $gs->rev_map_get($rev, $uuid);
1245                                if ($c && $c eq $hash) {
1246                                        close $fh; # break the pipe
1247                                        return ($url, $rev, $uuid, $gs);
1248                                } else {
1249                                        $max{$url} ||= $gs->rev_map_max;
1250                                }
1251                        }
1252                }
1253        }
1254        command_close_pipe($fh, $ctx);
1255        (undef, undef, undef, undef);
1256}
1257
1258sub read_commit_parents {
1259        my ($parents, $c) = @_;
1260        chomp(my $p = command_oneline(qw/rev-list --parents -1/, $c));
1261        $p =~ s/^($c)\s*// or die "rev-list --parents -1 $c failed!\n";
1262        @{$parents->{$c}} = split(/ /, $p);
1263}
1264
1265sub linearize_history {
1266        my ($gs, $refs) = @_;
1267        my %parents;
1268        foreach my $c (@$refs) {
1269                read_commit_parents(\%parents, $c);
1270        }
1271
1272        my @linear_refs;
1273        my %skip = ();
1274        my $last_svn_commit = $gs->last_commit;
1275        foreach my $c (reverse @$refs) {
1276                next if $c eq $last_svn_commit;
1277                last if $skip{$c};
1278
1279                unshift @linear_refs, $c;
1280                $skip{$c} = 1;
1281
1282                # we only want the first parent to diff against for linear
1283                # history, we save the rest to inject when we finalize the
1284                # svn commit
1285                my $fp_a = verify_ref("$c~1");
1286                my $fp_b = shift @{$parents{$c}} if $parents{$c};
1287                if (!$fp_a || !$fp_b) {
1288                        die "Commit $c\n",
1289                            "has no parent commit, and therefore ",
1290                            "nothing to diff against.\n",
1291                            "You should be working from a repository ",
1292                            "originally created by git-svn\n";
1293                }
1294                if ($fp_a ne $fp_b) {
1295                        die "$c~1 = $fp_a, however parsing commit $c ",
1296                            "revealed that:\n$c~1 = $fp_b\nBUG!\n";
1297                }
1298
1299                foreach my $p (@{$parents{$c}}) {
1300                        $skip{$p} = 1;
1301                }
1302        }
1303        (\@linear_refs, \%parents);
1304}
1305
1306sub find_file_type_and_diff_status {
1307        my ($path) = @_;
1308        return ('dir', '') if $path eq '';
1309
1310        my $diff_output =
1311            command_oneline(qw(diff --cached --name-status --), $path) || "";
1312        my $diff_status = (split(' ', $diff_output))[0] || "";
1313
1314        my $ls_tree = command_oneline(qw(ls-tree HEAD), $path) || "";
1315
1316        return (undef, undef) if !$diff_status && !$ls_tree;
1317
1318        if ($diff_status eq "A") {
1319                return ("link", $diff_status) if -l $path;
1320                return ("dir", $diff_status) if -d $path;
1321                return ("file", $diff_status);
1322        }
1323
1324        my $mode = (split(' ', $ls_tree))[0] || "";
1325
1326        return ("link", $diff_status) if $mode eq "120000";
1327        return ("dir", $diff_status) if $mode eq "040000";
1328        return ("file", $diff_status);
1329}
1330
1331sub md5sum {
1332        my $arg = shift;
1333        my $ref = ref $arg;
1334        my $md5 = Digest::MD5->new();
1335        if ($ref eq 'GLOB' || $ref eq 'IO::File' || $ref eq 'File::Temp') {
1336                $md5->addfile($arg) or croak $!;
1337        } elsif ($ref eq 'SCALAR') {
1338                $md5->add($$arg) or croak $!;
1339        } elsif (!$ref) {
1340                $md5->add($arg) or croak $!;
1341        } else {
1342                ::fatal "Can't provide MD5 hash for unknown ref type: '", $ref, "'";
1343        }
1344        return $md5->hexdigest();
1345}
1346
1347package Git::SVN;
1348use strict;
1349use warnings;
1350use Fcntl qw/:DEFAULT :seek/;
1351use constant rev_map_fmt => 'NH40';
1352use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
1353            $_repack $_repack_flags $_use_svm_props $_head
1354            $_use_svnsync_props $no_reuse_existing $_minimize_url
1355            $_use_log_author $_add_author_from/;
1356use Carp qw/croak/;
1357use File::Path qw/mkpath/;
1358use File::Copy qw/copy/;
1359use IPC::Open3;
1360
1361my ($_gc_nr, $_gc_period);
1362
1363# properties that we do not log:
1364my %SKIP_PROP;
1365BEGIN {
1366        %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
1367                                        svn:special svn:executable
1368                                        svn:entry:committed-rev
1369                                        svn:entry:last-author
1370                                        svn:entry:uuid
1371                                        svn:entry:committed-date/;
1372
1373        # some options are read globally, but can be overridden locally
1374        # per [svn-remote "..."] section.  Command-line options will *NOT*
1375        # override options set in an [svn-remote "..."] section
1376        no strict 'refs';
1377        for my $option (qw/follow_parent no_metadata use_svm_props
1378                           use_svnsync_props/) {
1379                my $key = $option;
1380                $key =~ tr/_//d;
1381                my $prop = "-$option";
1382                *$option = sub {
1383                        my ($self) = @_;
1384                        return $self->{$prop} if exists $self->{$prop};
1385                        my $k = "svn-remote.$self->{repo_id}.$key";
1386                        eval { command_oneline(qw/config --get/, $k) };
1387                        if ($@) {
1388                                $self->{$prop} = ${"Git::SVN::_$option"};
1389                        } else {
1390                                my $v = command_oneline(qw/config --bool/,$k);
1391                                $self->{$prop} = $v eq 'false' ? 0 : 1;
1392                        }
1393                        return $self->{$prop};
1394                }
1395        }
1396}
1397
1398
1399my (%LOCKFILES, %INDEX_FILES);
1400END {
1401        unlink keys %LOCKFILES if %LOCKFILES;
1402        unlink keys %INDEX_FILES if %INDEX_FILES;
1403}
1404
1405sub resolve_local_globs {
1406        my ($url, $fetch, $glob_spec) = @_;
1407        return unless defined $glob_spec;
1408        my $ref = $glob_spec->{ref};
1409        my $path = $glob_spec->{path};
1410        foreach (command(qw#for-each-ref --format=%(refname) refs/remotes#)) {
1411                next unless m#^refs/remotes/$ref->{regex}$#;
1412                my $p = $1;
1413                my $pathname = desanitize_refname($path->full_path($p));
1414                my $refname = desanitize_refname($ref->full_path($p));
1415                if (my $existing = $fetch->{$pathname}) {
1416                        if ($existing ne $refname) {
1417                                die "Refspec conflict:\n",
1418                                    "existing: refs/remotes/$existing\n",
1419                                    " globbed: refs/remotes/$refname\n";
1420                        }
1421                        my $u = (::cmt_metadata("refs/remotes/$refname"))[0];
1422                        $u =~ s!^\Q$url\E(/|$)!! or die
1423                          "refs/remotes/$refname: '$url' not found in '$u'\n";
1424                        if ($pathname ne $u) {
1425                                warn "W: Refspec glob conflict ",
1426                                     "(ref: refs/remotes/$refname):\n",
1427                                     "expected path: $pathname\n",
1428                                     "    real path: $u\n",
1429                                     "Continuing ahead with $u\n";
1430                                next;
1431                        }
1432                } else {
1433                        $fetch->{$pathname} = $refname;
1434                }
1435        }
1436}
1437
1438sub parse_revision_argument {
1439        my ($base, $head) = @_;
1440        if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
1441                return ($base, $head);
1442        }
1443        return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
1444        return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
1445        return ($head, $head) if ($::_revision eq 'HEAD');
1446        return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
1447        return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
1448        die "revision argument: $::_revision not understood by git-svn\n";
1449}
1450
1451sub fetch_all {
1452        my ($repo_id, $remotes) = @_;
1453        if (ref $repo_id) {
1454                my $gs = $repo_id;
1455                $repo_id = undef;
1456                $repo_id = $gs->{repo_id};
1457        }
1458        $remotes ||= read_all_remotes();
1459        my $remote = $remotes->{$repo_id} or
1460                     die "[svn-remote \"$repo_id\"] unknown\n";
1461        my $fetch = $remote->{fetch};
1462        my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
1463        my (@gs, @globs);
1464        my $ra = Git::SVN::Ra->new($url);
1465        my $uuid = $ra->get_uuid;
1466        my $head = $ra->get_latest_revnum;
1467        my $base = defined $fetch ? $head : 0;
1468
1469        # read the max revs for wildcard expansion (branches/*, tags/*)
1470        foreach my $t (qw/branches tags/) {
1471                defined $remote->{$t} or next;
1472                push @globs, $remote->{$t};
1473                my $max_rev = eval { tmp_config(qw/--int --get/,
1474                                         "svn-remote.$repo_id.${t}-maxRev") };
1475                if (defined $max_rev && ($max_rev < $base)) {
1476                        $base = $max_rev;
1477                } elsif (!defined $max_rev) {
1478                        $base = 0;
1479                }
1480        }
1481
1482        if ($fetch) {
1483                foreach my $p (sort keys %$fetch) {
1484                        my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
1485                        my $lr = $gs->rev_map_max;
1486                        if (defined $lr) {
1487                                $base = $lr if ($lr < $base);
1488                        }
1489                        push @gs, $gs;
1490                }
1491        }
1492
1493        ($base, $head) = parse_revision_argument($base, $head);
1494        $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
1495}
1496
1497sub read_all_remotes {
1498        my $r = {};
1499        my $use_svm_props = eval { command_oneline(qw/config --bool
1500            svn.useSvmProps/) };
1501        $use_svm_props = $use_svm_props eq 'true' if $use_svm_props;
1502        foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
1503                if (m!^(.+)\.fetch=\s*(.*)\s*:\s*(.+)\s*$!) {
1504                        my ($remote, $local_ref, $_remote_ref) = ($1, $2, $3);
1505                        die("svn-remote.$remote: remote ref '$_remote_ref' "
1506                            . "must start with 'refs/remotes/'\n")
1507                                unless $_remote_ref =~ m{^refs/remotes/(.+)};
1508                        my $remote_ref = $1;
1509                        $local_ref =~ s{^/}{};
1510                        $r->{$remote}->{fetch}->{$local_ref} = $remote_ref;
1511                        $r->{$remote}->{svm} = {} if $use_svm_props;
1512                } elsif (m!^(.+)\.usesvmprops=\s*(.*)\s*$!) {
1513                        $r->{$1}->{svm} = {};
1514                } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
1515                        $r->{$1}->{url} = $2;
1516                } elsif (m!^(.+)\.(branches|tags)=
1517                           (.*):refs/remotes/(.+)\s*$/!x) {
1518                        my ($p, $g) = ($3, $4);
1519                        my $rs = $r->{$1}->{$2} = {
1520                                          t => $2,
1521                                          remote => $1,
1522                                          path => Git::SVN::GlobSpec->new($p),
1523                                          ref => Git::SVN::GlobSpec->new($g) };
1524                        if (length($rs->{ref}->{right}) != 0) {
1525                                die "The '*' glob character must be the last ",
1526                                    "character of '$g'\n";
1527                        }
1528                }
1529        }
1530
1531        map {
1532                if (defined $r->{$_}->{svm}) {
1533                        my $svm;
1534                        eval {
1535                                my $section = "svn-remote.$_";
1536                                $svm = {
1537                                        source => tmp_config('--get',
1538                                            "$section.svm-source"),
1539                                        replace => tmp_config('--get',
1540                                            "$section.svm-replace"),
1541                                }
1542                        };
1543                        $r->{$_}->{svm} = $svm;
1544                }
1545        } keys %$r;
1546
1547        $r;
1548}
1549
1550sub init_vars {
1551        $_gc_nr = $_gc_period = 1000;
1552        if (defined $_repack || defined $_repack_flags) {
1553               warn "Repack options are obsolete; they have no effect.\n";
1554        }
1555}
1556
1557sub verify_remotes_sanity {
1558        return unless -d $ENV{GIT_DIR};
1559        my %seen;
1560        foreach (command(qw/config -l/)) {
1561                if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
1562                        if ($seen{$1}) {
1563                                die "Remote ref refs/remote/$1 is tracked by",
1564                                    "\n  \"$_\"\nand\n  \"$seen{$1}\"\n",
1565                                    "Please resolve this ambiguity in ",
1566                                    "your git configuration file before ",
1567                                    "continuing\n";
1568                        }
1569                        $seen{$1} = $_;
1570                }
1571        }
1572}
1573
1574sub find_existing_remote {
1575        my ($url, $remotes) = @_;
1576        return undef if $no_reuse_existing;
1577        my $existing;
1578        foreach my $repo_id (keys %$remotes) {
1579                my $u = $remotes->{$repo_id}->{url} or next;
1580                next if $u ne $url;
1581                $existing = $repo_id;
1582                last;
1583        }
1584        $existing;
1585}
1586
1587sub init_remote_config {
1588        my ($self, $url, $no_write) = @_;
1589        $url =~ s!/+$!!; # strip trailing slash
1590        my $r = read_all_remotes();
1591        my $existing = find_existing_remote($url, $r);
1592        if ($existing) {
1593                unless ($no_write) {
1594                        print STDERR "Using existing ",
1595                                     "[svn-remote \"$existing\"]\n";
1596                }
1597                $self->{repo_id} = $existing;
1598        } elsif ($_minimize_url) {
1599                my $min_url = Git::SVN::Ra->new($url)->minimize_url;
1600                $existing = find_existing_remote($min_url, $r);
1601                if ($existing) {
1602                        unless ($no_write) {
1603                                print STDERR "Using existing ",
1604                                             "[svn-remote \"$existing\"]\n";
1605                        }
1606                        $self->{repo_id} = $existing;
1607                }
1608                if ($min_url ne $url) {
1609                        unless ($no_write) {
1610                                print STDERR "Using higher level of URL: ",
1611                                             "$url => $min_url\n";
1612                        }
1613                        my $old_path = $self->{path};
1614                        $self->{path} = $url;
1615                        $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
1616                        if (length $old_path) {
1617                                $self->{path} .= "/$old_path";
1618                        }
1619                        $url = $min_url;
1620                }
1621        }
1622        my $orig_url;
1623        if (!$existing) {
1624                # verify that we aren't overwriting anything:
1625                $orig_url = eval {
1626                        command_oneline('config', '--get',
1627                                        "svn-remote.$self->{repo_id}.url")
1628                };
1629                if ($orig_url && ($orig_url ne $url)) {
1630                        die "svn-remote.$self->{repo_id}.url already set: ",
1631                            "$orig_url\nwanted to set to: $url\n";
1632                }
1633        }
1634        my ($xrepo_id, $xpath) = find_ref($self->refname);
1635        if (defined $xpath) {
1636                die "svn-remote.$xrepo_id.fetch already set to track ",
1637                    "$xpath:refs/remotes/", $self->refname, "\n";
1638        }
1639        unless ($no_write) {
1640                command_noisy('config',
1641                              "svn-remote.$self->{repo_id}.url", $url);
1642                $self->{path} =~ s{^/}{};
1643                command_noisy('config', '--add',
1644                              "svn-remote.$self->{repo_id}.fetch",
1645                              "$self->{path}:".$self->refname);
1646        }
1647        $self->{url} = $url;
1648}
1649
1650sub find_by_url { # repos_root and, path are optional
1651        my ($class, $full_url, $repos_root, $path) = @_;
1652
1653        return undef unless defined $full_url;
1654        remove_username($full_url);
1655        remove_username($repos_root) if defined $repos_root;
1656        my $remotes = read_all_remotes();
1657        if (defined $full_url && defined $repos_root && !defined $path) {
1658                $path = $full_url;
1659                $path =~ s#^\Q$repos_root\E(?:/|$)##;
1660        }
1661        foreach my $repo_id (keys %$remotes) {
1662                my $u = $remotes->{$repo_id}->{url} or next;
1663                remove_username($u);
1664                next if defined $repos_root && $repos_root ne $u;
1665
1666                my $fetch = $remotes->{$repo_id}->{fetch} || {};
1667                foreach (qw/branches tags/) {
1668                        resolve_local_globs($u, $fetch,
1669                                            $remotes->{$repo_id}->{$_});
1670                }
1671                my $p = $path;
1672                my $rwr = rewrite_root({repo_id => $repo_id});
1673                my $svm = $remotes->{$repo_id}->{svm}
1674                        if defined $remotes->{$repo_id}->{svm};
1675                unless (defined $p) {
1676                        $p = $full_url;
1677                        my $z = $u;
1678                        my $prefix = '';
1679                        if ($rwr) {
1680                                $z = $rwr;
1681                        } elsif (defined $svm) {
1682                                $z = $svm->{source};
1683                                $prefix = $svm->{replace};
1684                                $prefix =~ s#^\Q$u\E(?:/|$)##;
1685                                $prefix =~ s#/$##;
1686                        }
1687                        $p =~ s#^\Q$z\E(?:/|$)#$prefix# or next;
1688                }
1689                foreach my $f (keys %$fetch) {
1690                        next if $f ne $p;
1691                        return Git::SVN->new($fetch->{$f}, $repo_id, $f);
1692                }
1693        }
1694        undef;
1695}
1696
1697sub init {
1698        my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
1699        my $self = _new($class, $repo_id, $ref_id, $path);
1700        if (defined $url) {
1701                $self->init_remote_config($url, $no_write);
1702        }
1703        $self;
1704}
1705
1706sub find_ref {
1707        my ($ref_id) = @_;
1708        foreach (command(qw/config -l/)) {
1709                next unless m!^svn-remote\.(.+)\.fetch=
1710                              \s*(.*)\s*:\s*refs/remotes/(.+)\s*$!x;
1711                my ($repo_id, $path, $ref) = ($1, $2, $3);
1712                if ($ref eq $ref_id) {
1713                        $path = '' if ($path =~ m#^\./?#);
1714                        return ($repo_id, $path);
1715                }
1716        }
1717        (undef, undef, undef);
1718}
1719
1720sub new {
1721        my ($class, $ref_id, $repo_id, $path) = @_;
1722        if (defined $ref_id && !defined $repo_id && !defined $path) {
1723                ($repo_id, $path) = find_ref($ref_id);
1724                if (!defined $repo_id) {
1725                        die "Could not find a \"svn-remote.*.fetch\" key ",
1726                            "in the repository configuration matching: ",
1727                            "refs/remotes/$ref_id\n";
1728                }
1729        }
1730        my $self = _new($class, $repo_id, $ref_id, $path);
1731        if (!defined $self->{path} || !length $self->{path}) {
1732                my $fetch = command_oneline('config', '--get',
1733                                            "svn-remote.$repo_id.fetch",
1734                                            ":refs/remotes/$ref_id\$") or
1735                     die "Failed to read \"svn-remote.$repo_id.fetch\" ",
1736                         "\":refs/remotes/$ref_id\$\" in config\n";
1737                ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
1738        }
1739        $self->{url} = command_oneline('config', '--get',
1740                                       "svn-remote.$repo_id.url") or
1741                  die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
1742        $self->rebuild;
1743        $self;
1744}
1745
1746sub refname {
1747        my ($refname) = "refs/remotes/$_[0]->{ref_id}" ;
1748
1749        # It cannot end with a slash /, we'll throw up on this because
1750        # SVN can't have directories with a slash in their name, either:
1751        if ($refname =~ m{/$}) {
1752                die "ref: '$refname' ends with a trailing slash, this is ",
1753                    "not permitted by git nor Subversion\n";
1754        }
1755
1756        # It cannot have ASCII control character space, tilde ~, caret ^,
1757        # colon :, question-mark ?, asterisk *, space, or open bracket [
1758        # anywhere.
1759        #
1760        # Additionally, % must be escaped because it is used for escaping
1761        # and we want our escaped refname to be reversible
1762        $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
1763
1764        # no slash-separated component can begin with a dot .
1765        # /.* becomes /%2E*
1766        $refname =~ s{/\.}{/%2E}g;
1767
1768        # It cannot have two consecutive dots .. anywhere
1769        # .. becomes %2E%2E
1770        $refname =~ s{\.\.}{%2E%2E}g;
1771
1772        return $refname;
1773}
1774
1775sub desanitize_refname {
1776        my ($refname) = @_;
1777        $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
1778        return $refname;
1779}
1780
1781sub svm_uuid {
1782        my ($self) = @_;
1783        return $self->{svm}->{uuid} if $self->svm;
1784        $self->ra;
1785        unless ($self->{svm}) {
1786                die "SVM UUID not cached, and reading remotely failed\n";
1787        }
1788        $self->{svm}->{uuid};
1789}
1790
1791sub svm {
1792        my ($self) = @_;
1793        return $self->{svm} if $self->{svm};
1794        my $svm;
1795        # see if we have it in our config, first:
1796        eval {
1797                my $section = "svn-remote.$self->{repo_id}";
1798                $svm = {
1799                  source => tmp_config('--get', "$section.svm-source"),
1800                  uuid => tmp_config('--get', "$section.svm-uuid"),
1801                  replace => tmp_config('--get', "$section.svm-replace"),
1802                }
1803        };
1804        if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
1805                $self->{svm} = $svm;
1806        }
1807        $self->{svm};
1808}
1809
1810sub _set_svm_vars {
1811        my ($self, $ra) = @_;
1812        return $ra if $self->svm;
1813
1814        my @err = ( "useSvmProps set, but failed to read SVM properties\n",
1815                    "(svm:source, svm:uuid) ",
1816                    "from the following URLs:\n" );
1817        sub read_svm_props {
1818                my ($self, $ra, $path, $r) = @_;
1819                my $props = ($ra->get_dir($path, $r))[2];
1820                my $src = $props->{'svm:source'};
1821                my $uuid = $props->{'svm:uuid'};
1822                return undef if (!$src || !$uuid);
1823
1824                chomp($src, $uuid);
1825
1826                $uuid =~ m{^[0-9a-f\-]{30,}$}
1827                    or die "doesn't look right - svm:uuid is '$uuid'\n";
1828
1829                # the '!' is used to mark the repos_root!/relative/path
1830                $src =~ s{/?!/?}{/};
1831                $src =~ s{/+$}{}; # no trailing slashes please
1832                # username is of no interest
1833                $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
1834
1835                my $replace = $ra->{url};
1836                $replace .= "/$path" if length $path;
1837
1838                my $section = "svn-remote.$self->{repo_id}";
1839                tmp_config("$section.svm-source", $src);
1840                tmp_config("$section.svm-replace", $replace);
1841                tmp_config("$section.svm-uuid", $uuid);
1842                $self->{svm} = {
1843                        source => $src,
1844                        uuid => $uuid,
1845                        replace => $replace
1846                };
1847        }
1848
1849        my $r = $ra->get_latest_revnum;
1850        my $path = $self->{path};
1851        my %tried;
1852        while (length $path) {
1853                unless ($tried{"$self->{url}/$path"}) {
1854                        return $ra if $self->read_svm_props($ra, $path, $r);
1855                        $tried{"$self->{url}/$path"} = 1;
1856                }
1857                $path =~ s#/?[^/]+$##;
1858        }
1859        die "Path: '$path' should be ''\n" if $path ne '';
1860        return $ra if $self->read_svm_props($ra, $path, $r);
1861        $tried{"$self->{url}/$path"} = 1;
1862
1863        if ($ra->{repos_root} eq $self->{url}) {
1864                die @err, (map { "  $_\n" } keys %tried), "\n";
1865        }
1866
1867        # nope, make sure we're connected to the repository root:
1868        my $ok;
1869        my @tried_b;
1870        $path = $ra->{svn_path};
1871        $ra = Git::SVN::Ra->new($ra->{repos_root});
1872        while (length $path) {
1873                unless ($tried{"$ra->{url}/$path"}) {
1874                        $ok = $self->read_svm_props($ra, $path, $r);
1875                        last if $ok;
1876                        $tried{"$ra->{url}/$path"} = 1;
1877                }
1878                $path =~ s#/?[^/]+$##;
1879        }
1880        die "Path: '$path' should be ''\n" if $path ne '';
1881        $ok ||= $self->read_svm_props($ra, $path, $r);
1882        $tried{"$ra->{url}/$path"} = 1;
1883        if (!$ok) {
1884                die @err, (map { "  $_\n" } keys %tried), "\n";
1885        }
1886        Git::SVN::Ra->new($self->{url});
1887}
1888
1889sub svnsync {
1890        my ($self) = @_;
1891        return $self->{svnsync} if $self->{svnsync};
1892
1893        if ($self->no_metadata) {
1894                die "Can't have both 'noMetadata' and ",
1895                    "'useSvnsyncProps' options set!\n";
1896        }
1897        if ($self->rewrite_root) {
1898                die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
1899                    "options set!\n";
1900        }
1901
1902        my $svnsync;
1903        # see if we have it in our config, first:
1904        eval {
1905                my $section = "svn-remote.$self->{repo_id}";
1906
1907                my $url = tmp_config('--get', "$section.svnsync-url");
1908                ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
1909                   die "doesn't look right - svn:sync-from-url is '$url'\n";
1910
1911                my $uuid = tmp_config('--get', "$section.svnsync-uuid");
1912                ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}) or
1913                   die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
1914
1915                $svnsync = { url => $url, uuid => $uuid }
1916        };
1917        if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
1918                return $self->{svnsync} = $svnsync;
1919        }
1920
1921        my $err = "useSvnsyncProps set, but failed to read " .
1922                  "svnsync property: svn:sync-from-";
1923        my $rp = $self->ra->rev_proplist(0);
1924
1925        my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
1926        ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
1927                   die "doesn't look right - svn:sync-from-url is '$url'\n";
1928
1929        my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
1930        ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}) or
1931                   die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
1932
1933        my $section = "svn-remote.$self->{repo_id}";
1934        tmp_config('--add', "$section.svnsync-uuid", $uuid);
1935        tmp_config('--add', "$section.svnsync-url", $url);
1936        return $self->{svnsync} = { url => $url, uuid => $uuid };
1937}
1938
1939# this allows us to memoize our SVN::Ra UUID locally and avoid a
1940# remote lookup (useful for 'git svn log').
1941sub ra_uuid {
1942        my ($self) = @_;
1943        unless ($self->{ra_uuid}) {
1944                my $key = "svn-remote.$self->{repo_id}.uuid";
1945                my $uuid = eval { tmp_config('--get', $key) };
1946                if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/) {
1947                        $self->{ra_uuid} = $uuid;
1948                } else {
1949                        die "ra_uuid called without URL\n" unless $self->{url};
1950                        $self->{ra_uuid} = $self->ra->get_uuid;
1951                        tmp_config('--add', $key, $self->{ra_uuid});
1952                }
1953        }
1954        $self->{ra_uuid};
1955}
1956
1957sub _set_repos_root {
1958        my ($self, $repos_root) = @_;
1959        my $k = "svn-remote.$self->{repo_id}.reposRoot";
1960        $repos_root ||= $self->ra->{repos_root};
1961        tmp_config($k, $repos_root);
1962        $repos_root;
1963}
1964
1965sub repos_root {
1966        my ($self) = @_;
1967        my $k = "svn-remote.$self->{repo_id}.reposRoot";
1968        eval { tmp_config('--get', $k) } || $self->_set_repos_root;
1969}
1970
1971sub ra {
1972        my ($self) = shift;
1973        my $ra = Git::SVN::Ra->new($self->{url});
1974        $self->_set_repos_root($ra->{repos_root});
1975        if ($self->use_svm_props && !$self->{svm}) {
1976                if ($self->no_metadata) {
1977                        die "Can't have both 'noMetadata' and ",
1978                            "'useSvmProps' options set!\n";
1979                } elsif ($self->use_svnsync_props) {
1980                        die "Can't have both 'useSvnsyncProps' and ",
1981                            "'useSvmProps' options set!\n";
1982                }
1983                $ra = $self->_set_svm_vars($ra);
1984                $self->{-want_revprops} = 1;
1985        }
1986        $ra;
1987}
1988
1989sub rel_path {
1990        my ($self) = @_;
1991        my $repos_root = $self->ra->{repos_root};
1992        return $self->{path} if ($self->{url} eq $repos_root);
1993        my $url = $self->{url} .
1994                  (length $self->{path} ? "/$self->{path}" : $self->{path});
1995        $url =~ s!^\Q$repos_root\E(?:/+|$)!!g;
1996        $url;
1997}
1998
1999# prop_walk(PATH, REV, SUB)
2000# -------------------------
2001# Recursively traverse PATH at revision REV and invoke SUB for each
2002# directory that contains a SVN property.  SUB will be invoked as
2003# follows:  &SUB(gs, path, props);  where `gs' is this instance of
2004# Git::SVN, `path' the path to the directory where the properties
2005# `props' were found.  The `path' will be relative to point of checkout,
2006# that is, if url://repo/trunk is the current Git branch, and that
2007# directory contains a sub-directory `d', SUB will be invoked with `/d/'
2008# as `path' (note the trailing `/').
2009sub prop_walk {
2010        my ($self, $path, $rev, $sub) = @_;
2011
2012        $path =~ s#^/##;
2013        my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev);
2014        $path =~ s#^/*#/#g;
2015        my $p = $path;
2016        # Strip the irrelevant part of the path.
2017        $p =~ s#^/+\Q$self->{path}\E(/|$)#/#;
2018        # Ensure the path is terminated by a `/'.
2019        $p =~ s#/*$#/#;
2020
2021        # The properties contain all the internal SVN stuff nobody
2022        # (usually) cares about.
2023        my $interesting_props = 0;
2024        foreach (keys %{$props}) {
2025                # If it doesn't start with `svn:', it must be a
2026                # user-defined property.
2027                ++$interesting_props and next if $_ !~ /^svn:/;
2028                # FIXME: Fragile, if SVN adds new public properties,
2029                # this needs to be updated.
2030                ++$interesting_props if /^svn:(?:ignore|keywords|executable
2031                                                 |eol-style|mime-type
2032                                                 |externals|needs-lock)$/x;
2033        }
2034        &$sub($self, $p, $props) if $interesting_props;
2035
2036        foreach (sort keys %$dirent) {
2037                next if $dirent->{$_}->{kind} != $SVN::Node::dir;
2038                $self->prop_walk($self->{path} . $p . $_, $rev, $sub);
2039        }
2040}
2041
2042sub last_rev { ($_[0]->last_rev_commit)[0] }
2043sub last_commit { ($_[0]->last_rev_commit)[1] }
2044
2045# returns the newest SVN revision number and newest commit SHA1
2046sub last_rev_commit {
2047        my ($self) = @_;
2048        if (defined $self->{last_rev} && defined $self->{last_commit}) {
2049                return ($self->{last_rev}, $self->{last_commit});
2050        }
2051        my $c = ::verify_ref($self->refname.'^0');
2052        if ($c && !$self->use_svm_props && !$self->no_metadata) {
2053                my $rev = (::cmt_metadata($c))[1];
2054                if (defined $rev) {
2055                        ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
2056                        return ($rev, $c);
2057                }
2058        }
2059        my $map_path = $self->map_path;
2060        unless (-e $map_path) {
2061                ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
2062                return (undef, undef);
2063        }
2064        my ($rev, $commit) = $self->rev_map_max(1);
2065        ($self->{last_rev}, $self->{last_commit}) = ($rev, $commit);
2066        return ($rev, $commit);
2067}
2068
2069sub get_fetch_range {
2070        my ($self, $min, $max) = @_;
2071        $max ||= $self->ra->get_latest_revnum;
2072        $min ||= $self->rev_map_max;
2073        (++$min, $max);
2074}
2075
2076sub tmp_config {
2077        my (@args) = @_;
2078        my $old_def_config = "$ENV{GIT_DIR}/svn/config";
2079        my $config = "$ENV{GIT_DIR}/svn/.metadata";
2080        if (! -f $config && -f $old_def_config) {
2081                rename $old_def_config, $config or
2082                       die "Failed rename $old_def_config => $config: $!\n";
2083        }
2084        my $old_config = $ENV{GIT_CONFIG};
2085        $ENV{GIT_CONFIG} = $config;
2086        $@ = undef;
2087        my @ret = eval {
2088                unless (-f $config) {
2089                        mkfile($config);
2090                        open my $fh, '>', $config or
2091                            die "Can't open $config: $!\n";
2092                        print $fh "; This file is used internally by ",
2093                                  "git-svn\n" or die
2094                                  "Couldn't write to $config: $!\n";
2095                        print $fh "; You should not have to edit it\n" or
2096                              die "Couldn't write to $config: $!\n";
2097                        close $fh or die "Couldn't close $config: $!\n";
2098                }
2099                command('config', @args);
2100        };
2101        my $err = $@;
2102        if (defined $old_config) {
2103                $ENV{GIT_CONFIG} = $old_config;
2104        } else {
2105                delete $ENV{GIT_CONFIG};
2106        }
2107        die $err if $err;
2108        wantarray ? @ret : $ret[0];
2109}
2110
2111sub tmp_index_do {
2112        my ($self, $sub) = @_;
2113        my $old_index = $ENV{GIT_INDEX_FILE};
2114        $ENV{GIT_INDEX_FILE} = $self->{index};
2115        $@ = undef;
2116        my @ret = eval {
2117                my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
2118                mkpath([$dir]) unless -d $dir;
2119                &$sub;
2120        };
2121        my $err = $@;
2122        if (defined $old_index) {
2123                $ENV{GIT_INDEX_FILE} = $old_index;
2124        } else {
2125                delete $ENV{GIT_INDEX_FILE};
2126        }
2127        die $err if $err;
2128        wantarray ? @ret : $ret[0];
2129}
2130
2131sub assert_index_clean {
2132        my ($self, $treeish) = @_;
2133
2134        $self->tmp_index_do(sub {
2135                command_noisy('read-tree', $treeish) unless -e $self->{index};
2136                my $x = command_oneline('write-tree');
2137                my ($y) = (command(qw/cat-file commit/, $treeish) =~
2138                           /^tree ($::sha1)/mo);
2139                return if $y eq $x;
2140
2141                warn "Index mismatch: $y != $x\nrereading $treeish\n";
2142                unlink $self->{index} or die "unlink $self->{index}: $!\n";
2143                command_noisy('read-tree', $treeish);
2144                $x = command_oneline('write-tree');
2145                if ($y ne $x) {
2146                        ::fatal "trees ($treeish) $y != $x\n",
2147                                "Something is seriously wrong...";
2148                }
2149        });
2150}
2151
2152sub get_commit_parents {
2153        my ($self, $log_entry) = @_;
2154        my (%seen, @ret, @tmp);
2155        # legacy support for 'set-tree'; this is only used by set_tree_cb:
2156        if (my $ip = $self->{inject_parents}) {
2157                if (my $commit = delete $ip->{$log_entry->{revision}}) {
2158                        push @tmp, $commit;
2159                }
2160        }
2161        if (my $cur = ::verify_ref($self->refname.'^0')) {
2162                push @tmp, $cur;
2163        }
2164        if (my $ipd = $self->{inject_parents_dcommit}) {
2165                if (my $commit = delete $ipd->{$log_entry->{revision}}) {
2166                        push @tmp, @$commit;
2167                }
2168        }
2169        push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
2170        while (my $p = shift @tmp) {
2171                next if $seen{$p};
2172                $seen{$p} = 1;
2173                push @ret, $p;
2174                # MAXPARENT is defined to 16 in commit-tree.c:
2175                last if @ret >= 16;
2176        }
2177        if (@tmp) {
2178                die "r$log_entry->{revision}: No room for parents:\n\t",
2179                    join("\n\t", @tmp), "\n";
2180        }
2181        @ret;
2182}
2183
2184sub rewrite_root {
2185        my ($self) = @_;
2186        return $self->{-rewrite_root} if exists $self->{-rewrite_root};
2187        my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
2188        my $rwr = eval { command_oneline(qw/config --get/, $k) };
2189        if ($rwr) {
2190                $rwr =~ s#/+$##;
2191                if ($rwr !~ m#^[a-z\+]+://#) {
2192                        die "$rwr is not a valid URL (key: $k)\n";
2193                }
2194        }
2195        $self->{-rewrite_root} = $rwr;
2196}
2197
2198sub metadata_url {
2199        my ($self) = @_;
2200        ($self->rewrite_root || $self->{url}) .
2201           (length $self->{path} ? '/' . $self->{path} : '');
2202}
2203
2204sub full_url {
2205        my ($self) = @_;
2206        $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
2207}
2208
2209
2210sub set_commit_header_env {
2211        my ($log_entry) = @_;
2212        my %env;
2213        foreach my $ned (qw/NAME EMAIL DATE/) {
2214                foreach my $ac (qw/AUTHOR COMMITTER/) {
2215                        $env{"GIT_${ac}_${ned}"} = $ENV{"GIT_${ac}_${ned}"};
2216                }
2217        }
2218
2219        $ENV{GIT_AUTHOR_NAME} = $log_entry->{name};
2220        $ENV{GIT_AUTHOR_EMAIL} = $log_entry->{email};
2221        $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
2222
2223        $ENV{GIT_COMMITTER_NAME} = (defined $log_entry->{commit_name})
2224                                                ? $log_entry->{commit_name}
2225                                                : $log_entry->{name};
2226        $ENV{GIT_COMMITTER_EMAIL} = (defined $log_entry->{commit_email})
2227                                                ? $log_entry->{commit_email}
2228                                                : $log_entry->{email};
2229        \%env;
2230}
2231
2232sub restore_commit_header_env {
2233        my ($env) = @_;
2234        foreach my $ned (qw/NAME EMAIL DATE/) {
2235                foreach my $ac (qw/AUTHOR COMMITTER/) {
2236                        my $k = "GIT_${ac}_${ned}";
2237                        if (defined $env->{$k}) {
2238                                $ENV{$k} = $env->{$k};
2239                        } else {
2240                                delete $ENV{$k};
2241                        }
2242                }
2243        }
2244}
2245
2246sub gc {
2247        command_noisy('gc', '--auto');
2248};
2249
2250sub do_git_commit {
2251        my ($self, $log_entry) = @_;
2252        my $lr = $self->last_rev;
2253        if (defined $lr && $lr >= $log_entry->{revision}) {
2254                die "Last fetched revision of ", $self->refname,
2255                    " was r$lr, but we are about to fetch: ",
2256                    "r$log_entry->{revision}!\n";
2257        }
2258        if (my $c = $self->rev_map_get($log_entry->{revision})) {
2259                croak "$log_entry->{revision} = $c already exists! ",
2260                      "Why are we refetching it?\n";
2261        }
2262        my $old_env = set_commit_header_env($log_entry);
2263        my $tree = $log_entry->{tree};
2264        if (!defined $tree) {
2265                $tree = $self->tmp_index_do(sub {
2266                                            command_oneline('write-tree') });
2267        }
2268        die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
2269
2270        my @exec = ('git', 'commit-tree', $tree);
2271        foreach ($self->get_commit_parents($log_entry)) {
2272                push @exec, '-p', $_;
2273        }
2274        defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
2275                                                                   or croak $!;
2276        print $msg_fh $log_entry->{log} or croak $!;
2277        restore_commit_header_env($old_env);
2278        unless ($self->no_metadata) {
2279                print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
2280                              or croak $!;
2281        }
2282        $msg_fh->flush == 0 or croak $!;
2283        close $msg_fh or croak $!;
2284        chomp(my $commit = do { local $/; <$out_fh> });
2285        close $out_fh or croak $!;
2286        waitpid $pid, 0;
2287        croak $? if $?;
2288        if ($commit !~ /^$::sha1$/o) {
2289                die "Failed to commit, invalid sha1: $commit\n";
2290        }
2291
2292        $self->rev_map_set($log_entry->{revision}, $commit, 1);
2293
2294        $self->{last_rev} = $log_entry->{revision};
2295        $self->{last_commit} = $commit;
2296        print "r$log_entry->{revision}";
2297        if (defined $log_entry->{svm_revision}) {
2298                 print " (\@$log_entry->{svm_revision})";
2299                 $self->rev_map_set($log_entry->{svm_revision}, $commit,
2300                                   0, $self->svm_uuid);
2301        }
2302        print " = $commit ($self->{ref_id})\n";
2303        if (--$_gc_nr == 0) {
2304                $_gc_nr = $_gc_period;
2305                gc();
2306        }
2307        return $commit;
2308}
2309
2310sub match_paths {
2311        my ($self, $paths, $r) = @_;
2312        return 1 if $self->{path} eq '';
2313        if (my $path = $paths->{"/$self->{path}"}) {
2314                return ($path->{action} eq 'D') ? 0 : 1;
2315        }
2316        $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
2317        if (grep /$self->{path_regex}/, keys %$paths) {
2318                return 1;
2319        }
2320        my $c = '';
2321        foreach (split m#/#, $self->{path}) {
2322                $c .= "/$_";
2323                next unless ($paths->{$c} &&
2324                             ($paths->{$c}->{action} =~ /^[AR]$/));
2325                if ($self->ra->check_path($self->{path}, $r) ==
2326                    $SVN::Node::dir) {
2327                        return 1;
2328                }
2329        }
2330        return 0;
2331}
2332
2333sub find_parent_branch {
2334        my ($self, $paths, $rev) = @_;
2335        return undef unless $self->follow_parent;
2336        unless (defined $paths) {
2337                my $err_handler = $SVN::Error::handler;
2338                $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
2339                $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1, sub {
2340                                   $paths =
2341                                      Git::SVN::Ra::dup_changed_paths($_[0]) });
2342                $SVN::Error::handler = $err_handler;
2343        }
2344        return undef unless defined $paths;
2345
2346        # look for a parent from another branch:
2347        my @b_path_components = split m#/#, $self->rel_path;
2348        my @a_path_components;
2349        my $i;
2350        while (@b_path_components) {
2351                $i = $paths->{'/'.join('/', @b_path_components)};
2352                last if $i && defined $i->{copyfrom_path};
2353                unshift(@a_path_components, pop(@b_path_components));
2354        }
2355        return undef unless defined $i && defined $i->{copyfrom_path};
2356        my $branch_from = $i->{copyfrom_path};
2357        if (@a_path_components) {
2358                print STDERR "branch_from: $branch_from => ";
2359                $branch_from .= '/'.join('/', @a_path_components);
2360                print STDERR $branch_from, "\n";
2361        }
2362        my $r = $i->{copyfrom_rev};
2363        my $repos_root = $self->ra->{repos_root};
2364        my $url = $self->ra->{url};
2365        my $new_url = $repos_root . $branch_from;
2366        print STDERR  "Found possible branch point: ",
2367                      "$new_url => ", $self->full_url, ", $r\n";
2368        $branch_from =~ s#^/##;
2369        my $gs = Git::SVN->find_by_url($new_url, $repos_root, $branch_from);
2370        unless ($gs) {
2371                my $ref_id = $self->{ref_id};
2372                $ref_id =~ s/\@\d+$//;
2373                $ref_id .= "\@$r";
2374                # just grow a tail if we're not unique enough :x
2375                $ref_id .= '-' while find_ref($ref_id);
2376                print STDERR "Initializing parent: $ref_id\n";
2377                my ($u, $p, $repo_id) = ($new_url, '', $ref_id);
2378                if ($u =~ s#^\Q$url\E(/|$)##) {
2379                        $p = $u;
2380                        $u = $url;
2381                        $repo_id = $self->{repo_id};
2382                }
2383                $gs = Git::SVN->init($u, $p, $repo_id, $ref_id, 1);
2384        }
2385        my ($r0, $parent) = $gs->find_rev_before($r, 1);
2386        if (!defined $r0 || !defined $parent) {
2387                my ($base, $head) = parse_revision_argument(0, $r);
2388                if ($base <= $r) {
2389                        $gs->fetch($base, $r);
2390                }
2391                ($r0, $parent) = $gs->last_rev_commit;
2392        }
2393        if (defined $r0 && defined $parent) {
2394                print STDERR "Found branch parent: ($self->{ref_id}) $parent\n";
2395                my $ed;
2396                if ($self->ra->can_do_switch) {
2397                        $self->assert_index_clean($parent);
2398                        print STDERR "Following parent with do_switch\n";
2399                        # do_switch works with svn/trunk >= r22312, but that
2400                        # is not included with SVN 1.4.3 (the latest version
2401                        # at the moment), so we can't rely on it
2402                        $self->{last_commit} = $parent;
2403                        $ed = SVN::Git::Fetcher->new($self);
2404                        $gs->ra->gs_do_switch($r0, $rev, $gs,
2405                                              $self->full_url, $ed)
2406                          or die "SVN connection failed somewhere...\n";
2407                } elsif ($self->ra->trees_match($new_url, $r0,
2408                                                $self->full_url, $rev)) {
2409                        print STDERR "Trees match:\n",
2410                                     "  $new_url\@$r0\n",
2411                                     "  ${\$self->full_url}\@$rev\n",
2412                                     "Following parent with no changes\n";
2413                        $self->tmp_index_do(sub {
2414                            command_noisy('read-tree', $parent);
2415                        });
2416                        $self->{last_commit} = $parent;
2417                } else {
2418                        print STDERR "Following parent with do_update\n";
2419                        $ed = SVN::Git::Fetcher->new($self);
2420                        $self->ra->gs_do_update($rev, $rev, $self, $ed)
2421                          or die "SVN connection failed somewhere...\n";
2422                }
2423                print STDERR "Successfully followed parent\n";
2424                return $self->make_log_entry($rev, [$parent], $ed);
2425        }
2426        return undef;
2427}
2428
2429sub do_fetch {
2430        my ($self, $paths, $rev) = @_;
2431        my $ed;
2432        my ($last_rev, @parents);
2433        if (my $lc = $self->last_commit) {
2434                # we can have a branch that was deleted, then re-added
2435                # under the same name but copied from another path, in
2436                # which case we'll have multiple parents (we don't
2437                # want to break the original ref, nor lose copypath info):
2438                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2439                        push @{$log_entry->{parents}}, $lc;
2440                        return $log_entry;
2441                }
2442                $ed = SVN::Git::Fetcher->new($self);
2443                $last_rev = $self->{last_rev};
2444                $ed->{c} = $lc;
2445                @parents = ($lc);
2446        } else {
2447                $last_rev = $rev;
2448                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2449                        return $log_entry;
2450                }
2451                $ed = SVN::Git::Fetcher->new($self);
2452        }
2453        unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
2454                die "SVN connection failed somewhere...\n";
2455        }
2456        $self->make_log_entry($rev, \@parents, $ed);
2457}
2458
2459sub get_untracked {
2460        my ($self, $ed) = @_;
2461        my @out;
2462        my $h = $ed->{empty};
2463        foreach (sort keys %$h) {
2464                my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
2465                push @out, "  $act: " . uri_encode($_);
2466                warn "W: $act: $_\n";
2467        }
2468        foreach my $t (qw/dir_prop file_prop/) {
2469                $h = $ed->{$t} or next;
2470                foreach my $path (sort keys %$h) {
2471                        my $ppath = $path eq '' ? '.' : $path;
2472                        foreach my $prop (sort keys %{$h->{$path}}) {
2473                                next if $SKIP_PROP{$prop};
2474                                my $v = $h->{$path}->{$prop};
2475                                my $t_ppath_prop = "$t: " .
2476                                                    uri_encode($ppath) . ' ' .
2477                                                    uri_encode($prop);
2478                                if (defined $v) {
2479                                        push @out, "  +$t_ppath_prop " .
2480                                                   uri_encode($v);
2481                                } else {
2482                                        push @out, "  -$t_ppath_prop";
2483                                }
2484                        }
2485                }
2486        }
2487        foreach my $t (qw/absent_file absent_directory/) {
2488                $h = $ed->{$t} or next;
2489                foreach my $parent (sort keys %$h) {
2490                        foreach my $path (sort @{$h->{$parent}}) {
2491                                push @out, "  $t: " .
2492                                           uri_encode("$parent/$path");
2493                                warn "W: $t: $parent/$path ",
2494                                     "Insufficient permissions?\n";
2495                        }
2496                }
2497        }
2498        \@out;
2499}
2500
2501sub parse_svn_date {
2502        my $date = shift || return '+0000 1970-01-01 00:00:00';
2503        my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2504                                            (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x) or
2505                                         croak "Unable to parse date: $date\n";
2506        "+0000 $Y-$m-$d $H:$M:$S";
2507}
2508
2509sub check_author {
2510        my ($author) = @_;
2511        if (!defined $author || length $author == 0) {
2512                $author = '(no author)';
2513        } elsif (defined $::_authors && ! defined $::users{$author}) {
2514                die "Author: $author not defined in $::_authors file\n";
2515        }
2516        $author;
2517}
2518
2519sub make_log_entry {
2520        my ($self, $rev, $parents, $ed) = @_;
2521        my $untracked = $self->get_untracked($ed);
2522
2523        open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
2524        print $un "r$rev\n" or croak $!;
2525        print $un $_, "\n" foreach @$untracked;
2526        my %log_entry = ( parents => $parents || [], revision => $rev,
2527                          log => '');
2528
2529        my $headrev;
2530        my $logged = delete $self->{logged_rev_props};
2531        if (!$logged || $self->{-want_revprops}) {
2532                my $rp = $self->ra->rev_proplist($rev);
2533                foreach (sort keys %$rp) {
2534                        my $v = $rp->{$_};
2535                        if (/^svn:(author|date|log)$/) {
2536                                $log_entry{$1} = $v;
2537                        } elsif ($_ eq 'svm:headrev') {
2538                                $headrev = $v;
2539                        } else {
2540                                print $un "  rev_prop: ", uri_encode($_), ' ',
2541                                          uri_encode($v), "\n";
2542                        }
2543                }
2544        } else {
2545                map { $log_entry{$_} = $logged->{$_} } keys %$logged;
2546        }
2547        close $un or croak $!;
2548
2549        $log_entry{date} = parse_svn_date($log_entry{date});
2550        $log_entry{log} .= "\n";
2551        my $author = $log_entry{author} = check_author($log_entry{author});
2552        my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
2553                                                       : ($author, undef);
2554
2555        my ($commit_name, $commit_email) = ($name, $email);
2556        if ($_use_log_author) {
2557                my $name_field;
2558                if ($log_entry{log} =~ /From:\s+(.*\S)\s*\n/i) {
2559                        $name_field = $1;
2560                } elsif ($log_entry{log} =~ /Signed-off-by:\s+(.*\S)\s*\n/i) {
2561                        $name_field = $1;
2562                }
2563                if (!defined $name_field) {
2564                        if (!defined $email) {
2565                                $email = $name;
2566                        }
2567                } elsif ($name_field =~ /(.*?)\s+<(.*)>/) {
2568                        ($name, $email) = ($1, $2);
2569                } elsif ($name_field =~ /(.*)@/) {
2570                        ($name, $email) = ($1, $name_field);
2571                } else {
2572                        ($name, $email) = ($name_field, $name_field);
2573                }
2574        }
2575        if (defined $headrev && $self->use_svm_props) {
2576                if ($self->rewrite_root) {
2577                        die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
2578                            "options set!\n";
2579                }
2580                my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$};
2581                # we don't want "SVM: initializing mirror for junk" ...
2582                return undef if $r == 0;
2583                my $svm = $self->svm;
2584                if ($uuid ne $svm->{uuid}) {
2585                        die "UUID mismatch on SVM path:\n",
2586                            "expected: $svm->{uuid}\n",
2587                            "     got: $uuid\n";
2588                }
2589                my $full_url = $self->full_url;
2590                $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
2591                             die "Failed to replace '$svm->{replace}' with ",
2592                                 "'$svm->{source}' in $full_url\n";
2593                # throw away username for storing in records
2594                remove_username($full_url);
2595                $log_entry{metadata} = "$full_url\@$r $uuid";
2596                $log_entry{svm_revision} = $r;
2597                $email ||= "$author\@$uuid";
2598                $commit_email ||= "$author\@$uuid";
2599        } elsif ($self->use_svnsync_props) {
2600                my $full_url = $self->svnsync->{url};
2601                $full_url .= "/$self->{path}" if length $self->{path};
2602                remove_username($full_url);
2603                my $uuid = $self->svnsync->{uuid};
2604                $log_entry{metadata} = "$full_url\@$rev $uuid";
2605                $email ||= "$author\@$uuid";
2606                $commit_email ||= "$author\@$uuid";
2607        } else {
2608                my $url = $self->metadata_url;
2609                remove_username($url);
2610                $log_entry{metadata} = "$url\@$rev " .
2611                                       $self->ra->get_uuid;
2612                $email ||= "$author\@" . $self->ra->get_uuid;
2613                $commit_email ||= "$author\@" . $self->ra->get_uuid;
2614        }
2615        $log_entry{name} = $name;
2616        $log_entry{email} = $email;
2617        $log_entry{commit_name} = $commit_name;
2618        $log_entry{commit_email} = $commit_email;
2619        \%log_entry;
2620}
2621
2622sub fetch {
2623        my ($self, $min_rev, $max_rev, @parents) = @_;
2624        my ($last_rev, $last_commit) = $self->last_rev_commit;
2625        my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
2626        $self->ra->gs_fetch_loop_common($base, $head, [$self]);
2627}
2628
2629sub set_tree_cb {
2630        my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
2631        $self->{inject_parents} = { $rev => $tree };
2632        $self->fetch(undef, undef);
2633}
2634
2635sub set_tree {
2636        my ($self, $tree) = (shift, shift);
2637        my $log_entry = ::get_commit_entry($tree);
2638        unless ($self->{last_rev}) {
2639                ::fatal("Must have an existing revision to commit");
2640        }
2641        my %ed_opts = ( r => $self->{last_rev},
2642                        log => $log_entry->{log},
2643                        ra => $self->ra,
2644                        tree_a => $self->{last_commit},
2645                        tree_b => $tree,
2646                        editor_cb => sub {
2647                               $self->set_tree_cb($log_entry, $tree, @_) },
2648                        svn_path => $self->{path} );
2649        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
2650                print "No changes\nr$self->{last_rev} = $tree\n";
2651        }
2652}
2653
2654sub rebuild_from_rev_db {
2655        my ($self, $path) = @_;
2656        my $r = -1;
2657        open my $fh, '<', $path or croak "open: $!";
2658        binmode $fh or croak "binmode: $!";
2659        while (<$fh>) {
2660                length($_) == 41 or croak "inconsistent size in ($_) != 41";
2661                chomp($_);
2662                ++$r;
2663                next if $_ eq ('0' x 40);
2664                $self->rev_map_set($r, $_);
2665                print "r$r = $_\n";
2666        }
2667        close $fh or croak "close: $!";
2668        unlink $path or croak "unlink: $!";
2669}
2670
2671sub rebuild {
2672        my ($self) = @_;
2673        my $map_path = $self->map_path;
2674        my $partial = (-e $map_path && ! -z $map_path);
2675        return unless ::verify_ref($self->refname.'^0');
2676        if (!$partial && ($self->use_svm_props || $self->no_metadata)) {
2677                my $rev_db = $self->rev_db_path;
2678                $self->rebuild_from_rev_db($rev_db);
2679                if ($self->use_svm_props) {
2680                        my $svm_rev_db = $self->rev_db_path($self->svm_uuid);
2681                        $self->rebuild_from_rev_db($svm_rev_db);
2682                }
2683                $self->unlink_rev_db_symlink;
2684                return;
2685        }
2686        print "Rebuilding $map_path ...\n" if (!$partial);
2687        my ($base_rev, $head) = ($partial ? $self->rev_map_max_norebuild(1) :
2688                (undef, undef));
2689        my ($log, $ctx) =
2690            command_output_pipe(qw/rev-list --pretty=raw --no-color --reverse/,
2691                                ($head ? "$head.." : "") . $self->refname,
2692                                '--');
2693        my $metadata_url = $self->metadata_url;
2694        remove_username($metadata_url);
2695        my $svn_uuid = $self->ra_uuid;
2696        my $c;
2697        while (<$log>) {
2698                if ( m{^commit ($::sha1)$} ) {
2699                        $c = $1;
2700                        next;
2701                }
2702                next unless s{^\s*(git-svn-id:)}{$1};
2703                my ($url, $rev, $uuid) = ::extract_metadata($_);
2704                remove_username($url);
2705
2706                # ignore merges (from set-tree)
2707                next if (!defined $rev || !$uuid);
2708
2709                # if we merged or otherwise started elsewhere, this is
2710                # how we break out of it
2711                if (($uuid ne $svn_uuid) ||
2712                    ($metadata_url && $url && ($url ne $metadata_url))) {
2713                        next;
2714                }
2715                if ($partial && $head) {
2716                        print "Partial-rebuilding $map_path ...\n";
2717                        print "Currently at $base_rev = $head\n";
2718                        $head = undef;
2719                }
2720
2721                $self->rev_map_set($rev, $c);
2722                print "r$rev = $c\n";
2723        }
2724        command_close_pipe($log, $ctx);
2725        print "Done rebuilding $map_path\n" if (!$partial || !$head);
2726        my $rev_db_path = $self->rev_db_path;
2727        if (-f $self->rev_db_path) {
2728                unlink $self->rev_db_path or croak "unlink: $!";
2729        }
2730        $self->unlink_rev_db_symlink;
2731}
2732
2733# rev_map:
2734# Tie::File seems to be prone to offset errors if revisions get sparse,
2735# it's not that fast, either.  Tie::File is also not in Perl 5.6.  So
2736# one of my favorite modules is out :<  Next up would be one of the DBM
2737# modules, but I'm not sure which is most portable...
2738#
2739# This is the replacement for the rev_db format, which was too big
2740# and inefficient for large repositories with a lot of sparse history
2741# (mainly tags)
2742#
2743# The format is this:
2744#   - 24 bytes for every record,
2745#     * 4 bytes for the integer representing an SVN revision number
2746#     * 20 bytes representing the sha1 of a git commit
2747#   - No empty padding records like the old format
2748#     (except the last record, which can be overwritten)
2749#   - new records are written append-only since SVN revision numbers
2750#     increase monotonically
2751#   - lookups on SVN revision number are done via a binary search
2752#   - Piping the file to xxd -c24 is a good way of dumping it for
2753#     viewing or editing (piped back through xxd -r), should the need
2754#     ever arise.
2755#   - The last record can be padding revision with an all-zero sha1
2756#     This is used to optimize fetch performance when using multiple
2757#     "fetch" directives in .git/config
2758#
2759# These files are disposable unless noMetadata or useSvmProps is set
2760
2761sub _rev_map_set {
2762        my ($fh, $rev, $commit) = @_;
2763
2764        binmode $fh or croak "binmode: $!";
2765        my $size = (stat($fh))[7];
2766        ($size % 24) == 0 or croak "inconsistent size: $size";
2767
2768        my $wr_offset = 0;
2769        if ($size > 0) {
2770                sysseek($fh, -24, SEEK_END) or croak "seek: $!";
2771                my $read = sysread($fh, my $buf, 24) or croak "read: $!";
2772                $read == 24 or croak "read only $read bytes (!= 24)";
2773                my ($last_rev, $last_commit) = unpack(rev_map_fmt, $buf);
2774                if ($last_commit eq ('0' x40)) {
2775                        if ($size >= 48) {
2776                                sysseek($fh, -48, SEEK_END) or croak "seek: $!";
2777                                $read = sysread($fh, $buf, 24) or
2778                                    croak "read: $!";
2779                                $read == 24 or
2780                                    croak "read only $read bytes (!= 24)";
2781                                ($last_rev, $last_commit) =
2782                                    unpack(rev_map_fmt, $buf);
2783                                if ($last_commit eq ('0' x40)) {
2784                                        croak "inconsistent .rev_map\n";
2785                                }
2786                        }
2787                        if ($last_rev >= $rev) {
2788                                croak "last_rev is higher!: $last_rev >= $rev";
2789                        }
2790                        $wr_offset = -24;
2791                }
2792        }
2793        sysseek($fh, $wr_offset, SEEK_END) or croak "seek: $!";
2794        syswrite($fh, pack(rev_map_fmt, $rev, $commit), 24) == 24 or
2795          croak "write: $!";
2796}
2797
2798sub mkfile {
2799        my ($path) = @_;
2800        unless (-e $path) {
2801                my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
2802                mkpath([$dir]) unless -d $dir;
2803                open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
2804                close $fh or die "Couldn't close (create) $path: $!\n";
2805        }
2806}
2807
2808sub rev_map_set {
2809        my ($self, $rev, $commit, $update_ref, $uuid) = @_;
2810        length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
2811        my $db = $self->map_path($uuid);
2812        my $db_lock = "$db.lock";
2813        my $sig;
2814        if ($update_ref) {
2815                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
2816                            $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
2817        }
2818        mkfile($db);
2819
2820        $LOCKFILES{$db_lock} = 1;
2821        my $sync;
2822        # both of these options make our .rev_db file very, very important
2823        # and we can't afford to lose it because rebuild() won't work
2824        if ($self->use_svm_props || $self->no_metadata) {
2825                $sync = 1;
2826                copy($db, $db_lock) or die "rev_map_set(@_): ",
2827                                           "Failed to copy: ",
2828                                           "$db => $db_lock ($!)\n";
2829        } else {
2830                rename $db, $db_lock or die "rev_map_set(@_): ",
2831                                            "Failed to rename: ",
2832                                            "$db => $db_lock ($!)\n";
2833        }
2834
2835        sysopen(my $fh, $db_lock, O_RDWR | O_CREAT)
2836             or croak "Couldn't open $db_lock: $!\n";
2837        _rev_map_set($fh, $rev, $commit);
2838        if ($sync) {
2839                $fh->flush or die "Couldn't flush $db_lock: $!\n";
2840                $fh->sync or die "Couldn't sync $db_lock: $!\n";
2841        }
2842        close $fh or croak $!;
2843        if ($update_ref) {
2844                $_head = $self;
2845                command_noisy('update-ref', '-m', "r$rev",
2846                              $self->refname, $commit);
2847        }
2848        rename $db_lock, $db or die "rev_map_set(@_): ", "Failed to rename: ",
2849                                    "$db_lock => $db ($!)\n";
2850        delete $LOCKFILES{$db_lock};
2851        if ($update_ref) {
2852                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
2853                            $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
2854                kill $sig, $$ if defined $sig;
2855        }
2856}
2857
2858# If want_commit, this will return an array of (rev, commit) where
2859# commit _must_ be a valid commit in the archive.
2860# Otherwise, it'll return the max revision (whether or not the
2861# commit is valid or just a 0x40 placeholder).
2862sub rev_map_max {
2863        my ($self, $want_commit) = @_;
2864        $self->rebuild;
2865        my ($r, $c) = $self->rev_map_max_norebuild($want_commit);
2866        $want_commit ? ($r, $c) : $r;
2867}
2868
2869sub rev_map_max_norebuild {
2870        my ($self, $want_commit) = @_;
2871        my $map_path = $self->map_path;
2872        stat $map_path or return $want_commit ? (0, undef) : 0;
2873        sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
2874        binmode $fh or croak "binmode: $!";
2875        my $size = (stat($fh))[7];
2876        ($size % 24) == 0 or croak "inconsistent size: $size";
2877
2878        if ($size == 0) {
2879                close $fh or croak "close: $!";
2880                return $want_commit ? (0, undef) : 0;
2881        }
2882
2883        sysseek($fh, -24, SEEK_END) or croak "seek: $!";
2884        sysread($fh, my $buf, 24) == 24 or croak "read: $!";
2885        my ($r, $c) = unpack(rev_map_fmt, $buf);
2886        if ($want_commit && $c eq ('0' x40)) {
2887                if ($size < 48) {
2888                        return $want_commit ? (0, undef) : 0;
2889                }
2890                sysseek($fh, -48, SEEK_END) or croak "seek: $!";
2891                sysread($fh, $buf, 24) == 24 or croak "read: $!";
2892                ($r, $c) = unpack(rev_map_fmt, $buf);
2893                if ($c eq ('0'x40)) {
2894                        croak "Penultimate record is all-zeroes in $map_path";
2895                }
2896        }
2897        close $fh or croak "close: $!";
2898        $want_commit ? ($r, $c) : $r;
2899}
2900
2901sub rev_map_get {
2902        my ($self, $rev, $uuid) = @_;
2903        my $map_path = $self->map_path($uuid);
2904        return undef unless -e $map_path;
2905
2906        sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
2907        binmode $fh or croak "binmode: $!";
2908        my $size = (stat($fh))[7];
2909        ($size % 24) == 0 or croak "inconsistent size: $size";
2910
2911        if ($size == 0) {
2912                close $fh or croak "close: $fh";
2913                return undef;
2914        }
2915
2916        my ($l, $u) = (0, $size - 24);
2917        my ($r, $c, $buf);
2918
2919        while ($l <= $u) {
2920                my $i = int(($l/24 + $u/24) / 2) * 24;
2921                sysseek($fh, $i, SEEK_SET) or croak "seek: $!";
2922                sysread($fh, my $buf, 24) == 24 or croak "read: $!";
2923                my ($r, $c) = unpack('NH40', $buf);
2924
2925                if ($r < $rev) {
2926                        $l = $i + 24;
2927                } elsif ($r > $rev) {
2928                        $u = $i - 24;
2929                } else { # $r == $rev
2930                        close($fh) or croak "close: $!";
2931                        return $c eq ('0' x 40) ? undef : $c;
2932                }
2933        }
2934        close($fh) or croak "close: $!";
2935        undef;
2936}
2937
2938# Finds the first svn revision that exists on (if $eq_ok is true) or
2939# before $rev for the current branch.  It will not search any lower
2940# than $min_rev.  Returns the git commit hash and svn revision number
2941# if found, else (undef, undef).
2942sub find_rev_before {
2943        my ($self, $rev, $eq_ok, $min_rev) = @_;
2944        --$rev unless $eq_ok;
2945        $min_rev ||= 1;
2946        while ($rev >= $min_rev) {
2947                if (my $c = $self->rev_map_get($rev)) {
2948                        return ($rev, $c);
2949                }
2950                --$rev;
2951        }
2952        return (undef, undef);
2953}
2954
2955# Finds the first svn revision that exists on (if $eq_ok is true) or
2956# after $rev for the current branch.  It will not search any higher
2957# than $max_rev.  Returns the git commit hash and svn revision number
2958# if found, else (undef, undef).
2959sub find_rev_after {
2960        my ($self, $rev, $eq_ok, $max_rev) = @_;
2961        ++$rev unless $eq_ok;
2962        $max_rev ||= $self->rev_map_max;
2963        while ($rev <= $max_rev) {
2964                if (my $c = $self->rev_map_get($rev)) {
2965                        return ($rev, $c);
2966                }
2967                ++$rev;
2968        }
2969        return (undef, undef);
2970}
2971
2972sub _new {
2973        my ($class, $repo_id, $ref_id, $path) = @_;
2974        unless (defined $repo_id && length $repo_id) {
2975                $repo_id = $Git::SVN::default_repo_id;
2976        }
2977        unless (defined $ref_id && length $ref_id) {
2978                $_[2] = $ref_id = $Git::SVN::default_ref_id;
2979        }
2980        $_[1] = $repo_id;
2981        my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
2982        $_[3] = $path = '' unless (defined $path);
2983        mkpath(["$ENV{GIT_DIR}/svn"]);
2984        bless {
2985                ref_id => $ref_id, dir => $dir, index => "$dir/index",
2986                path => $path, config => "$ENV{GIT_DIR}/svn/config",
2987                map_root => "$dir/.rev_map", repo_id => $repo_id }, $class;
2988}
2989
2990# for read-only access of old .rev_db formats
2991sub unlink_rev_db_symlink {
2992        my ($self) = @_;
2993        my $link = $self->rev_db_path;
2994        $link =~ s/\.[\w-]+$// or croak "missing UUID at the end of $link";
2995        if (-l $link) {
2996                unlink $link or croak "unlink: $link failed!";
2997        }
2998}
2999
3000sub rev_db_path {
3001        my ($self, $uuid) = @_;
3002        my $db_path = $self->map_path($uuid);
3003        $db_path =~ s{/\.rev_map\.}{/\.rev_db\.}
3004            or croak "map_path: $db_path does not contain '/.rev_map.' !";
3005        $db_path;
3006}
3007
3008# the new replacement for .rev_db
3009sub map_path {
3010        my ($self, $uuid) = @_;
3011        $uuid ||= $self->ra_uuid;
3012        "$self->{map_root}.$uuid";
3013}
3014
3015sub uri_encode {
3016        my ($f) = @_;
3017        $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
3018        $f
3019}
3020
3021sub remove_username {
3022        $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
3023}
3024
3025package Git::SVN::Prompt;
3026use strict;
3027use warnings;
3028require SVN::Core;
3029use vars qw/$_no_auth_cache $_username/;
3030
3031sub simple {
3032        my ($cred, $realm, $default_username, $may_save, $pool) = @_;
3033        $may_save = undef if $_no_auth_cache;
3034        $default_username = $_username if defined $_username;
3035        if (defined $default_username && length $default_username) {
3036                if (defined $realm && length $realm) {
3037                        print STDERR "Authentication realm: $realm\n";
3038                        STDERR->flush;
3039                }
3040                $cred->username($default_username);
3041        } else {
3042                username($cred, $realm, $may_save, $pool);
3043        }
3044        $cred->password(_read_password("Password for '" .
3045                                       $cred->username . "': ", $realm));
3046        $cred->may_save($may_save);
3047        $SVN::_Core::SVN_NO_ERROR;
3048}
3049
3050sub ssl_server_trust {
3051        my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
3052        $may_save = undef if $_no_auth_cache;
3053        print STDERR "Error validating server certificate for '$realm':\n";
3054        {
3055                no warnings 'once';
3056                # All variables SVN::Auth::SSL::* are used only once,
3057                # so we're shutting up Perl warnings about this.
3058                if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
3059                        print STDERR " - The certificate is not issued ",
3060                            "by a trusted authority. Use the\n",
3061                            "   fingerprint to validate ",
3062                            "the certificate manually!\n";
3063                }
3064                if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
3065                        print STDERR " - The certificate hostname ",
3066                            "does not match.\n";
3067                }
3068                if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
3069                        print STDERR " - The certificate is not yet valid.\n";
3070                }
3071                if ($failures & $SVN::Auth::SSL::EXPIRED) {
3072                        print STDERR " - The certificate has expired.\n";
3073                }
3074                if ($failures & $SVN::Auth::SSL::OTHER) {
3075                        print STDERR " - The certificate has ",
3076                            "an unknown error.\n";
3077                }
3078        } # no warnings 'once'
3079        printf STDERR
3080                "Certificate information:\n".
3081                " - Hostname: %s\n".
3082                " - Valid: from %s until %s\n".
3083                " - Issuer: %s\n".
3084                " - Fingerprint: %s\n",
3085                map $cert_info->$_, qw(hostname valid_from valid_until
3086                                       issuer_dname fingerprint);
3087        my $choice;
3088prompt:
3089        print STDERR $may_save ?
3090              "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
3091              "(R)eject or accept (t)emporarily? ";
3092        STDERR->flush;
3093        $choice = lc(substr(<STDIN> || 'R', 0, 1));
3094        if ($choice =~ /^t$/i) {
3095                $cred->may_save(undef);
3096        } elsif ($choice =~ /^r$/i) {
3097                return -1;
3098        } elsif ($may_save && $choice =~ /^p$/i) {
3099                $cred->may_save($may_save);
3100        } else {
3101                goto prompt;
3102        }
3103        $cred->accepted_failures($failures);
3104        $SVN::_Core::SVN_NO_ERROR;
3105}
3106
3107sub ssl_client_cert {
3108        my ($cred, $realm, $may_save, $pool) = @_;
3109        $may_save = undef if $_no_auth_cache;
3110        print STDERR "Client certificate filename: ";
3111        STDERR->flush;
3112        chomp(my $filename = <STDIN>);
3113        $cred->cert_file($filename);
3114        $cred->may_save($may_save);
3115        $SVN::_Core::SVN_NO_ERROR;
3116}
3117
3118sub ssl_client_cert_pw {
3119        my ($cred, $realm, $may_save, $pool) = @_;
3120        $may_save = undef if $_no_auth_cache;
3121        $cred->password(_read_password("Password: ", $realm));
3122        $cred->may_save($may_save);
3123        $SVN::_Core::SVN_NO_ERROR;
3124}
3125
3126sub username {
3127        my ($cred, $realm, $may_save, $pool) = @_;
3128        $may_save = undef if $_no_auth_cache;
3129        if (defined $realm && length $realm) {
3130                print STDERR "Authentication realm: $realm\n";
3131        }
3132        my $username;
3133        if (defined $_username) {
3134                $username = $_username;
3135        } else {
3136                print STDERR "Username: ";
3137                STDERR->flush;
3138                chomp($username = <STDIN>);
3139        }
3140        $cred->username($username);
3141        $cred->may_save($may_save);
3142        $SVN::_Core::SVN_NO_ERROR;
3143}
3144
3145sub _read_password {
3146        my ($prompt, $realm) = @_;
3147        print STDERR $prompt;
3148        STDERR->flush;
3149        require Term::ReadKey;
3150        Term::ReadKey::ReadMode('noecho');
3151        my $password = '';
3152        while (defined(my $key = Term::ReadKey::ReadKey(0))) {
3153                last if $key =~ /[\012\015]/; # \n\r
3154                $password .= $key;
3155        }
3156        Term::ReadKey::ReadMode('restore');
3157        print STDERR "\n";
3158        STDERR->flush;
3159        $password;
3160}
3161
3162package SVN::Git::Fetcher;
3163use vars qw/@ISA/;
3164use strict;
3165use warnings;
3166use Carp qw/croak/;
3167use File::Temp qw/tempfile/;
3168use IO::File qw//;
3169
3170# file baton members: path, mode_a, mode_b, pool, fh, blob, base
3171sub new {
3172        my ($class, $git_svn) = @_;
3173        my $self = SVN::Delta::Editor->new;
3174        bless $self, $class;
3175        $self->{c} = $git_svn->{last_commit} if exists $git_svn->{last_commit};
3176        $self->{empty} = {};
3177        $self->{dir_prop} = {};
3178        $self->{file_prop} = {};
3179        $self->{absent_dir} = {};
3180        $self->{absent_file} = {};
3181        $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
3182        $self;
3183}
3184
3185sub set_path_strip {
3186        my ($self, $path) = @_;
3187        $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
3188}
3189
3190sub open_root {
3191        { path => '' };
3192}
3193
3194sub open_directory {
3195        my ($self, $path, $pb, $rev) = @_;
3196        { path => $path };
3197}
3198
3199sub git_path {
3200        my ($self, $path) = @_;
3201        if ($self->{path_strip}) {
3202                $path =~ s!$self->{path_strip}!! or
3203                  die "Failed to strip path '$path' ($self->{path_strip})\n";
3204        }
3205        $path;
3206}
3207
3208sub delete_entry {
3209        my ($self, $path, $rev, $pb) = @_;
3210
3211        my $gpath = $self->git_path($path);
3212        return undef if ($gpath eq '');
3213
3214        # remove entire directories.
3215        if (command('ls-tree', $self->{c}, '--', $gpath) =~ /^040000 tree/) {
3216                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3217                                                     -r --name-only -z/,
3218                                                     $self->{c}, '--', $gpath);
3219                local $/ = "\0";
3220                while (<$ls>) {
3221                        chomp;
3222                        $self->{gii}->remove($_);
3223                        print "\tD\t$_\n" unless $::_q;
3224                }
3225                print "\tD\t$gpath/\n" unless $::_q;
3226                command_close_pipe($ls, $ctx);
3227                $self->{empty}->{$path} = 0
3228        } else {
3229                $self->{gii}->remove($gpath);
3230                print "\tD\t$gpath\n" unless $::_q;
3231        }
3232        undef;
3233}
3234
3235sub open_file {
3236        my ($self, $path, $pb, $rev) = @_;
3237        my $gpath = $self->git_path($path);
3238        my ($mode, $blob) = (command('ls-tree', $self->{c}, '--', $gpath)
3239                             =~ /^(\d{6}) blob ([a-f\d]{40})\t/);
3240        unless (defined $mode && defined $blob) {
3241                die "$path was not found in commit $self->{c} (r$rev)\n";
3242        }
3243        { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
3244          pool => SVN::Pool->new, action => 'M' };
3245}
3246
3247sub add_file {
3248        my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
3249        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3250        delete $self->{empty}->{$dir};
3251        { path => $path, mode_a => 100644, mode_b => 100644,
3252          pool => SVN::Pool->new, action => 'A' };
3253}
3254
3255sub add_directory {
3256        my ($self, $path, $cp_path, $cp_rev) = @_;
3257        my $gpath = $self->git_path($path);
3258        if ($gpath eq '') {
3259                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3260                                                     -r --name-only -z/,
3261                                                     $self->{c});
3262                local $/ = "\0";
3263                while (<$ls>) {
3264                        chomp;
3265                        $self->{gii}->remove($_);
3266                        print "\tD\t$_\n" unless $::_q;
3267                }
3268                command_close_pipe($ls, $ctx);
3269                $self->{empty}->{$path} = 0;
3270        }
3271        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3272        delete $self->{empty}->{$dir};
3273        $self->{empty}->{$path} = 1;
3274        { path => $path };
3275}
3276
3277sub change_dir_prop {
3278        my ($self, $db, $prop, $value) = @_;
3279        $self->{dir_prop}->{$db->{path}} ||= {};
3280        $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
3281        undef;
3282}
3283
3284sub absent_directory {
3285        my ($self, $path, $pb) = @_;
3286        $self->{absent_dir}->{$pb->{path}} ||= [];
3287        push @{$self->{absent_dir}->{$pb->{path}}}, $path;
3288        undef;
3289}
3290
3291sub absent_file {
3292        my ($self, $path, $pb) = @_;
3293        $self->{absent_file}->{$pb->{path}} ||= [];
3294        push @{$self->{absent_file}->{$pb->{path}}}, $path;
3295        undef;
3296}
3297
3298sub change_file_prop {
3299        my ($self, $fb, $prop, $value) = @_;
3300        if ($prop eq 'svn:executable') {
3301                if ($fb->{mode_b} != 120000) {
3302                        $fb->{mode_b} = defined $value ? 100755 : 100644;
3303                }
3304        } elsif ($prop eq 'svn:special') {
3305                $fb->{mode_b} = defined $value ? 120000 : 100644;
3306        } else {
3307                $self->{file_prop}->{$fb->{path}} ||= {};
3308                $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
3309        }
3310        undef;
3311}
3312
3313sub apply_textdelta {
3314        my ($self, $fb, $exp) = @_;
3315        my $fh = Git::temp_acquire('svn_delta');
3316        # $fh gets auto-closed() by SVN::TxDelta::apply(),
3317        # (but $base does not,) so dup() it for reading in close_file
3318        open my $dup, '<&', $fh or croak $!;
3319        my $base = Git::temp_acquire('git_blob');
3320        if ($fb->{blob}) {
3321                print $base 'link ' if ($fb->{mode_a} == 120000);
3322                my $size = $::_repository->cat_blob($fb->{blob}, $base);
3323                die "Failed to read object $fb->{blob}" if ($size < 0);
3324
3325                if (defined $exp) {
3326                        seek $base, 0, 0 or croak $!;
3327                        my $got = ::md5sum($base);
3328                        die "Checksum mismatch: $fb->{path} $fb->{blob}\n",
3329                            "expected: $exp\n",
3330                            "     got: $got\n" if ($got ne $exp);
3331                }
3332        }
3333        seek $base, 0, 0 or croak $!;
3334        $fb->{fh} = $fh;
3335        $fb->{base} = $base;
3336        [ SVN::TxDelta::apply($base, $dup, undef, $fb->{path}, $fb->{pool}) ];
3337}
3338
3339sub close_file {
3340        my ($self, $fb, $exp) = @_;
3341        my $hash;
3342        my $path = $self->git_path($fb->{path});
3343        if (my $fh = $fb->{fh}) {
3344                if (defined $exp) {
3345                        seek($fh, 0, 0) or croak $!;
3346                        my $got = ::md5sum($fh);
3347                        if ($got ne $exp) {
3348                                die "Checksum mismatch: $path\n",
3349                                    "expected: $exp\n    got: $got\n";
3350                        }
3351                }
3352                if ($fb->{mode_b} == 120000) {
3353                        sysseek($fh, 0, 0) or croak $!;
3354                        sysread($fh, my $buf, 5) == 5 or croak $!;
3355
3356                        unless ($buf eq 'link ') {
3357                                warn "$path has mode 120000",
3358                                                " but is not a link\n";
3359                        } else {
3360                                my $tmp_fh = Git::temp_acquire('svn_hash');
3361                                my $res;
3362                                while ($res = sysread($fh, my $str, 1024)) {
3363                                        my $out = syswrite($tmp_fh, $str, $res);
3364                                        defined($out) && $out == $res
3365                                                or croak("write ",
3366                                                        Git::temp_path($tmp_fh),
3367                                                        ": $!\n");
3368                                }
3369                                defined $res or croak $!;
3370
3371                                ($fh, $tmp_fh) = ($tmp_fh, $fh);
3372                                Git::temp_release($tmp_fh, 1);
3373                        }
3374                }
3375
3376                $hash = $::_repository->hash_and_insert_object(
3377                                Git::temp_path($fh));
3378                $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
3379
3380                Git::temp_release($fb->{base}, 1);
3381                Git::temp_release($fh, 1);
3382        } else {
3383                $hash = $fb->{blob} or die "no blob information\n";
3384        }
3385        $fb->{pool}->clear;
3386        $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
3387        print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
3388        undef;
3389}
3390
3391sub abort_edit {
3392        my $self = shift;
3393        $self->{nr} = $self->{gii}->{nr};
3394        delete $self->{gii};
3395        $self->SUPER::abort_edit(@_);
3396}
3397
3398sub close_edit {
3399        my $self = shift;
3400        $self->{git_commit_ok} = 1;
3401        $self->{nr} = $self->{gii}->{nr};
3402        delete $self->{gii};
3403        $self->SUPER::close_edit(@_);
3404}
3405
3406package SVN::Git::Editor;
3407use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
3408use strict;
3409use warnings;
3410use Carp qw/croak/;
3411use IO::File;
3412
3413sub new {
3414        my ($class, $opts) = @_;
3415        foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
3416                die "$_ required!\n" unless (defined $opts->{$_});
3417        }
3418
3419        my $pool = SVN::Pool->new;
3420        my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
3421        my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
3422                                     $opts->{r}, $mods);
3423
3424        # $opts->{ra} functions should not be used after this:
3425        my @ce  = $opts->{ra}->get_commit_editor($opts->{log},
3426                                                $opts->{editor_cb}, $pool);
3427        my $self = SVN::Delta::Editor->new(@ce, $pool);
3428        bless $self, $class;
3429        foreach (qw/svn_path r tree_a tree_b/) {
3430                $self->{$_} = $opts->{$_};
3431        }
3432        $self->{url} = $opts->{ra}->{url};
3433        $self->{mods} = $mods;
3434        $self->{types} = $types;
3435        $self->{pool} = $pool;
3436        $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
3437        $self->{rm} = { };
3438        $self->{path_prefix} = length $self->{svn_path} ?
3439                               "$self->{svn_path}/" : '';
3440        $self->{config} = $opts->{config};
3441        return $self;
3442}
3443
3444sub generate_diff {
3445        my ($tree_a, $tree_b) = @_;
3446        my @diff_tree = qw(diff-tree -z -r);
3447        if ($_cp_similarity) {
3448                push @diff_tree, "-C$_cp_similarity";
3449        } else {
3450                push @diff_tree, '-C';
3451        }
3452        push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
3453        push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
3454        push @diff_tree, $tree_a, $tree_b;
3455        my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
3456        local $/ = "\0";
3457        my $state = 'meta';
3458        my @mods;
3459        while (<$diff_fh>) {
3460                chomp $_; # this gets rid of the trailing "\0"
3461                if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
3462                                        ($::sha1)\s($::sha1)\s
3463                                        ([MTCRAD])\d*$/xo) {
3464                        push @mods, {   mode_a => $1, mode_b => $2,
3465                                        sha1_a => $3, sha1_b => $4,
3466                                        chg => $5 };
3467                        if ($5 =~ /^(?:C|R)$/) {
3468                                $state = 'file_a';
3469                        } else {
3470                                $state = 'file_b';
3471                        }
3472                } elsif ($state eq 'file_a') {
3473                        my $x = $mods[$#mods] or croak "Empty array\n";
3474                        if ($x->{chg} !~ /^(?:C|R)$/) {
3475                                croak "Error parsing $_, $x->{chg}\n";
3476                        }
3477                        $x->{file_a} = $_;
3478                        $state = 'file_b';
3479                } elsif ($state eq 'file_b') {
3480                        my $x = $mods[$#mods] or croak "Empty array\n";
3481                        if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
3482                                croak "Error parsing $_, $x->{chg}\n";
3483                        }
3484                        if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
3485                                croak "Error parsing $_, $x->{chg}\n";
3486                        }
3487                        $x->{file_b} = $_;
3488                        $state = 'meta';
3489                } else {
3490                        croak "Error parsing $_\n";
3491                }
3492        }
3493        command_close_pipe($diff_fh, $ctx);
3494        \@mods;
3495}
3496
3497sub check_diff_paths {
3498        my ($ra, $pfx, $rev, $mods) = @_;
3499        my %types;
3500        $pfx .= '/' if length $pfx;
3501
3502        sub type_diff_paths {
3503                my ($ra, $types, $path, $rev) = @_;
3504                my @p = split m#/+#, $path;
3505                my $c = shift @p;
3506                unless (defined $types->{$c}) {
3507                        $types->{$c} = $ra->check_path($c, $rev);
3508                }
3509                while (@p) {
3510                        $c .= '/' . shift @p;
3511                        next if defined $types->{$c};
3512                        $types->{$c} = $ra->check_path($c, $rev);
3513                }
3514        }
3515
3516        foreach my $m (@$mods) {
3517                foreach my $f (qw/file_a file_b/) {
3518                        next unless defined $m->{$f};
3519                        my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
3520                        if (length $pfx.$dir && ! defined $types{$dir}) {
3521                                type_diff_paths($ra, \%types, $pfx.$dir, $rev);
3522                        }
3523                }
3524        }
3525        \%types;
3526}
3527
3528sub split_path {
3529        return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
3530}
3531
3532sub repo_path {
3533        my ($self, $path) = @_;
3534        $self->{path_prefix}.(defined $path ? $path : '');
3535}
3536
3537sub url_path {
3538        my ($self, $path) = @_;
3539        if ($self->{url} =~ m#^https?://#) {
3540                $path =~ s/([^a-zA-Z0-9_.-])/uc sprintf("%%%02x",ord($1))/eg;
3541        }
3542        $self->{url} . '/' . $self->repo_path($path);
3543}
3544
3545sub rmdirs {
3546        my ($self) = @_;
3547        my $rm = $self->{rm};
3548        delete $rm->{''}; # we never delete the url we're tracking
3549        return unless %$rm;
3550
3551        foreach (keys %$rm) {
3552                my @d = split m#/#, $_;
3553                my $c = shift @d;
3554                $rm->{$c} = 1;
3555                while (@d) {
3556                        $c .= '/' . shift @d;
3557                        $rm->{$c} = 1;
3558                }
3559        }
3560        delete $rm->{$self->{svn_path}};
3561        delete $rm->{''}; # we never delete the url we're tracking
3562        return unless %$rm;
3563
3564        my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
3565                                             $self->{tree_b});
3566        local $/ = "\0";
3567        while (<$fh>) {
3568                chomp;
3569                my @dn = split m#/#, $_;
3570                while (pop @dn) {
3571                        delete $rm->{join '/', @dn};
3572                }
3573                unless (%$rm) {
3574                        close $fh;
3575                        return;
3576                }
3577        }
3578        command_close_pipe($fh, $ctx);
3579
3580        my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
3581        foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
3582                $self->close_directory($bat->{$d}, $p);
3583                my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
3584                print "\tD+\t$d/\n" unless $::_q;
3585                $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
3586                delete $bat->{$d};
3587        }
3588}
3589
3590sub open_or_add_dir {
3591        my ($self, $full_path, $baton) = @_;
3592        my $t = $self->{types}->{$full_path};
3593        if (!defined $t) {
3594                die "$full_path not known in r$self->{r} or we have a bug!\n";
3595        }
3596        {
3597                no warnings 'once';
3598                # SVN::Node::none and SVN::Node::file are used only once,
3599                # so we're shutting up Perl's warnings about them.
3600                if ($t == $SVN::Node::none) {
3601                        return $self->add_directory($full_path, $baton,
3602                            undef, -1, $self->{pool});
3603                } elsif ($t == $SVN::Node::dir) {
3604                        return $self->open_directory($full_path, $baton,
3605                            $self->{r}, $self->{pool});
3606                } # no warnings 'once'
3607                print STDERR "$full_path already exists in repository at ",
3608                    "r$self->{r} and it is not a directory (",
3609                    ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
3610        } # no warnings 'once'
3611        exit 1;
3612}
3613
3614sub ensure_path {
3615        my ($self, $path) = @_;
3616        my $bat = $self->{bat};
3617        my $repo_path = $self->repo_path($path);
3618        return $bat->{''} unless (length $repo_path);
3619        my @p = split m#/+#, $repo_path;
3620        my $c = shift @p;
3621        $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
3622        while (@p) {
3623                my $c0 = $c;
3624                $c .= '/' . shift @p;
3625                $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
3626        }
3627        return $bat->{$c};
3628}
3629
3630# Subroutine to convert a globbing pattern to a regular expression.
3631# From perl cookbook.
3632sub glob2pat {
3633        my $globstr = shift;
3634        my %patmap = ('*' => '.*', '?' => '.', '[' => '[', ']' => ']');
3635        $globstr =~ s{(.)} { $patmap{$1} || "\Q$1" }ge;
3636        return '^' . $globstr . '$';
3637}
3638
3639sub check_autoprop {
3640        my ($self, $pattern, $properties, $file, $fbat) = @_;
3641        # Convert the globbing pattern to a regular expression.
3642        my $regex = glob2pat($pattern);
3643        # Check if the pattern matches the file name.
3644        if($file =~ m/($regex)/) {
3645                # Parse the list of properties to set.
3646                my @props = split(/;/, $properties);
3647                foreach my $prop (@props) {
3648                        # Parse 'name=value' syntax and set the property.
3649                        if ($prop =~ /([^=]+)=(.*)/) {
3650                                my ($n,$v) = ($1,$2);
3651                                for ($n, $v) {
3652                                        s/^\s+//; s/\s+$//;
3653                                }
3654                                $self->change_file_prop($fbat, $n, $v);
3655                        }
3656                }
3657        }
3658}
3659
3660sub apply_autoprops {
3661        my ($self, $file, $fbat) = @_;
3662        my $conf_t = ${$self->{config}}{'config'};
3663        no warnings 'once';
3664        # Check [miscellany]/enable-auto-props in svn configuration.
3665        if (SVN::_Core::svn_config_get_bool(
3666                $conf_t,
3667                $SVN::_Core::SVN_CONFIG_SECTION_MISCELLANY,
3668                $SVN::_Core::SVN_CONFIG_OPTION_ENABLE_AUTO_PROPS,
3669                0)) {
3670                # Auto-props are enabled.  Enumerate them to look for matches.
3671                my $callback = sub {
3672                        $self->check_autoprop($_[0], $_[1], $file, $fbat);
3673                };
3674                SVN::_Core::svn_config_enumerate(
3675                        $conf_t,
3676                        $SVN::_Core::SVN_CONFIG_SECTION_AUTO_PROPS,
3677                        $callback);
3678        }
3679}
3680
3681sub A {
3682        my ($self, $m) = @_;
3683        my ($dir, $file) = split_path($m->{file_b});
3684        my $pbat = $self->ensure_path($dir);
3685        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3686                                        undef, -1);
3687        print "\tA\t$m->{file_b}\n" unless $::_q;
3688        $self->apply_autoprops($file, $fbat);
3689        $self->chg_file($fbat, $m);
3690        $self->close_file($fbat,undef,$self->{pool});
3691}
3692
3693sub C {
3694        my ($self, $m) = @_;
3695        my ($dir, $file) = split_path($m->{file_b});
3696        my $pbat = $self->ensure_path($dir);
3697        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3698                                $self->url_path($m->{file_a}), $self->{r});
3699        print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
3700        $self->chg_file($fbat, $m);
3701        $self->close_file($fbat,undef,$self->{pool});
3702}
3703
3704sub delete_entry {
3705        my ($self, $path, $pbat) = @_;
3706        my $rpath = $self->repo_path($path);
3707        my ($dir, $file) = split_path($rpath);
3708        $self->{rm}->{$dir} = 1;
3709        $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
3710}
3711
3712sub R {
3713        my ($self, $m) = @_;
3714        my ($dir, $file) = split_path($m->{file_b});
3715        my $pbat = $self->ensure_path($dir);
3716        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3717                                $self->url_path($m->{file_a}), $self->{r});
3718        print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
3719        $self->apply_autoprops($file, $fbat);
3720        $self->chg_file($fbat, $m);
3721        $self->close_file($fbat,undef,$self->{pool});
3722
3723        ($dir, $file) = split_path($m->{file_a});
3724        $pbat = $self->ensure_path($dir);
3725        $self->delete_entry($m->{file_a}, $pbat);
3726}
3727
3728sub M {
3729        my ($self, $m) = @_;
3730        my ($dir, $file) = split_path($m->{file_b});
3731        my $pbat = $self->ensure_path($dir);
3732        my $fbat = $self->open_file($self->repo_path($m->{file_b}),
3733                                $pbat,$self->{r},$self->{pool});
3734        print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
3735        $self->chg_file($fbat, $m);
3736        $self->close_file($fbat,undef,$self->{pool});
3737}
3738
3739sub T { shift->M(@_) }
3740
3741sub change_file_prop {
3742        my ($self, $fbat, $pname, $pval) = @_;
3743        $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
3744}
3745
3746sub _chg_file_get_blob ($$$$) {
3747        my ($self, $fbat, $m, $which) = @_;
3748        my $fh = Git::temp_acquire("git_blob_$which");
3749        if ($m->{"mode_$which"} =~ /^120/) {
3750                print $fh 'link ' or croak $!;
3751                $self->change_file_prop($fbat,'svn:special','*');
3752        } elsif ($m->{mode_a} =~ /^120/ && $m->{"mode_$which"} !~ /^120/) {
3753                $self->change_file_prop($fbat,'svn:special',undef);
3754        }
3755        my $blob = $m->{"sha1_$which"};
3756        return ($fh,) if ($blob =~ /^0{40}$/);
3757        my $size = $::_repository->cat_blob($blob, $fh);
3758        croak "Failed to read object $blob" if ($size < 0);
3759        $fh->flush == 0 or croak $!;
3760        seek $fh, 0, 0 or croak $!;
3761
3762        my $exp = ::md5sum($fh);
3763        seek $fh, 0, 0 or croak $!;
3764        return ($fh, $exp);
3765}
3766
3767sub chg_file {
3768        my ($self, $fbat, $m) = @_;
3769        if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
3770                $self->change_file_prop($fbat,'svn:executable','*');
3771        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
3772                $self->change_file_prop($fbat,'svn:executable',undef);
3773        }
3774        my ($fh_a, $exp_a) = _chg_file_get_blob $self, $fbat, $m, 'a';
3775        my ($fh_b, $exp_b) = _chg_file_get_blob $self, $fbat, $m, 'b';
3776        my $pool = SVN::Pool->new;
3777        my $atd = $self->apply_textdelta($fbat, $exp_a, $pool);
3778        if (-s $fh_a) {
3779                my $txstream = SVN::TxDelta::new ($fh_a, $fh_b, $pool);
3780                my $res = SVN::TxDelta::send_txstream($txstream, @$atd, $pool);
3781                if (defined $res) {
3782                        die "Unexpected result from send_txstream: $res\n",
3783                            "(SVN::Core::VERSION: $SVN::Core::VERSION)\n";
3784                }
3785        } else {
3786                my $got = SVN::TxDelta::send_stream($fh_b, @$atd, $pool);
3787                die "Checksum mismatch\nexpected: $exp_b\ngot: $got\n"
3788                    if ($got ne $exp_b);
3789        }
3790        Git::temp_release($fh_b, 1);
3791        Git::temp_release($fh_a, 1);
3792        $pool->clear;
3793}
3794
3795sub D {
3796        my ($self, $m) = @_;
3797        my ($dir, $file) = split_path($m->{file_b});
3798        my $pbat = $self->ensure_path($dir);
3799        print "\tD\t$m->{file_b}\n" unless $::_q;
3800        $self->delete_entry($m->{file_b}, $pbat);
3801}
3802
3803sub close_edit {
3804        my ($self) = @_;
3805        my ($p,$bat) = ($self->{pool}, $self->{bat});
3806        foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
3807                next if $_ eq '';
3808                $self->close_directory($bat->{$_}, $p);
3809        }
3810        $self->close_directory($bat->{''}, $p);
3811        $self->SUPER::close_edit($p);
3812        $p->clear;
3813}
3814
3815sub abort_edit {
3816        my ($self) = @_;
3817        $self->SUPER::abort_edit($self->{pool});
3818}
3819
3820sub DESTROY {
3821        my $self = shift;
3822        $self->SUPER::DESTROY(@_);
3823        $self->{pool}->clear;
3824}
3825
3826# this drives the editor
3827sub apply_diff {
3828        my ($self) = @_;
3829        my $mods = $self->{mods};
3830        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
3831        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
3832                my $f = $m->{chg};
3833                if (defined $o{$f}) {
3834                        $self->$f($m);
3835                } else {
3836                        fatal("Invalid change type: $f");
3837                }
3838        }
3839        $self->rmdirs if $_rmdir;
3840        if (@$mods == 0) {
3841                $self->abort_edit;
3842        } else {
3843                $self->close_edit;
3844        }
3845        return scalar @$mods;
3846}
3847
3848package Git::SVN::Ra;
3849use vars qw/@ISA $config_dir $_log_window_size/;
3850use strict;
3851use warnings;
3852my ($ra_invalid, $can_do_switch, %ignored_err, $RA);
3853
3854BEGIN {
3855        # enforce temporary pool usage for some simple functions
3856        no strict 'refs';
3857        for my $f (qw/rev_proplist get_latest_revnum get_uuid get_repos_root/) {
3858                my $SUPER = "SUPER::$f";
3859                *$f = sub {
3860                        my $self = shift;
3861                        my $pool = SVN::Pool->new;
3862                        my @ret = $self->$SUPER(@_,$pool);
3863                        $pool->clear;
3864                        wantarray ? @ret : $ret[0];
3865                };
3866        }
3867}
3868
3869sub _auth_providers () {
3870        [
3871          SVN::Client::get_simple_provider(),
3872          SVN::Client::get_ssl_server_trust_file_provider(),
3873          SVN::Client::get_simple_prompt_provider(
3874            \&Git::SVN::Prompt::simple, 2),
3875          SVN::Client::get_ssl_client_cert_file_provider(),
3876          SVN::Client::get_ssl_client_cert_prompt_provider(
3877            \&Git::SVN::Prompt::ssl_client_cert, 2),
3878          SVN::Client::get_ssl_client_cert_pw_file_provider(),
3879          SVN::Client::get_ssl_client_cert_pw_prompt_provider(
3880            \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
3881          SVN::Client::get_username_provider(),
3882          SVN::Client::get_ssl_server_trust_prompt_provider(
3883            \&Git::SVN::Prompt::ssl_server_trust),
3884          SVN::Client::get_username_prompt_provider(
3885            \&Git::SVN::Prompt::username, 2)
3886        ]
3887}
3888
3889sub escape_uri_only {
3890        my ($uri) = @_;
3891        my @tmp;
3892        foreach (split m{/}, $uri) {
3893                s/([^\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
3894                push @tmp, $_;
3895        }
3896        join('/', @tmp);
3897}
3898
3899sub escape_url {
3900        my ($url) = @_;
3901        if ($url =~ m#^(https?)://([^/]+)(.*)$#) {
3902                my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
3903                $url = "$scheme://$domain$uri";
3904        }
3905        $url;
3906}
3907
3908sub new {
3909        my ($class, $url) = @_;
3910        $url =~ s!/+$!!;
3911        return $RA if ($RA && $RA->{url} eq $url);
3912
3913        SVN::_Core::svn_config_ensure($config_dir, undef);
3914        my ($baton, $callbacks) = SVN::Core::auth_open_helper(_auth_providers);
3915        my $config = SVN::Core::config_get_config($config_dir);
3916        $RA = undef;
3917        my $dont_store_passwords = 1;
3918        my $conf_t = ${$config}{'config'};
3919        {
3920                no warnings 'once';
3921                # The usage of $SVN::_Core::SVN_CONFIG_* variables
3922                # produces warnings that variables are used only once.
3923                # I had not found the better way to shut them up, so
3924                # the warnings of type 'once' are disabled in this block.
3925                if (SVN::_Core::svn_config_get_bool($conf_t,
3926                    $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
3927                    $SVN::_Core::SVN_CONFIG_OPTION_STORE_PASSWORDS,
3928                    1) == 0) {
3929                        SVN::_Core::svn_auth_set_parameter($baton,
3930                            $SVN::_Core::SVN_AUTH_PARAM_DONT_STORE_PASSWORDS,
3931                            bless (\$dont_store_passwords, "_p_void"));
3932                }
3933                if (SVN::_Core::svn_config_get_bool($conf_t,
3934                    $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
3935                    $SVN::_Core::SVN_CONFIG_OPTION_STORE_AUTH_CREDS,
3936                    1) == 0) {
3937                        $Git::SVN::Prompt::_no_auth_cache = 1;
3938                }
3939        } # no warnings 'once'
3940        my $self = SVN::Ra->new(url => escape_url($url), auth => $baton,
3941                              config => $config,
3942                              pool => SVN::Pool->new,
3943                              auth_provider_callbacks => $callbacks);
3944        $self->{url} = $url;
3945        $self->{svn_path} = $url;
3946        $self->{repos_root} = $self->get_repos_root;
3947        $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
3948        $self->{cache} = { check_path => { r => 0, data => {} },
3949                           get_dir => { r => 0, data => {} } };
3950        $RA = bless $self, $class;
3951}
3952
3953sub check_path {
3954        my ($self, $path, $r) = @_;
3955        my $cache = $self->{cache}->{check_path};
3956        if ($r == $cache->{r} && exists $cache->{data}->{$path}) {
3957                return $cache->{data}->{$path};
3958        }
3959        my $pool = SVN::Pool->new;
3960        my $t = $self->SUPER::check_path($path, $r, $pool);
3961        $pool->clear;
3962        if ($r != $cache->{r}) {
3963                %{$cache->{data}} = ();
3964                $cache->{r} = $r;
3965        }
3966        $cache->{data}->{$path} = $t;
3967}
3968
3969sub get_dir {
3970        my ($self, $dir, $r) = @_;
3971        my $cache = $self->{cache}->{get_dir};
3972        if ($r == $cache->{r}) {
3973                if (my $x = $cache->{data}->{$dir}) {
3974                        return wantarray ? @$x : $x->[0];
3975                }
3976        }
3977        my $pool = SVN::Pool->new;
3978        my ($d, undef, $props) = $self->SUPER::get_dir($dir, $r, $pool);
3979        my %dirents = map { $_ => { kind => $d->{$_}->kind } } keys %$d;
3980        $pool->clear;
3981        if ($r != $cache->{r}) {
3982                %{$cache->{data}} = ();
3983                $cache->{r} = $r;
3984        }
3985        $cache->{data}->{$dir} = [ \%dirents, $r, $props ];
3986        wantarray ? (\%dirents, $r, $props) : \%dirents;
3987}
3988
3989sub DESTROY {
3990        # do not call the real DESTROY since we store ourselves in $RA
3991}
3992
3993sub get_log {
3994        my ($self, @args) = @_;
3995        my $pool = SVN::Pool->new;
3996        splice(@args, 3, 1) if ($SVN::Core::VERSION le '1.2.0');
3997        my $ret = $self->SUPER::get_log(@args, $pool);
3998        $pool->clear;
3999        $ret;
4000}
4001
4002sub trees_match {
4003        my ($self, $url1, $rev1, $url2, $rev2) = @_;
4004        my $ctx = SVN::Client->new(auth => _auth_providers);
4005        my $out = IO::File->new_tmpfile;
4006
4007        # older SVN (1.1.x) doesn't take $pool as the last parameter for
4008        # $ctx->diff(), so we'll create a default one
4009        my $pool = SVN::Pool->new_default_sub;
4010
4011        $ra_invalid = 1; # this will open a new SVN::Ra connection to $url1
4012        $ctx->diff([], $url1, $rev1, $url2, $rev2, 1, 1, 0, $out, $out);
4013        $out->flush;
4014        my $ret = (($out->stat)[7] == 0);
4015        close $out or croak $!;
4016
4017        $ret;
4018}
4019
4020sub get_commit_editor {
4021        my ($self, $log, $cb, $pool) = @_;
4022        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
4023        $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
4024}
4025
4026sub gs_do_update {
4027        my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
4028        my $new = ($rev_a == $rev_b);
4029        my $path = $gs->{path};
4030
4031        if ($new && -e $gs->{index}) {
4032                unlink $gs->{index} or die
4033                  "Couldn't unlink index: $gs->{index}: $!\n";
4034        }
4035        my $pool = SVN::Pool->new;
4036        $editor->set_path_strip($path);
4037        my (@pc) = split m#/#, $path;
4038        my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
4039                                        1, $editor, $pool);
4040        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
4041
4042        # Since we can't rely on svn_ra_reparent being available, we'll
4043        # just have to do some magic with set_path to make it so
4044        # we only want a partial path.
4045        my $sp = '';
4046        my $final = join('/', @pc);
4047        while (@pc) {
4048                $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
4049                $sp .= '/' if length $sp;
4050                $sp .= shift @pc;
4051        }
4052        die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
4053
4054        $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
4055
4056        $reporter->finish_report($pool);
4057        $pool->clear;
4058        $editor->{git_commit_ok};
4059}
4060
4061# this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
4062# svn_ra_reparent didn't work before 1.4)
4063sub gs_do_switch {
4064        my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
4065        my $path = $gs->{path};
4066        my $pool = SVN::Pool->new;
4067
4068        my $full_url = $self->{url};
4069        my $old_url = $full_url;
4070        $full_url .= '/' . escape_uri_only($path) if length $path;
4071        my ($ra, $reparented);
4072
4073        if ($old_url =~ m#^svn(\+ssh)?://#) {
4074                $_[0] = undef;
4075                $self = undef;
4076                $RA = undef;
4077                $ra = Git::SVN::Ra->new($full_url);
4078                $ra_invalid = 1;
4079        } elsif ($old_url ne $full_url) {
4080                SVN::_Ra::svn_ra_reparent($self->{session}, $full_url, $pool);
4081                $self->{url} = $full_url;
4082                $reparented = 1;
4083        }
4084
4085        $ra ||= $self;
4086        $url_b = escape_url($url_b);
4087        my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
4088        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
4089        $reporter->set_path('', $rev_a, 0, @lock, $pool);
4090        $reporter->finish_report($pool);
4091
4092        if ($reparented) {
4093                SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
4094                $self->{url} = $old_url;
4095        }
4096
4097        $pool->clear;
4098        $editor->{git_commit_ok};
4099}
4100
4101sub longest_common_path {
4102        my ($gsv, $globs) = @_;
4103        my %common;
4104        my $common_max = scalar @$gsv;
4105
4106        foreach my $gs (@$gsv) {
4107                my @tmp = split m#/#, $gs->{path};
4108                my $p = '';
4109                foreach (@tmp) {
4110                        $p .= length($p) ? "/$_" : $_;
4111                        $common{$p} ||= 0;
4112                        $common{$p}++;
4113                }
4114        }
4115        $globs ||= [];
4116        $common_max += scalar @$globs;
4117        foreach my $glob (@$globs) {
4118                my @tmp = split m#/#, $glob->{path}->{left};
4119                my $p = '';
4120                foreach (@tmp) {
4121                        $p .= length($p) ? "/$_" : $_;
4122                        $common{$p} ||= 0;
4123                        $common{$p}++;
4124                }
4125        }
4126
4127        my $longest_path = '';
4128        foreach (sort {length $b <=> length $a} keys %common) {
4129                if ($common{$_} == $common_max) {
4130                        $longest_path = $_;
4131                        last;
4132                }
4133        }
4134        $longest_path;
4135}
4136
4137sub gs_fetch_loop_common {
4138        my ($self, $base, $head, $gsv, $globs) = @_;
4139        return if ($base > $head);
4140        my $inc = $_log_window_size;
4141        my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
4142        my $longest_path = longest_common_path($gsv, $globs);
4143        my $ra_url = $self->{url};
4144        while (1) {
4145                my %revs;
4146                my $err;
4147                my $err_handler = $SVN::Error::handler;
4148                $SVN::Error::handler = sub {
4149                        ($err) = @_;
4150                        skip_unknown_revs($err);
4151                };
4152                sub _cb {
4153                        my ($paths, $r, $author, $date, $log) = @_;
4154                        [ dup_changed_paths($paths),
4155                          { author => $author, date => $date, log => $log } ];
4156                }
4157                $self->get_log([$longest_path], $min, $max, 0, 1, 1,
4158                               sub { $revs{$_[1]} = _cb(@_) });
4159                if ($err && $max >= $head) {
4160                        print STDERR "Path '$longest_path' ",
4161                                     "was probably deleted:\n",
4162                                     $err->expanded_message,
4163                                     "\nWill attempt to follow ",
4164                                     "revisions r$min .. r$max ",
4165                                     "committed before the deletion\n";
4166                        my $hi = $max;
4167                        while (--$hi >= $min) {
4168                                my $ok;
4169                                $self->get_log([$longest_path], $min, $hi,
4170                                               0, 1, 1, sub {
4171                                               $ok ||= $_[1];
4172                                               $revs{$_[1]} = _cb(@_) });
4173                                if ($ok) {
4174                                        print STDERR "r$min .. r$ok OK\n";
4175                                        last;
4176                                }
4177                        }
4178                }
4179                $SVN::Error::handler = $err_handler;
4180
4181                my %exists = map { $_->{path} => $_ } @$gsv;
4182                foreach my $r (sort {$a <=> $b} keys %revs) {
4183                        my ($paths, $logged) = @{$revs{$r}};
4184
4185                        foreach my $gs ($self->match_globs(\%exists, $paths,
4186                                                           $globs, $r)) {
4187                                if ($gs->rev_map_max >= $r) {
4188                                        next;
4189                                }
4190                                next unless $gs->match_paths($paths, $r);
4191                                $gs->{logged_rev_props} = $logged;
4192                                if (my $last_commit = $gs->last_commit) {
4193                                        $gs->assert_index_clean($last_commit);
4194                                }
4195                                my $log_entry = $gs->do_fetch($paths, $r);
4196                                if ($log_entry) {
4197                                        $gs->do_git_commit($log_entry);
4198                                }
4199                                $INDEX_FILES{$gs->{index}} = 1;
4200                        }
4201                        foreach my $g (@$globs) {
4202                                my $k = "svn-remote.$g->{remote}." .
4203                                        "$g->{t}-maxRev";
4204                                Git::SVN::tmp_config($k, $r);
4205                        }
4206                        if ($ra_invalid) {
4207                                $_[0] = undef;
4208                                $self = undef;
4209                                $RA = undef;
4210                                $self = Git::SVN::Ra->new($ra_url);
4211                                $ra_invalid = undef;
4212                        }
4213                }
4214                # pre-fill the .rev_db since it'll eventually get filled in
4215                # with '0' x40 if something new gets committed
4216                foreach my $gs (@$gsv) {
4217                        next if $gs->rev_map_max >= $max;
4218                        next if defined $gs->rev_map_get($max);
4219                        $gs->rev_map_set($max, 0 x40);
4220                }
4221                foreach my $g (@$globs) {
4222                        my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
4223                        Git::SVN::tmp_config($k, $max);
4224                }
4225                last if $max >= $head;
4226                $min = $max + 1;
4227                $max += $inc;
4228                $max = $head if ($max > $head);
4229        }
4230        Git::SVN::gc();
4231}
4232
4233sub get_dir_globbed {
4234        my ($self, $left, $depth, $r) = @_;
4235
4236        my @x = eval { $self->get_dir($left, $r) };
4237        return unless scalar @x == 3;
4238        my $dirents = $x[0];
4239        my @finalents;
4240        foreach my $de (keys %$dirents) {
4241                next if $dirents->{$de}->{kind} != $SVN::Node::dir;
4242                if ($depth > 1) {
4243                        my @args = ("$left/$de", $depth - 1, $r);
4244                        foreach my $dir ($self->get_dir_globbed(@args)) {
4245                                push @finalents, "$de/$dir";
4246                        }
4247                } else {
4248                        push @finalents, $de;
4249                }
4250        }
4251        @finalents;
4252}
4253
4254sub match_globs {
4255        my ($self, $exists, $paths, $globs, $r) = @_;
4256
4257        sub get_dir_check {
4258                my ($self, $exists, $g, $r) = @_;
4259
4260                my @dirs = $self->get_dir_globbed($g->{path}->{left},
4261                                                  $g->{path}->{depth},
4262                                                  $r);
4263
4264                foreach my $de (@dirs) {
4265                        my $p = $g->{path}->full_path($de);
4266                        next if $exists->{$p};
4267                        next if (length $g->{path}->{right} &&
4268                                 ($self->check_path($p, $r) !=
4269                                  $SVN::Node::dir));
4270                        $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
4271                                         $g->{ref}->full_path($de), 1);
4272                }
4273        }
4274        foreach my $g (@$globs) {
4275                if (my $path = $paths->{"/$g->{path}->{left}"}) {
4276                        if ($path->{action} =~ /^[AR]$/) {
4277                                get_dir_check($self, $exists, $g, $r);
4278                        }
4279                }
4280                foreach (keys %$paths) {
4281                        if (/$g->{path}->{left_regex}/ &&
4282                            !/$g->{path}->{regex}/) {
4283                                next if $paths->{$_}->{action} !~ /^[AR]$/;
4284                                get_dir_check($self, $exists, $g, $r);
4285                        }
4286                        next unless /$g->{path}->{regex}/;
4287                        my $p = $1;
4288                        my $pathname = $g->{path}->full_path($p);
4289                        next if $exists->{$pathname};
4290                        next if ($self->check_path($pathname, $r) !=
4291                                 $SVN::Node::dir);
4292                        $exists->{$pathname} = Git::SVN->init(
4293                                              $self->{url}, $pathname, undef,
4294                                              $g->{ref}->full_path($p), 1);
4295                }
4296                my $c = '';
4297                foreach (split m#/#, $g->{path}->{left}) {
4298                        $c .= "/$_";
4299                        next unless ($paths->{$c} &&
4300                                     ($paths->{$c}->{action} =~ /^[AR]$/));
4301                        get_dir_check($self, $exists, $g, $r);
4302                }
4303        }
4304        values %$exists;
4305}
4306
4307sub minimize_url {
4308        my ($self) = @_;
4309        return $self->{url} if ($self->{url} eq $self->{repos_root});
4310        my $url = $self->{repos_root};
4311        my @components = split(m!/!, $self->{svn_path});
4312        my $c = '';
4313        do {
4314                $url .= "/$c" if length $c;
4315                eval { (ref $self)->new($url)->get_latest_revnum };
4316        } while ($@ && ($c = shift @components));
4317        $url;
4318}
4319
4320sub can_do_switch {
4321        my $self = shift;
4322        unless (defined $can_do_switch) {
4323                my $pool = SVN::Pool->new;
4324                my $rep = eval {
4325                        $self->do_switch(1, '', 0, $self->{url},
4326                                         SVN::Delta::Editor->new, $pool);
4327                };
4328                if ($@) {
4329                        $can_do_switch = 0;
4330                } else {
4331                        $rep->abort_report($pool);
4332                        $can_do_switch = 1;
4333                }
4334                $pool->clear;
4335        }
4336        $can_do_switch;
4337}
4338
4339sub skip_unknown_revs {
4340        my ($err) = @_;
4341        my $errno = $err->apr_err();
4342        # Maybe the branch we're tracking didn't
4343        # exist when the repo started, so it's
4344        # not an error if it doesn't, just continue
4345        #
4346        # Wonderfully consistent library, eh?
4347        # 160013 - svn:// and file://
4348        # 175002 - http(s)://
4349        # 175007 - http(s):// (this repo required authorization, too...)
4350        #   More codes may be discovered later...
4351        if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
4352                my $err_key = $err->expanded_message;
4353                # revision numbers change every time, filter them out
4354                $err_key =~ s/\d+/\0/g;
4355                $err_key = "$errno\0$err_key";
4356                unless ($ignored_err{$err_key}) {
4357                        warn "W: Ignoring error from SVN, path probably ",
4358                             "does not exist: ($errno): ",
4359                             $err->expanded_message,"\n";
4360                        warn "W: Do not be alarmed at the above message ",
4361                             "git-svn is just searching aggressively for ",
4362                             "old history.\n",
4363                             "This may take a while on large repositories\n";
4364                        $ignored_err{$err_key} = 1;
4365                }
4366                return;
4367        }
4368        die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
4369}
4370
4371# svn_log_changed_path_t objects passed to get_log are likely to be
4372# overwritten even if only the refs are copied to an external variable,
4373# so we should dup the structures in their entirety.  Using an externally
4374# passed pool (instead of our temporary and quickly cleared pool in
4375# Git::SVN::Ra) does not help matters at all...
4376sub dup_changed_paths {
4377        my ($paths) = @_;
4378        return undef unless $paths;
4379        my %ret;
4380        foreach my $p (keys %$paths) {
4381                my $i = $paths->{$p};
4382                my %s = map { $_ => $i->$_ }
4383                              qw/copyfrom_path copyfrom_rev action/;
4384                $ret{$p} = \%s;
4385        }
4386        \%ret;
4387}
4388
4389package Git::SVN::Log;
4390use strict;
4391use warnings;
4392use POSIX qw/strftime/;
4393use constant commit_log_separator => ('-' x 72) . "\n";
4394use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
4395            %rusers $show_commit $incremental/;
4396my $l_fmt;
4397
4398sub cmt_showable {
4399        my ($c) = @_;
4400        return 1 if defined $c->{r};
4401
4402        # big commit message got truncated by the 16k pretty buffer in rev-list
4403        if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
4404                                $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
4405                @{$c->{l}} = ();
4406                my @log = command(qw/cat-file commit/, $c->{c});
4407
4408                # shift off the headers
4409                shift @log while ($log[0] ne '');
4410                shift @log;
4411
4412                # TODO: make $c->{l} not have a trailing newline in the future
4413                @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
4414
4415                (undef, $c->{r}, undef) = ::extract_metadata(
4416                                (grep(/^git-svn-id: /, @log))[-1]);
4417        }
4418        return defined $c->{r};
4419}
4420
4421sub log_use_color {
4422        return $color || Git->repository->get_colorbool('color.diff');
4423}
4424
4425sub git_svn_log_cmd {
4426        my ($r_min, $r_max, @args) = @_;
4427        my $head = 'HEAD';
4428        my (@files, @log_opts);
4429        foreach my $x (@args) {
4430                if ($x eq '--' || @files) {
4431                        push @files, $x;
4432                } else {
4433                        if (::verify_ref("$x^0")) {
4434                                $head = $x;
4435                        } else {
4436                                push @log_opts, $x;
4437                        }
4438                }
4439        }
4440
4441        my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
4442        $gs ||= Git::SVN->_new;
4443        my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
4444                   $gs->refname);
4445        push @cmd, '-r' unless $non_recursive;
4446        push @cmd, qw/--raw --name-status/ if $verbose;
4447        push @cmd, '--color' if log_use_color();
4448        push @cmd, @log_opts;
4449        if (defined $r_max && $r_max == $r_min) {
4450                push @cmd, '--max-count=1';
4451                if (my $c = $gs->rev_map_get($r_max)) {
4452                        push @cmd, $c;
4453                }
4454        } elsif (defined $r_max) {
4455                if ($r_max < $r_min) {
4456                        ($r_min, $r_max) = ($r_max, $r_min);
4457                }
4458                my (undef, $c_max) = $gs->find_rev_before($r_max, 1, $r_min);
4459                my (undef, $c_min) = $gs->find_rev_after($r_min, 1, $r_max);
4460                # If there are no commits in the range, both $c_max and $c_min
4461                # will be undefined.  If there is at least 1 commit in the
4462                # range, both will be defined.
4463                return () if !defined $c_min || !defined $c_max;
4464                if ($c_min eq $c_max) {
4465                        push @cmd, '--max-count=1', $c_min;
4466                } else {
4467                        push @cmd, '--boundary', "$c_min..$c_max";
4468                }
4469        }
4470        return (@cmd, @files);
4471}
4472
4473# adapted from pager.c
4474sub config_pager {
4475        $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
4476        if (!defined $pager) {
4477                $pager = 'less';
4478        } elsif (length $pager == 0 || $pager eq 'cat') {
4479                $pager = undef;
4480        }
4481        $ENV{GIT_PAGER_IN_USE} = defined($pager);
4482}
4483
4484sub run_pager {
4485        return unless -t *STDOUT && defined $pager;
4486        pipe my ($rfd, $wfd) or return;
4487        defined(my $pid = fork) or ::fatal "Can't fork: $!";
4488        if (!$pid) {
4489                open STDOUT, '>&', $wfd or
4490                                     ::fatal "Can't redirect to stdout: $!";
4491                return;
4492        }
4493        open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!";
4494        $ENV{LESS} ||= 'FRSX';
4495        exec $pager or ::fatal "Can't run pager: $! ($pager)";
4496}
4497
4498sub format_svn_date {
4499        return strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)", localtime(shift));
4500}
4501
4502sub parse_git_date {
4503        my ($t, $tz) = @_;
4504        # Date::Parse isn't in the standard Perl distro :(
4505        if ($tz =~ s/^\+//) {
4506                $t += tz_to_s_offset($tz);
4507        } elsif ($tz =~ s/^\-//) {
4508                $t -= tz_to_s_offset($tz);
4509        }
4510        return $t;
4511}
4512
4513sub set_local_timezone {
4514        if (defined $TZ) {
4515                $ENV{TZ} = $TZ;
4516        } else {
4517                delete $ENV{TZ};
4518        }
4519}
4520
4521sub tz_to_s_offset {
4522        my ($tz) = @_;
4523        $tz =~ s/(\d\d)$//;
4524        return ($1 * 60) + ($tz * 3600);
4525}
4526
4527sub get_author_info {
4528        my ($dest, $author, $t, $tz) = @_;
4529        $author =~ s/(?:^\s*|\s*$)//g;
4530        $dest->{a_raw} = $author;
4531        my $au;
4532        if ($::_authors) {
4533                $au = $rusers{$author} || undef;
4534        }
4535        if (!$au) {
4536                ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
4537        }
4538        $dest->{t} = $t;
4539        $dest->{tz} = $tz;
4540        $dest->{a} = $au;
4541        $dest->{t_utc} = parse_git_date($t, $tz);
4542}
4543
4544sub process_commit {
4545        my ($c, $r_min, $r_max, $defer) = @_;
4546        if (defined $r_min && defined $r_max) {
4547                if ($r_min == $c->{r} && $r_min == $r_max) {
4548                        show_commit($c);
4549                        return 0;
4550                }
4551                return 1 if $r_min == $r_max;
4552                if ($r_min < $r_max) {
4553                        # we need to reverse the print order
4554                        return 0 if (defined $limit && --$limit < 0);
4555                        push @$defer, $c;
4556                        return 1;
4557                }
4558                if ($r_min != $r_max) {
4559                        return 1 if ($r_min < $c->{r});
4560                        return 1 if ($r_max > $c->{r});
4561                }
4562        }
4563        return 0 if (defined $limit && --$limit < 0);
4564        show_commit($c);
4565        return 1;
4566}
4567
4568sub show_commit {
4569        my $c = shift;
4570        if ($oneline) {
4571                my $x = "\n";
4572                if (my $l = $c->{l}) {
4573                        while ($l->[0] =~ /^\s*$/) { shift @$l }
4574                        $x = $l->[0];
4575                }
4576                $l_fmt ||= 'A' . length($c->{r});
4577                print 'r',pack($l_fmt, $c->{r}),' | ';
4578                print "$c->{c} | " if $show_commit;
4579                print $x;
4580        } else {
4581                show_commit_normal($c);
4582        }
4583}
4584
4585sub show_commit_changed_paths {
4586        my ($c) = @_;
4587        return unless $c->{changed};
4588        print "Changed paths:\n", @{$c->{changed}};
4589}
4590
4591sub show_commit_normal {
4592        my ($c) = @_;
4593        print commit_log_separator, "r$c->{r} | ";
4594        print "$c->{c} | " if $show_commit;
4595        print "$c->{a} | ", format_svn_date($c->{t_utc}), ' | ';
4596        my $nr_line = 0;
4597
4598        if (my $l = $c->{l}) {
4599                while ($l->[$#$l] eq "\n" && $#$l > 0
4600                                          && $l->[($#$l - 1)] eq "\n") {
4601                        pop @$l;
4602                }
4603                $nr_line = scalar @$l;
4604                if (!$nr_line) {
4605                        print "1 line\n\n\n";
4606                } else {
4607                        if ($nr_line == 1) {
4608                                $nr_line = '1 line';
4609                        } else {
4610                                $nr_line .= ' lines';
4611                        }
4612                        print $nr_line, "\n";
4613                        show_commit_changed_paths($c);
4614                        print "\n";
4615                        print $_ foreach @$l;
4616                }
4617        } else {
4618                print "1 line\n";
4619                show_commit_changed_paths($c);
4620                print "\n";
4621
4622        }
4623        foreach my $x (qw/raw stat diff/) {
4624                if ($c->{$x}) {
4625                        print "\n";
4626                        print $_ foreach @{$c->{$x}}
4627                }
4628        }
4629}
4630
4631sub cmd_show_log {
4632        my (@args) = @_;
4633        my ($r_min, $r_max);
4634        my $r_last = -1; # prevent dupes
4635        set_local_timezone();
4636        if (defined $::_revision) {
4637                if ($::_revision =~ /^(\d+):(\d+)$/) {
4638                        ($r_min, $r_max) = ($1, $2);
4639                } elsif ($::_revision =~ /^\d+$/) {
4640                        $r_min = $r_max = $::_revision;
4641                } else {
4642                        ::fatal "-r$::_revision is not supported, use ",
4643                                "standard 'git log' arguments instead";
4644                }
4645        }
4646
4647        config_pager();
4648        @args = git_svn_log_cmd($r_min, $r_max, @args);
4649        if (!@args) {
4650                print commit_log_separator unless $incremental || $oneline;
4651                return;
4652        }
4653        my $log = command_output_pipe(@args);
4654        run_pager();
4655        my (@k, $c, $d, $stat);
4656        my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
4657        while (<$log>) {
4658                if (/^${esc_color}commit -?($::sha1_short)/o) {
4659                        my $cmt = $1;
4660                        if ($c && cmt_showable($c) && $c->{r} != $r_last) {
4661                                $r_last = $c->{r};
4662                                process_commit($c, $r_min, $r_max, \@k) or
4663                                                                goto out;
4664                        }
4665                        $d = undef;
4666                        $c = { c => $cmt };
4667                } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
4668                        get_author_info($c, $1, $2, $3);
4669                } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
4670                        # ignore
4671                } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
4672                        push @{$c->{raw}}, $_;
4673                } elsif (/^${esc_color}[ACRMDT]\t/) {
4674                        # we could add $SVN->{svn_path} here, but that requires
4675                        # remote access at the moment (repo_path_split)...
4676                        s#^(${esc_color})([ACRMDT])\t#$1   $2 #o;
4677                        push @{$c->{changed}}, $_;
4678                } elsif (/^${esc_color}diff /o) {
4679                        $d = 1;
4680                        push @{$c->{diff}}, $_;
4681                } elsif ($d) {
4682                        push @{$c->{diff}}, $_;
4683                } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
4684                          $esc_color*[\+\-]*$esc_color$/x) {
4685                        $stat = 1;
4686                        push @{$c->{stat}}, $_;
4687                } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
4688                        push @{$c->{stat}}, $_;
4689                        $stat = undef;
4690                } elsif (/^${esc_color}    (git-svn-id:.+)$/o) {
4691                        ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
4692                } elsif (s/^${esc_color}    //o) {
4693                        push @{$c->{l}}, $_;
4694                }
4695        }
4696        if ($c && defined $c->{r} && $c->{r} != $r_last) {
4697                $r_last = $c->{r};
4698                process_commit($c, $r_min, $r_max, \@k);
4699        }
4700        if (@k) {
4701                ($r_min, $r_max) = ($r_max, $r_min);
4702                process_commit($_, $r_min, $r_max) foreach reverse @k;
4703        }
4704out:
4705        close $log;
4706        print commit_log_separator unless $incremental || $oneline;
4707}
4708
4709sub cmd_blame {
4710        my $path = pop;
4711
4712        config_pager();
4713        run_pager();
4714
4715        my ($fh, $ctx, $rev);
4716
4717        if ($_git_format) {
4718                ($fh, $ctx) = command_output_pipe('blame', @_, $path);
4719                while (my $line = <$fh>) {
4720                        if ($line =~ /^\^?([[:xdigit:]]+)\s/) {
4721                                # Uncommitted edits show up as a rev ID of
4722                                # all zeros, which we can't look up with
4723                                # cmt_metadata
4724                                if ($1 !~ /^0+$/) {
4725                                        (undef, $rev, undef) =
4726                                                ::cmt_metadata($1);
4727                                        $rev = '0' if (!$rev);
4728                                } else {
4729                                        $rev = '0';
4730                                }
4731                                $rev = sprintf('%-10s', $rev);
4732                                $line =~ s/^\^?[[:xdigit:]]+(\s)/$rev$1/;
4733                        }
4734                        print $line;
4735                }
4736        } else {
4737                ($fh, $ctx) = command_output_pipe('blame', '-p', @_, 'HEAD',
4738                                                  '--', $path);
4739                my ($sha1);
4740                my %authors;
4741                while (my $line = <$fh>) {
4742                        if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
4743                                $sha1 = $1;
4744                                (undef, $rev, undef) = ::cmt_metadata($1);
4745                                $rev = '0' if (!$rev);
4746                        }
4747                        elsif ($line =~ /^author (.*)/) {
4748                                $authors{$rev} = $1;
4749                                $authors{$rev} =~ s/\s/_/g;
4750                        }
4751                        elsif ($line =~ /^\t(.*)$/) {
4752                                printf("%6s %10s %s\n", $rev, $authors{$rev}, $1);
4753                        }
4754                }
4755        }
4756        command_close_pipe($fh, $ctx);
4757}
4758
4759package Git::SVN::Migration;
4760# these version numbers do NOT correspond to actual version numbers
4761# of git nor git-svn.  They are just relative.
4762#
4763# v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
4764#
4765# v1 layout: .git/$id/info/url, refs/remotes/$id
4766#
4767# v2 layout: .git/svn/$id/info/url, refs/remotes/$id
4768#
4769# v3 layout: .git/svn/$id, refs/remotes/$id
4770#            - info/url may remain for backwards compatibility
4771#            - this is what we migrate up to this layout automatically,
4772#            - this will be used by git svn init on single branches
4773# v3.1 layout (auto migrated):
4774#            - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
4775#              for backwards compatibility
4776#
4777# v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
4778#            - this is only created for newly multi-init-ed
4779#              repositories.  Similar in spirit to the
4780#              --use-separate-remotes option in git-clone (now default)
4781#            - we do not automatically migrate to this (following
4782#              the example set by core git)
4783#
4784# v5 layout: .rev_db.$UUID => .rev_map.$UUID
4785#            - newer, more-efficient format that uses 24-bytes per record
4786#              with no filler space.
4787#            - use xxd -c24 < .rev_map.$UUID to view and debug
4788#            - This is a one-way migration, repositories updated to the
4789#              new format will not be able to use old git-svn without
4790#              rebuilding the .rev_db.  Rebuilding the rev_db is not
4791#              possible if noMetadata or useSvmProps are set; but should
4792#              be no problem for users that use the (sensible) defaults.
4793use strict;
4794use warnings;
4795use Carp qw/croak/;
4796use File::Path qw/mkpath/;
4797use File::Basename qw/dirname basename/;
4798use vars qw/$_minimize/;
4799
4800sub migrate_from_v0 {
4801        my $git_dir = $ENV{GIT_DIR};
4802        return undef unless -d $git_dir;
4803        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
4804        my $migrated = 0;
4805        while (<$fh>) {
4806                chomp;
4807                my ($id, $orig_ref) = ($_, $_);
4808                next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
4809                next unless -f "$git_dir/$id/info/url";
4810                my $new_ref = "refs/remotes/$id";
4811                if (::verify_ref("$new_ref^0")) {
4812                        print STDERR "W: $orig_ref is probably an old ",
4813                                     "branch used by an ancient version of ",
4814                                     "git-svn.\n",
4815                                     "However, $new_ref also exists.\n",
4816                                     "We will not be able ",
4817                                     "to use this branch until this ",
4818                                     "ambiguity is resolved.\n";
4819                        next;
4820                }
4821                print STDERR "Migrating from v0 layout...\n" if !$migrated;
4822                print STDERR "Renaming ref: $orig_ref => $new_ref\n";
4823                command_noisy('update-ref', $new_ref, $orig_ref);
4824                command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
4825                $migrated++;
4826        }
4827        command_close_pipe($fh, $ctx);
4828        print STDERR "Done migrating from v0 layout...\n" if $migrated;
4829        $migrated;
4830}
4831
4832sub migrate_from_v1 {
4833        my $git_dir = $ENV{GIT_DIR};
4834        my $migrated = 0;
4835        return $migrated unless -d $git_dir;
4836        my $svn_dir = "$git_dir/svn";
4837
4838        # just in case somebody used 'svn' as their $id at some point...
4839        return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
4840
4841        print STDERR "Migrating from a git-svn v1 layout...\n";
4842        mkpath([$svn_dir]);
4843        print STDERR "Data from a previous version of git-svn exists, but\n\t",
4844                     "$svn_dir\n\t(required for this version ",
4845                     "($::VERSION) of git-svn) does not exist.\n";
4846        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
4847        while (<$fh>) {
4848                my $x = $_;
4849                next unless $x =~ s#^refs/remotes/##;
4850                chomp $x;
4851                next unless -f "$git_dir/$x/info/url";
4852                my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
4853                next unless $u;
4854                my $dn = dirname("$git_dir/svn/$x");
4855                mkpath([$dn]) unless -d $dn;
4856                if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
4857                        mkpath(["$git_dir/svn/svn"]);
4858                        print STDERR " - $git_dir/$x/info => ",
4859                                        "$git_dir/svn/$x/info\n";
4860                        rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
4861                               croak "$!: $x";
4862                        # don't worry too much about these, they probably
4863                        # don't exist with repos this old (save for index,
4864                        # and we can easily regenerate that)
4865                        foreach my $f (qw/unhandled.log index .rev_db/) {
4866                                rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
4867                        }
4868                } else {
4869                        print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
4870                        rename "$git_dir/$x", "$git_dir/svn/$x" or
4871                               croak "$!: $x";
4872                }
4873                $migrated++;
4874        }
4875        command_close_pipe($fh, $ctx);
4876        print STDERR "Done migrating from a git-svn v1 layout\n";
4877        $migrated;
4878}
4879
4880sub read_old_urls {
4881        my ($l_map, $pfx, $path) = @_;
4882        my @dir;
4883        foreach (<$path/*>) {
4884                if (-r "$_/info/url") {
4885                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
4886                        my $ref_id = $pfx . basename $_;
4887                        my $url = ::file_to_s("$_/info/url");
4888                        $l_map->{$ref_id} = $url;
4889                } elsif (-d $_) {
4890                        push @dir, $_;
4891                }
4892        }
4893        foreach (@dir) {
4894                my $x = $_;
4895                $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
4896                read_old_urls($l_map, $x, $_);
4897        }
4898}
4899
4900sub migrate_from_v2 {
4901        my @cfg = command(qw/config -l/);
4902        return if grep /^svn-remote\..+\.url=/, @cfg;
4903        my %l_map;
4904        read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
4905        my $migrated = 0;
4906
4907        foreach my $ref_id (sort keys %l_map) {
4908                eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
4909                if ($@) {
4910                        Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
4911                }
4912                $migrated++;
4913        }
4914        $migrated;
4915}
4916
4917sub minimize_connections {
4918        my $r = Git::SVN::read_all_remotes();
4919        my $new_urls = {};
4920        my $root_repos = {};
4921        foreach my $repo_id (keys %$r) {
4922                my $url = $r->{$repo_id}->{url} or next;
4923                my $fetch = $r->{$repo_id}->{fetch} or next;
4924                my $ra = Git::SVN::Ra->new($url);
4925
4926                # skip existing cases where we already connect to the root
4927                if (($ra->{url} eq $ra->{repos_root}) ||
4928                    ($ra->{repos_root} eq $repo_id)) {
4929                        $root_repos->{$ra->{url}} = $repo_id;
4930                        next;
4931                }
4932
4933                my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
4934                my $root_path = $ra->{url};
4935                $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
4936                foreach my $path (keys %$fetch) {
4937                        my $ref_id = $fetch->{$path};
4938                        my $gs = Git::SVN->new($ref_id, $repo_id, $path);
4939
4940                        # make sure we can read when connecting to
4941                        # a higher level of a repository
4942                        my ($last_rev, undef) = $gs->last_rev_commit;
4943                        if (!defined $last_rev) {
4944                                $last_rev = eval {
4945                                        $root_ra->get_latest_revnum;
4946                                };
4947                                next if $@;
4948                        }
4949                        my $new = $root_path;
4950                        $new .= length $path ? "/$path" : '';
4951                        eval {
4952                                $root_ra->get_log([$new], $last_rev, $last_rev,
4953                                                  0, 0, 1, sub { });
4954                        };
4955                        next if $@;
4956                        $new_urls->{$ra->{repos_root}}->{$new} =
4957                                { ref_id => $ref_id,
4958                                  old_repo_id => $repo_id,
4959                                  old_path => $path };
4960                }
4961        }
4962
4963        my @emptied;
4964        foreach my $url (keys %$new_urls) {
4965                # see if we can re-use an existing [svn-remote "repo_id"]
4966                # instead of creating a(n ugly) new section:
4967                my $repo_id = $root_repos->{$url} || $url;
4968
4969                my $fetch = $new_urls->{$url};
4970                foreach my $path (keys %$fetch) {
4971                        my $x = $fetch->{$path};
4972                        Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
4973                        my $pfx = "svn-remote.$x->{old_repo_id}";
4974
4975                        my $old_fetch = quotemeta("$x->{old_path}:".
4976                                                  "refs/remotes/$x->{ref_id}");
4977                        command_noisy(qw/config --unset/,
4978                                      "$pfx.fetch", '^'. $old_fetch . '$');
4979                        delete $r->{$x->{old_repo_id}}->
4980                               {fetch}->{$x->{old_path}};
4981                        if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
4982                                command_noisy(qw/config --unset/,
4983                                              "$pfx.url");
4984                                push @emptied, $x->{old_repo_id}
4985                        }
4986                }
4987        }
4988        if (@emptied) {
4989                my $file = $ENV{GIT_CONFIG} || $ENV{GIT_CONFIG_LOCAL} ||
4990                           "$ENV{GIT_DIR}/config";
4991                print STDERR <<EOF;
4992The following [svn-remote] sections in your config file ($file) are empty
4993and can be safely removed:
4994EOF
4995                print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
4996        }
4997}
4998
4999sub migration_check {
5000        migrate_from_v0();
5001        migrate_from_v1();
5002        migrate_from_v2();
5003        minimize_connections() if $_minimize;
5004}
5005
5006package Git::IndexInfo;
5007use strict;
5008use warnings;
5009use Git qw/command_input_pipe command_close_pipe/;
5010
5011sub new {
5012        my ($class) = @_;
5013        my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
5014        bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
5015}
5016
5017sub remove {
5018        my ($self, $path) = @_;
5019        if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
5020                return ++$self->{nr};
5021        }
5022        undef;
5023}
5024
5025sub update {
5026        my ($self, $mode, $hash, $path) = @_;
5027        if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
5028                return ++$self->{nr};
5029        }
5030        undef;
5031}
5032
5033sub DESTROY {
5034        my ($self) = @_;
5035        command_close_pipe($self->{gui}, $self->{ctx});
5036}
5037
5038package Git::SVN::GlobSpec;
5039use strict;
5040use warnings;
5041
5042sub new {
5043        my ($class, $glob) = @_;
5044        my $re = $glob;
5045        $re =~ s!/+$!!g; # no need for trailing slashes
5046        $re =~ m!^([^*]*)(\*(?:/\*)*)([^*]*)$!;
5047        my $temp = $re;
5048        my ($left, $right) = ($1, $3);
5049        $re = $2;
5050        my $depth = $re =~ tr/*/*/;
5051        if ($depth != $temp =~ tr/*/*/) {
5052                die "Only one set of wildcard directories " .
5053                        "(e.g. '*' or '*/*/*') is supported: '$glob'\n";
5054        }
5055        if ($depth == 0) {
5056                die "One '*' is needed for glob: '$glob'\n";
5057        }
5058        $re =~ s!\*!\[^/\]*!g;
5059        $re = quotemeta($left) . "($re)" . quotemeta($right);
5060        if (length $left && !($left =~ s!/+$!!g)) {
5061                die "Missing trailing '/' on left side of: '$glob' ($left)\n";
5062        }
5063        if (length $right && !($right =~ s!^/+!!g)) {
5064                die "Missing leading '/' on right side of: '$glob' ($right)\n";
5065        }
5066        my $left_re = qr/^\/\Q$left\E(\/|$)/;
5067        bless { left => $left, right => $right, left_regex => $left_re,
5068                regex => qr/$re/, glob => $glob, depth => $depth }, $class;
5069}
5070
5071sub full_path {
5072        my ($self, $path) = @_;
5073        return (length $self->{left} ? "$self->{left}/" : '') .
5074               $path . (length $self->{right} ? "/$self->{right}" : '');
5075}
5076
5077__END__
5078
5079Data structures:
5080
5081
5082$remotes = { # returned by read_all_remotes()
5083        'svn' => {
5084                # svn-remote.svn.url=https://svn.musicpd.org
5085                url => 'https://svn.musicpd.org',
5086                # svn-remote.svn.fetch=mpd/trunk:trunk
5087                fetch => {
5088                        'mpd/trunk' => 'trunk',
5089                },
5090                # svn-remote.svn.tags=mpd/tags/*:tags/*
5091                tags => {
5092                        path => {
5093                                left => 'mpd/tags',
5094                                right => '',
5095                                regex => qr!mpd/tags/([^/]+)$!,
5096                                glob => 'tags/*',
5097                        },
5098                        ref => {
5099                                left => 'tags',
5100                                right => '',
5101                                regex => qr!tags/([^/]+)$!,
5102                                glob => 'tags/*',
5103                        },
5104                }
5105        }
5106};
5107
5108$log_entry hashref as returned by libsvn_log_entry()
5109{
5110        log => 'whitespace-formatted log entry
5111',                                              # trailing newline is preserved
5112        revision => '8',                        # integer
5113        date => '2004-02-24T17:01:44.108345Z',  # commit date
5114        author => 'committer name'
5115};
5116
5117
5118# this is generated by generate_diff();
5119@mods = array of diff-index line hashes, each element represents one line
5120        of diff-index output
5121
5122diff-index line ($m hash)
5123{
5124        mode_a => first column of diff-index output, no leading ':',
5125        mode_b => second column of diff-index output,
5126        sha1_b => sha1sum of the final blob,
5127        chg => change type [MCRADT],
5128        file_a => original file name of a file (iff chg is 'C' or 'R')
5129        file_b => new/current file name of a file (any chg)
5130}
5131;
5132
5133# retval of read_url_paths{,_all}();
5134$l_map = {
5135        # repository root url
5136        'https://svn.musicpd.org' => {
5137                # repository path               # GIT_SVN_ID
5138                'mpd/trunk'             =>      'trunk',
5139                'mpd/tags/0.11.5'       =>      'tags/0.11.5',
5140        },
5141}
5142
5143Notes:
5144        I don't trust the each() function on unless I created %hash myself
5145        because the internal iterator may not have started at base.