git-svn.perlon commit Merge branch 'bd/diff-strbuf' (3814c07)
   1#!/usr/bin/env perl
   2# Copyright (C) 2006, Eric Wong <normalperson@yhbt.net>
   3# License: GPL v2 or later
   4use warnings;
   5use strict;
   6use vars qw/    $AUTHOR $VERSION
   7                $sha1 $sha1_short $_revision $_repository
   8                $_q $_authors %users/;
   9$AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
  10$VERSION = '@@GIT_VERSION@@';
  11
  12# From which subdir have we been invoked?
  13my $cmd_dir_prefix = eval {
  14        command_oneline([qw/rev-parse --show-prefix/], STDERR => 0)
  15} || '';
  16
  17my $git_dir_user_set = 1 if defined $ENV{GIT_DIR};
  18$ENV{GIT_DIR} ||= '.git';
  19$Git::SVN::default_repo_id = 'svn';
  20$Git::SVN::default_ref_id = $ENV{GIT_SVN_ID} || 'git-svn';
  21$Git::SVN::Ra::_log_window_size = 100;
  22
  23$Git::SVN::Log::TZ = $ENV{TZ};
  24$ENV{TZ} = 'UTC';
  25$| = 1; # unbuffer STDOUT
  26
  27sub fatal (@) { print STDERR "@_\n"; exit 1 }
  28require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
  29require SVN::Ra;
  30require SVN::Delta;
  31if ($SVN::Core::VERSION lt '1.1.0') {
  32        fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)";
  33}
  34push @Git::SVN::Ra::ISA, 'SVN::Ra';
  35push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
  36push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
  37use Carp qw/croak/;
  38use Digest::MD5;
  39use IO::File qw//;
  40use File::Basename qw/dirname basename/;
  41use File::Path qw/mkpath/;
  42use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev/;
  43use IPC::Open3;
  44use Git;
  45
  46BEGIN {
  47        # import functions from Git into our packages, en masse
  48        no strict 'refs';
  49        foreach (qw/command command_oneline command_noisy command_output_pipe
  50                    command_input_pipe command_close_pipe/) {
  51                for my $package ( qw(SVN::Git::Editor SVN::Git::Fetcher
  52                        Git::SVN::Migration Git::SVN::Log Git::SVN),
  53                        __PACKAGE__) {
  54                        *{"${package}::$_"} = \&{"Git::$_"};
  55                }
  56        }
  57}
  58
  59my ($SVN);
  60
  61$sha1 = qr/[a-f\d]{40}/;
  62$sha1_short = qr/[a-f\d]{4,40}/;
  63my ($_stdin, $_help, $_edit,
  64        $_message, $_file,
  65        $_template, $_shared,
  66        $_version, $_fetch_all, $_no_rebase,
  67        $_merge, $_strategy, $_dry_run, $_local,
  68        $_prefix, $_no_checkout, $_url, $_verbose,
  69        $_git_format, $_commit_url);
  70$Git::SVN::_follow_parent = 1;
  71my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
  72                    'config-dir=s' => \$Git::SVN::Ra::config_dir,
  73                    'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache );
  74my %fc_opts = ( 'follow-parent|follow!' => \$Git::SVN::_follow_parent,
  75                'authors-file|A=s' => \$_authors,
  76                'repack:i' => \$Git::SVN::_repack,
  77                'noMetadata' => \$Git::SVN::_no_metadata,
  78                'useSvmProps' => \$Git::SVN::_use_svm_props,
  79                'useSvnsyncProps' => \$Git::SVN::_use_svnsync_props,
  80                'log-window-size=i' => \$Git::SVN::Ra::_log_window_size,
  81                'no-checkout' => \$_no_checkout,
  82                'quiet|q' => \$_q,
  83                'repack-flags|repack-args|repack-opts=s' =>
  84                   \$Git::SVN::_repack_flags,
  85                'use-log-author' => \$Git::SVN::_use_log_author,
  86                'add-author-from' => \$Git::SVN::_add_author_from,
  87                %remote_opts );
  88
  89my ($_trunk, $_tags, $_branches, $_stdlayout);
  90my %icv;
  91my %init_opts = ( 'template=s' => \$_template, 'shared:s' => \$_shared,
  92                  'trunk|T=s' => \$_trunk, 'tags|t=s' => \$_tags,
  93                  'branches|b=s' => \$_branches, 'prefix=s' => \$_prefix,
  94                  'stdlayout|s' => \$_stdlayout,
  95                  'minimize-url|m' => \$Git::SVN::_minimize_url,
  96                  'no-metadata' => sub { $icv{noMetadata} = 1 },
  97                  'use-svm-props' => sub { $icv{useSvmProps} = 1 },
  98                  'use-svnsync-props' => sub { $icv{useSvnsyncProps} = 1 },
  99                  'rewrite-root=s' => sub { $icv{rewriteRoot} = $_[1] },
 100                  %remote_opts );
 101my %cmt_opts = ( 'edit|e' => \$_edit,
 102                'rmdir' => \$SVN::Git::Editor::_rmdir,
 103                'find-copies-harder' => \$SVN::Git::Editor::_find_copies_harder,
 104                'l=i' => \$SVN::Git::Editor::_rename_limit,
 105                'copy-similarity|C=i'=> \$SVN::Git::Editor::_cp_similarity
 106);
 107
 108my %cmd = (
 109        fetch => [ \&cmd_fetch, "Download new revisions from SVN",
 110                        { 'revision|r=s' => \$_revision,
 111                          'fetch-all|all' => \$_fetch_all,
 112                           %fc_opts } ],
 113        clone => [ \&cmd_clone, "Initialize and fetch revisions",
 114                        { 'revision|r=s' => \$_revision,
 115                           %fc_opts, %init_opts } ],
 116        init => [ \&cmd_init, "Initialize a repo for tracking" .
 117                          " (requires URL argument)",
 118                          \%init_opts ],
 119        'multi-init' => [ \&cmd_multi_init,
 120                          "Deprecated alias for ".
 121                          "'$0 init -T<trunk> -b<branches> -t<tags>'",
 122                          \%init_opts ],
 123        dcommit => [ \&cmd_dcommit,
 124                     'Commit several diffs to merge with upstream',
 125                        { 'merge|m|M' => \$_merge,
 126                          'strategy|s=s' => \$_strategy,
 127                          'verbose|v' => \$_verbose,
 128                          'dry-run|n' => \$_dry_run,
 129                          'fetch-all|all' => \$_fetch_all,
 130                          'commit-url=s' => \$_commit_url,
 131                          'revision|r=i' => \$_revision,
 132                          'no-rebase' => \$_no_rebase,
 133                        %cmt_opts, %fc_opts } ],
 134        'set-tree' => [ \&cmd_set_tree,
 135                        "Set an SVN repository to a git tree-ish",
 136                        { 'stdin|' => \$_stdin, %cmt_opts, %fc_opts, } ],
 137        'create-ignore' => [ \&cmd_create_ignore,
 138                             'Create a .gitignore per svn:ignore',
 139                             { 'revision|r=i' => \$_revision
 140                             } ],
 141        'propget' => [ \&cmd_propget,
 142                       'Print the value of a property on a file or directory',
 143                       { 'revision|r=i' => \$_revision } ],
 144        'proplist' => [ \&cmd_proplist,
 145                       'List all properties of a file or directory',
 146                       { 'revision|r=i' => \$_revision } ],
 147        'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
 148                        { 'revision|r=i' => \$_revision
 149                        } ],
 150        'show-externals' => [ \&cmd_show_externals, "Show svn:externals listings",
 151                        { 'revision|r=i' => \$_revision
 152                        } ],
 153        'multi-fetch' => [ \&cmd_multi_fetch,
 154                           "Deprecated alias for $0 fetch --all",
 155                           { 'revision|r=s' => \$_revision, %fc_opts } ],
 156        'migrate' => [ sub { },
 157                       # no-op, we automatically run this anyways,
 158                       'Migrate configuration/metadata/layout from
 159                        previous versions of git-svn',
 160                       { 'minimize' => \$Git::SVN::Migration::_minimize,
 161                         %remote_opts } ],
 162        'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
 163                        { 'limit=i' => \$Git::SVN::Log::limit,
 164                          'revision|r=s' => \$_revision,
 165                          'verbose|v' => \$Git::SVN::Log::verbose,
 166                          'incremental' => \$Git::SVN::Log::incremental,
 167                          'oneline' => \$Git::SVN::Log::oneline,
 168                          'show-commit' => \$Git::SVN::Log::show_commit,
 169                          'non-recursive' => \$Git::SVN::Log::non_recursive,
 170                          'authors-file|A=s' => \$_authors,
 171                          'color' => \$Git::SVN::Log::color,
 172                          'pager=s' => \$Git::SVN::Log::pager
 173                        } ],
 174        'find-rev' => [ \&cmd_find_rev,
 175                        "Translate between SVN revision numbers and tree-ish",
 176                        {} ],
 177        'rebase' => [ \&cmd_rebase, "Fetch and rebase your working directory",
 178                        { 'merge|m|M' => \$_merge,
 179                          'verbose|v' => \$_verbose,
 180                          'strategy|s=s' => \$_strategy,
 181                          'local|l' => \$_local,
 182                          'fetch-all|all' => \$_fetch_all,
 183                          'dry-run|n' => \$_dry_run,
 184                          %fc_opts } ],
 185        'commit-diff' => [ \&cmd_commit_diff,
 186                           'Commit a diff between two trees',
 187                        { 'message|m=s' => \$_message,
 188                          'file|F=s' => \$_file,
 189                          'revision|r=s' => \$_revision,
 190                        %cmt_opts } ],
 191        'info' => [ \&cmd_info,
 192                    "Show info about the latest SVN revision
 193                     on the current branch",
 194                    { 'url' => \$_url, } ],
 195        'blame' => [ \&Git::SVN::Log::cmd_blame,
 196                    "Show what revision and author last modified each line of a file",
 197                    { 'git-format' => \$_git_format } ],
 198);
 199
 200my $cmd;
 201for (my $i = 0; $i < @ARGV; $i++) {
 202        if (defined $cmd{$ARGV[$i]}) {
 203                $cmd = $ARGV[$i];
 204                splice @ARGV, $i, 1;
 205                last;
 206        }
 207};
 208
 209# make sure we're always running at the top-level working directory
 210unless ($cmd && $cmd =~ /(?:clone|init|multi-init)$/) {
 211        unless (-d $ENV{GIT_DIR}) {
 212                if ($git_dir_user_set) {
 213                        die "GIT_DIR=$ENV{GIT_DIR} explicitly set, ",
 214                            "but it is not a directory\n";
 215                }
 216                my $git_dir = delete $ENV{GIT_DIR};
 217                chomp(my $cdup = command_oneline(qw/rev-parse --show-cdup/));
 218                unless (length $cdup) {
 219                        die "Already at toplevel, but $git_dir ",
 220                            "not found '$cdup'\n";
 221                }
 222                chdir $cdup or die "Unable to chdir up to '$cdup'\n";
 223                unless (-d $git_dir) {
 224                        die "$git_dir still not found after going to ",
 225                            "'$cdup'\n";
 226                }
 227                $ENV{GIT_DIR} = $git_dir;
 228        }
 229        $_repository = Git->repository(Repository => $ENV{GIT_DIR});
 230}
 231
 232my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
 233
 234read_repo_config(\%opts);
 235if ($cmd && ($cmd eq 'log' || $cmd eq 'blame')) {
 236        Getopt::Long::Configure('pass_through');
 237}
 238my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
 239                    'minimize-connections' => \$Git::SVN::Migration::_minimize,
 240                    'id|i=s' => \$Git::SVN::default_ref_id,
 241                    'svn-remote|remote|R=s' => sub {
 242                       $Git::SVN::no_reuse_existing = 1;
 243                       $Git::SVN::default_repo_id = $_[1] });
 244exit 1 if (!$rv && $cmd && $cmd ne 'log');
 245
 246usage(0) if $_help;
 247version() if $_version;
 248usage(1) unless defined $cmd;
 249load_authors() if $_authors;
 250
 251unless ($cmd =~ /^(?:clone|init|multi-init|commit-diff)$/) {
 252        Git::SVN::Migration::migration_check();
 253}
 254Git::SVN::init_vars();
 255eval {
 256        Git::SVN::verify_remotes_sanity();
 257        $cmd{$cmd}->[0]->(@ARGV);
 258};
 259fatal $@ if $@;
 260post_fetch_checkout();
 261exit 0;
 262
 263####################### primary functions ######################
 264sub usage {
 265        my $exit = shift || 0;
 266        my $fd = $exit ? \*STDERR : \*STDOUT;
 267        print $fd <<"";
 268git-svn - bidirectional operations between a single Subversion tree and git
 269Usage: git svn <command> [options] [arguments]\n
 270
 271        print $fd "Available commands:\n" unless $cmd;
 272
 273        foreach (sort keys %cmd) {
 274                next if $cmd && $cmd ne $_;
 275                next if /^multi-/; # don't show deprecated commands
 276                print $fd '  ',pack('A17',$_),$cmd{$_}->[1],"\n";
 277                foreach (sort keys %{$cmd{$_}->[2]}) {
 278                        # mixed-case options are for .git/config only
 279                        next if /[A-Z]/ && /^[a-z]+$/i;
 280                        # prints out arguments as they should be passed:
 281                        my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
 282                        print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
 283                                                        "--$_" : "-$_" }
 284                                                split /\|/,$_)," $x\n";
 285                }
 286        }
 287        print $fd <<"";
 288\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
 289arbitrary identifier if you're tracking multiple SVN branches/repositories in
 290one git repository and want to keep them separate.  See git-svn(1) for more
 291information.
 292
 293        exit $exit;
 294}
 295
 296sub version {
 297        print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
 298        exit 0;
 299}
 300
 301sub do_git_init_db {
 302        unless (-d $ENV{GIT_DIR}) {
 303                my @init_db = ('init');
 304                push @init_db, "--template=$_template" if defined $_template;
 305                if (defined $_shared) {
 306                        if ($_shared =~ /[a-z]/) {
 307                                push @init_db, "--shared=$_shared";
 308                        } else {
 309                                push @init_db, "--shared";
 310                        }
 311                }
 312                command_noisy(@init_db);
 313                $_repository = Git->repository(Repository => ".git");
 314        }
 315        my $set;
 316        my $pfx = "svn-remote.$Git::SVN::default_repo_id";
 317        foreach my $i (keys %icv) {
 318                die "'$set' and '$i' cannot both be set\n" if $set;
 319                next unless defined $icv{$i};
 320                command_noisy('config', "$pfx.$i", $icv{$i});
 321                $set = $i;
 322        }
 323}
 324
 325sub init_subdir {
 326        my $repo_path = shift or return;
 327        mkpath([$repo_path]) unless -d $repo_path;
 328        chdir $repo_path or die "Couldn't chdir to $repo_path: $!\n";
 329        $ENV{GIT_DIR} = '.git';
 330        $_repository = Git->repository(Repository => $ENV{GIT_DIR});
 331}
 332
 333sub cmd_clone {
 334        my ($url, $path) = @_;
 335        if (!defined $path &&
 336            (defined $_trunk || defined $_branches || defined $_tags ||
 337             defined $_stdlayout) &&
 338            $url !~ m#^[a-z\+]+://#) {
 339                $path = $url;
 340        }
 341        $path = basename($url) if !defined $path || !length $path;
 342        cmd_init($url, $path);
 343        Git::SVN::fetch_all($Git::SVN::default_repo_id);
 344}
 345
 346sub cmd_init {
 347        if (defined $_stdlayout) {
 348                $_trunk = 'trunk' if (!defined $_trunk);
 349                $_tags = 'tags' if (!defined $_tags);
 350                $_branches = 'branches' if (!defined $_branches);
 351        }
 352        if (defined $_trunk || defined $_branches || defined $_tags) {
 353                return cmd_multi_init(@_);
 354        }
 355        my $url = shift or die "SVN repository location required ",
 356                               "as a command-line argument\n";
 357        init_subdir(@_);
 358        do_git_init_db();
 359
 360        Git::SVN->init($url);
 361}
 362
 363sub cmd_fetch {
 364        if (grep /^\d+=./, @_) {
 365                die "'<rev>=<commit>' fetch arguments are ",
 366                    "no longer supported.\n";
 367        }
 368        my ($remote) = @_;
 369        if (@_ > 1) {
 370                die "Usage: $0 fetch [--all] [svn-remote]\n";
 371        }
 372        $remote ||= $Git::SVN::default_repo_id;
 373        if ($_fetch_all) {
 374                cmd_multi_fetch();
 375        } else {
 376                Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
 377        }
 378}
 379
 380sub cmd_set_tree {
 381        my (@commits) = @_;
 382        if ($_stdin || !@commits) {
 383                print "Reading from stdin...\n";
 384                @commits = ();
 385                while (<STDIN>) {
 386                        if (/\b($sha1_short)\b/o) {
 387                                unshift @commits, $1;
 388                        }
 389                }
 390        }
 391        my @revs;
 392        foreach my $c (@commits) {
 393                my @tmp = command('rev-parse',$c);
 394                if (scalar @tmp == 1) {
 395                        push @revs, $tmp[0];
 396                } elsif (scalar @tmp > 1) {
 397                        push @revs, reverse(command('rev-list',@tmp));
 398                } else {
 399                        fatal "Failed to rev-parse $c";
 400                }
 401        }
 402        my $gs = Git::SVN->new;
 403        my ($r_last, $cmt_last) = $gs->last_rev_commit;
 404        $gs->fetch;
 405        if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
 406                fatal "There are new revisions that were fetched ",
 407                      "and need to be merged (or acknowledged) ",
 408                      "before committing.\nlast rev: $r_last\n",
 409                      " current: $gs->{last_rev}";
 410        }
 411        $gs->set_tree($_) foreach @revs;
 412        print "Done committing ",scalar @revs," revisions to SVN\n";
 413        unlink $gs->{index};
 414}
 415
 416sub cmd_dcommit {
 417        my $head = shift;
 418        git_cmd_try { command_oneline(qw/diff-index --quiet HEAD/) }
 419                'Cannot dcommit with a dirty index.  Commit your changes first, '
 420                . "or stash them with `git stash'.\n";
 421        $head ||= 'HEAD';
 422        my @refs;
 423        my ($url, $rev, $uuid, $gs) = working_head_info($head, \@refs);
 424        $url = $_commit_url if defined $_commit_url;
 425        my $last_rev = $_revision if defined $_revision;
 426        if ($url) {
 427                print "Committing to $url ...\n";
 428        }
 429        unless ($gs) {
 430                die "Unable to determine upstream SVN information from ",
 431                    "$head history.\nPerhaps the repository is empty.";
 432        }
 433        my ($linear_refs, $parents) = linearize_history($gs, \@refs);
 434        if ($_no_rebase && scalar(@$linear_refs) > 1) {
 435                warn "Attempting to commit more than one change while ",
 436                     "--no-rebase is enabled.\n",
 437                     "If these changes depend on each other, re-running ",
 438                     "without --no-rebase may be required."
 439        }
 440        while (1) {
 441                my $d = shift @$linear_refs or last;
 442                unless (defined $last_rev) {
 443                        (undef, $last_rev, undef) = cmt_metadata("$d~1");
 444                        unless (defined $last_rev) {
 445                                fatal "Unable to extract revision information ",
 446                                      "from commit $d~1";
 447                        }
 448                }
 449                if ($_dry_run) {
 450                        print "diff-tree $d~1 $d\n";
 451                } else {
 452                        my $cmt_rev;
 453                        my %ed_opts = ( r => $last_rev,
 454                                        log => get_commit_entry($d)->{log},
 455                                        ra => Git::SVN::Ra->new($url),
 456                                        config => SVN::Core::config_get_config(
 457                                                $Git::SVN::Ra::config_dir
 458                                        ),
 459                                        tree_a => "$d~1",
 460                                        tree_b => $d,
 461                                        editor_cb => sub {
 462                                               print "Committed r$_[0]\n";
 463                                               $cmt_rev = $_[0];
 464                                        },
 465                                        svn_path => '');
 466                        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
 467                                print "No changes\n$d~1 == $d\n";
 468                        } elsif ($parents->{$d} && @{$parents->{$d}}) {
 469                                $gs->{inject_parents_dcommit}->{$cmt_rev} =
 470                                                               $parents->{$d};
 471                        }
 472                        $_fetch_all ? $gs->fetch_all : $gs->fetch;
 473                        $last_rev = $cmt_rev;
 474                        next if $_no_rebase;
 475
 476                        # we always want to rebase against the current HEAD,
 477                        # not any head that was passed to us
 478                        my @diff = command('diff-tree', $d,
 479                                           $gs->refname, '--');
 480                        my @finish;
 481                        if (@diff) {
 482                                @finish = rebase_cmd();
 483                                print STDERR "W: $d and ", $gs->refname,
 484                                             " differ, using @finish:\n",
 485                                             join("\n", @diff), "\n";
 486                        } else {
 487                                print "No changes between current HEAD and ",
 488                                      $gs->refname,
 489                                      "\nResetting to the latest ",
 490                                      $gs->refname, "\n";
 491                                @finish = qw/reset --mixed/;
 492                        }
 493                        command_noisy(@finish, $gs->refname);
 494                        if (@diff) {
 495                                @refs = ();
 496                                my ($url_, $rev_, $uuid_, $gs_) =
 497                                              working_head_info($head, \@refs);
 498                                my ($linear_refs_, $parents_) =
 499                                              linearize_history($gs_, \@refs);
 500                                if (scalar(@$linear_refs) !=
 501                                    scalar(@$linear_refs_)) {
 502                                        fatal "# of revisions changed ",
 503                                          "\nbefore:\n",
 504                                          join("\n", @$linear_refs),
 505                                          "\n\nafter:\n",
 506                                          join("\n", @$linear_refs_), "\n",
 507                                          'If you are attempting to commit ',
 508                                          "merges, try running:\n\t",
 509                                          'git rebase --interactive',
 510                                          '--preserve-merges ',
 511                                          $gs->refname,
 512                                          "\nBefore dcommitting";
 513                                }
 514                                if ($url_ ne $url) {
 515                                        fatal "URL mismatch after rebase: ",
 516                                              "$url_ != $url";
 517                                }
 518                                if ($uuid_ ne $uuid) {
 519                                        fatal "uuid mismatch after rebase: ",
 520                                              "$uuid_ != $uuid";
 521                                }
 522                                # remap parents
 523                                my (%p, @l, $i);
 524                                for ($i = 0; $i < scalar @$linear_refs; $i++) {
 525                                        my $new = $linear_refs_->[$i] or next;
 526                                        $p{$new} =
 527                                                $parents->{$linear_refs->[$i]};
 528                                        push @l, $new;
 529                                }
 530                                $parents = \%p;
 531                                $linear_refs = \@l;
 532                        }
 533                }
 534        }
 535        unlink $gs->{index};
 536}
 537
 538sub cmd_find_rev {
 539        my $revision_or_hash = shift or die "SVN or git revision required ",
 540                                            "as a command-line argument\n";
 541        my $result;
 542        if ($revision_or_hash =~ /^r\d+$/) {
 543                my $head = shift;
 544                $head ||= 'HEAD';
 545                my @refs;
 546                my (undef, undef, $uuid, $gs) = working_head_info($head, \@refs);
 547                unless ($gs) {
 548                        die "Unable to determine upstream SVN information from ",
 549                            "$head history\n";
 550                }
 551                my $desired_revision = substr($revision_or_hash, 1);
 552                $result = $gs->rev_map_get($desired_revision, $uuid);
 553        } else {
 554                my (undef, $rev, undef) = cmt_metadata($revision_or_hash);
 555                $result = $rev;
 556        }
 557        print "$result\n" if $result;
 558}
 559
 560sub cmd_rebase {
 561        command_noisy(qw/update-index --refresh/);
 562        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 563        unless ($gs) {
 564                die "Unable to determine upstream SVN information from ",
 565                    "working tree history\n";
 566        }
 567        if ($_dry_run) {
 568                print "Remote Branch: " . $gs->refname . "\n";
 569                print "SVN URL: " . $url . "\n";
 570                return;
 571        }
 572        if (command(qw/diff-index HEAD --/)) {
 573                print STDERR "Cannot rebase with uncommited changes:\n";
 574                command_noisy('status');
 575                exit 1;
 576        }
 577        unless ($_local) {
 578                # rebase will checkout for us, so no need to do it explicitly
 579                $_no_checkout = 'true';
 580                $_fetch_all ? $gs->fetch_all : $gs->fetch;
 581        }
 582        command_noisy(rebase_cmd(), $gs->refname);
 583}
 584
 585sub cmd_show_ignore {
 586        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 587        $gs ||= Git::SVN->new;
 588        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 589        $gs->prop_walk($gs->{path}, $r, sub {
 590                my ($gs, $path, $props) = @_;
 591                print STDOUT "\n# $path\n";
 592                my $s = $props->{'svn:ignore'} or return;
 593                $s =~ s/[\r\n]+/\n/g;
 594                chomp $s;
 595                $s =~ s#^#$path#gm;
 596                print STDOUT "$s\n";
 597        });
 598}
 599
 600sub cmd_show_externals {
 601        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 602        $gs ||= Git::SVN->new;
 603        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 604        $gs->prop_walk($gs->{path}, $r, sub {
 605                my ($gs, $path, $props) = @_;
 606                print STDOUT "\n# $path\n";
 607                my $s = $props->{'svn:externals'} or return;
 608                $s =~ s/[\r\n]+/\n/g;
 609                chomp $s;
 610                $s =~ s#^#$path#gm;
 611                print STDOUT "$s\n";
 612        });
 613}
 614
 615sub cmd_create_ignore {
 616        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 617        $gs ||= Git::SVN->new;
 618        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 619        $gs->prop_walk($gs->{path}, $r, sub {
 620                my ($gs, $path, $props) = @_;
 621                # $path is of the form /path/to/dir/
 622                my $ignore = '.' . $path . '.gitignore';
 623                my $s = $props->{'svn:ignore'} or return;
 624                open(GITIGNORE, '>', $ignore)
 625                  or fatal("Failed to open `$ignore' for writing: $!");
 626                $s =~ s/[\r\n]+/\n/g;
 627                chomp $s;
 628                # Prefix all patterns so that the ignore doesn't apply
 629                # to sub-directories.
 630                $s =~ s#^#/#gm;
 631                print GITIGNORE "$s\n";
 632                close(GITIGNORE)
 633                  or fatal("Failed to close `$ignore': $!");
 634                command_noisy('add', '-f', $ignore);
 635        });
 636}
 637
 638sub canonicalize_path {
 639        my ($path) = @_;
 640        my $dot_slash_added = 0;
 641        if (substr($path, 0, 1) ne "/") {
 642                $path = "./" . $path;
 643                $dot_slash_added = 1;
 644        }
 645        # File::Spec->canonpath doesn't collapse x/../y into y (for a
 646        # good reason), so let's do this manually.
 647        $path =~ s#/+#/#g;
 648        $path =~ s#/\.(?:/|$)#/#g;
 649        $path =~ s#/[^/]+/\.\.##g;
 650        $path =~ s#/$##g;
 651        $path =~ s#^\./## if $dot_slash_added;
 652        $path =~ s#^/##;
 653        $path =~ s#^\.$##;
 654        return $path;
 655}
 656
 657# get_svnprops(PATH)
 658# ------------------
 659# Helper for cmd_propget and cmd_proplist below.
 660sub get_svnprops {
 661        my $path = shift;
 662        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 663        $gs ||= Git::SVN->new;
 664
 665        # prefix THE PATH by the sub-directory from which the user
 666        # invoked us.
 667        $path = $cmd_dir_prefix . $path;
 668        fatal("No such file or directory: $path") unless -e $path;
 669        my $is_dir = -d $path ? 1 : 0;
 670        $path = $gs->{path} . '/' . $path;
 671
 672        # canonicalize the path (otherwise libsvn will abort or fail to
 673        # find the file)
 674        $path = canonicalize_path($path);
 675
 676        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 677        my $props;
 678        if ($is_dir) {
 679                (undef, undef, $props) = $gs->ra->get_dir($path, $r);
 680        }
 681        else {
 682                (undef, $props) = $gs->ra->get_file($path, $r, undef);
 683        }
 684        return $props;
 685}
 686
 687# cmd_propget (PROP, PATH)
 688# ------------------------
 689# Print the SVN property PROP for PATH.
 690sub cmd_propget {
 691        my ($prop, $path) = @_;
 692        $path = '.' if not defined $path;
 693        usage(1) if not defined $prop;
 694        my $props = get_svnprops($path);
 695        if (not defined $props->{$prop}) {
 696                fatal("`$path' does not have a `$prop' SVN property.");
 697        }
 698        print $props->{$prop} . "\n";
 699}
 700
 701# cmd_proplist (PATH)
 702# -------------------
 703# Print the list of SVN properties for PATH.
 704sub cmd_proplist {
 705        my $path = shift;
 706        $path = '.' if not defined $path;
 707        my $props = get_svnprops($path);
 708        print "Properties on '$path':\n";
 709        foreach (sort keys %{$props}) {
 710                print "  $_\n";
 711        }
 712}
 713
 714sub cmd_multi_init {
 715        my $url = shift;
 716        unless (defined $_trunk || defined $_branches || defined $_tags) {
 717                usage(1);
 718        }
 719
 720        # there are currently some bugs that prevent multi-init/multi-fetch
 721        # setups from working well without this.
 722        $Git::SVN::_minimize_url = 1;
 723
 724        $_prefix = '' unless defined $_prefix;
 725        if (defined $url) {
 726                $url =~ s#/+$##;
 727                init_subdir(@_);
 728        }
 729        do_git_init_db();
 730        if (defined $_trunk) {
 731                my $trunk_ref = $_prefix . 'trunk';
 732                # try both old-style and new-style lookups:
 733                my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
 734                unless ($gs_trunk) {
 735                        my ($trunk_url, $trunk_path) =
 736                                              complete_svn_url($url, $_trunk);
 737                        $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
 738                                                   undef, $trunk_ref);
 739                }
 740        }
 741        return unless defined $_branches || defined $_tags;
 742        my $ra = $url ? Git::SVN::Ra->new($url) : undef;
 743        complete_url_ls_init($ra, $_branches, '--branches/-b', $_prefix);
 744        complete_url_ls_init($ra, $_tags, '--tags/-t', $_prefix . 'tags/');
 745}
 746
 747sub cmd_multi_fetch {
 748        my $remotes = Git::SVN::read_all_remotes();
 749        foreach my $repo_id (sort keys %$remotes) {
 750                if ($remotes->{$repo_id}->{url}) {
 751                        Git::SVN::fetch_all($repo_id, $remotes);
 752                }
 753        }
 754}
 755
 756# this command is special because it requires no metadata
 757sub cmd_commit_diff {
 758        my ($ta, $tb, $url) = @_;
 759        my $usage = "Usage: $0 commit-diff -r<revision> ".
 760                    "<tree-ish> <tree-ish> [<URL>]";
 761        fatal($usage) if (!defined $ta || !defined $tb);
 762        my $svn_path = '';
 763        if (!defined $url) {
 764                my $gs = eval { Git::SVN->new };
 765                if (!$gs) {
 766                        fatal("Needed URL or usable git-svn --id in ",
 767                              "the command-line\n", $usage);
 768                }
 769                $url = $gs->{url};
 770                $svn_path = $gs->{path};
 771        }
 772        unless (defined $_revision) {
 773                fatal("-r|--revision is a required argument\n", $usage);
 774        }
 775        if (defined $_message && defined $_file) {
 776                fatal("Both --message/-m and --file/-F specified ",
 777                      "for the commit message.\n",
 778                      "I have no idea what you mean");
 779        }
 780        if (defined $_file) {
 781                $_message = file_to_s($_file);
 782        } else {
 783                $_message ||= get_commit_entry($tb)->{log};
 784        }
 785        my $ra ||= Git::SVN::Ra->new($url);
 786        my $r = $_revision;
 787        if ($r eq 'HEAD') {
 788                $r = $ra->get_latest_revnum;
 789        } elsif ($r !~ /^\d+$/) {
 790                die "revision argument: $r not understood by git-svn\n";
 791        }
 792        my %ed_opts = ( r => $r,
 793                        log => $_message,
 794                        ra => $ra,
 795                        tree_a => $ta,
 796                        tree_b => $tb,
 797                        editor_cb => sub { print "Committed r$_[0]\n" },
 798                        svn_path => $svn_path );
 799        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
 800                print "No changes\n$ta == $tb\n";
 801        }
 802}
 803
 804sub cmd_info {
 805        my $path = canonicalize_path(defined($_[0]) ? $_[0] : ".");
 806        if (exists $_[1]) {
 807                die "Too many arguments specified\n";
 808        }
 809
 810        my ($file_type, $diff_status) = find_file_type_and_diff_status($path);
 811
 812        if (!$file_type && !$diff_status) {
 813                print STDERR "$path:  (Not a versioned resource)\n\n";
 814                return;
 815        }
 816
 817        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 818        unless ($gs) {
 819                die "Unable to determine upstream SVN information from ",
 820                    "working tree history\n";
 821        }
 822
 823        # canonicalize_path() will return "" to make libsvn 1.5.x happy,
 824        $path = "." if $path eq "";
 825
 826        my $full_url = $url . ($path eq "." ? "" : "/$path");
 827
 828        if ($_url) {
 829                print $full_url, "\n";
 830                return;
 831        }
 832
 833        my $result = "Path: $path\n";
 834        $result .= "Name: " . basename($path) . "\n" if $file_type ne "dir";
 835        $result .= "URL: " . $full_url . "\n";
 836
 837        eval {
 838                my $repos_root = $gs->repos_root;
 839                Git::SVN::remove_username($repos_root);
 840                $result .= "Repository Root: $repos_root\n";
 841        };
 842        if ($@) {
 843                $result .= "Repository Root: (offline)\n";
 844        }
 845        $result .= "Repository UUID: $uuid\n" unless $diff_status eq "A";
 846        $result .= "Revision: " . ($diff_status eq "A" ? 0 : $rev) . "\n";
 847
 848        $result .= "Node Kind: " .
 849                   ($file_type eq "dir" ? "directory" : "file") . "\n";
 850
 851        my $schedule = $diff_status eq "A"
 852                       ? "add"
 853                       : ($diff_status eq "D" ? "delete" : "normal");
 854        $result .= "Schedule: $schedule\n";
 855
 856        if ($diff_status eq "A") {
 857                print $result, "\n";
 858                return;
 859        }
 860
 861        my ($lc_author, $lc_rev, $lc_date_utc);
 862        my @args = Git::SVN::Log::git_svn_log_cmd($rev, $rev, "--", $path);
 863        my $log = command_output_pipe(@args);
 864        my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
 865        while (<$log>) {
 866                if (/^${esc_color}author (.+) <[^>]+> (\d+) ([\-\+]?\d+)$/o) {
 867                        $lc_author = $1;
 868                        $lc_date_utc = Git::SVN::Log::parse_git_date($2, $3);
 869                } elsif (/^${esc_color}    (git-svn-id:.+)$/o) {
 870                        (undef, $lc_rev, undef) = ::extract_metadata($1);
 871                }
 872        }
 873        close $log;
 874
 875        Git::SVN::Log::set_local_timezone();
 876
 877        $result .= "Last Changed Author: $lc_author\n";
 878        $result .= "Last Changed Rev: $lc_rev\n";
 879        $result .= "Last Changed Date: " .
 880                   Git::SVN::Log::format_svn_date($lc_date_utc) . "\n";
 881
 882        if ($file_type ne "dir") {
 883                my $text_last_updated_date =
 884                    ($diff_status eq "D" ? $lc_date_utc : (stat $path)[9]);
 885                $result .=
 886                    "Text Last Updated: " .
 887                    Git::SVN::Log::format_svn_date($text_last_updated_date) .
 888                    "\n";
 889                my $checksum;
 890                if ($diff_status eq "D") {
 891                        my ($fh, $ctx) =
 892                            command_output_pipe(qw(cat-file blob), "HEAD:$path");
 893                        if ($file_type eq "link") {
 894                                my $file_name = <$fh>;
 895                                $checksum = md5sum("link $file_name");
 896                        } else {
 897                                $checksum = md5sum($fh);
 898                        }
 899                        command_close_pipe($fh, $ctx);
 900                } elsif ($file_type eq "link") {
 901                        my $file_name =
 902                            command(qw(cat-file blob), "HEAD:$path");
 903                        $checksum =
 904                            md5sum("link " . $file_name);
 905                } else {
 906                        open FILE, "<", $path or die $!;
 907                        $checksum = md5sum(\*FILE);
 908                        close FILE or die $!;
 909                }
 910                $result .= "Checksum: " . $checksum . "\n";
 911        }
 912
 913        print $result, "\n";
 914}
 915
 916########################### utility functions #########################
 917
 918sub rebase_cmd {
 919        my @cmd = qw/rebase/;
 920        push @cmd, '-v' if $_verbose;
 921        push @cmd, qw/--merge/ if $_merge;
 922        push @cmd, "--strategy=$_strategy" if $_strategy;
 923        @cmd;
 924}
 925
 926sub post_fetch_checkout {
 927        return if $_no_checkout;
 928        my $gs = $Git::SVN::_head or return;
 929        return if verify_ref('refs/heads/master^0');
 930
 931        my $valid_head = verify_ref('HEAD^0');
 932        command_noisy(qw(update-ref refs/heads/master), $gs->refname);
 933        return if ($valid_head || !verify_ref('HEAD^0'));
 934
 935        return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
 936        my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
 937        return if -f $index;
 938
 939        return if command_oneline(qw/rev-parse --is-inside-work-tree/) eq 'false';
 940        return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
 941        command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
 942        print STDERR "Checked out HEAD:\n  ",
 943                     $gs->full_url, " r", $gs->last_rev, "\n";
 944}
 945
 946sub complete_svn_url {
 947        my ($url, $path) = @_;
 948        $path =~ s#/+$##;
 949        if ($path !~ m#^[a-z\+]+://#) {
 950                if (!defined $url || $url !~ m#^[a-z\+]+://#) {
 951                        fatal("E: '$path' is not a complete URL ",
 952                              "and a separate URL is not specified");
 953                }
 954                return ($url, $path);
 955        }
 956        return ($path, '');
 957}
 958
 959sub complete_url_ls_init {
 960        my ($ra, $repo_path, $switch, $pfx) = @_;
 961        unless ($repo_path) {
 962                print STDERR "W: $switch not specified\n";
 963                return;
 964        }
 965        $repo_path =~ s#/+$##;
 966        if ($repo_path =~ m#^[a-z\+]+://#) {
 967                $ra = Git::SVN::Ra->new($repo_path);
 968                $repo_path = '';
 969        } else {
 970                $repo_path =~ s#^/+##;
 971                unless ($ra) {
 972                        fatal("E: '$repo_path' is not a complete URL ",
 973                              "and a separate URL is not specified");
 974                }
 975        }
 976        my $url = $ra->{url};
 977        my $gs = Git::SVN->init($url, undef, undef, undef, 1);
 978        my $k = "svn-remote.$gs->{repo_id}.url";
 979        my $orig_url = eval { command_oneline(qw/config --get/, $k) };
 980        if ($orig_url && ($orig_url ne $gs->{url})) {
 981                die "$k already set: $orig_url\n",
 982                    "wanted to set to: $gs->{url}\n";
 983        }
 984        command_oneline('config', $k, $gs->{url}) unless $orig_url;
 985        my $remote_path = "$ra->{svn_path}/$repo_path";
 986        $remote_path =~ s#/+#/#g;
 987        $remote_path =~ s#^/##g;
 988        $remote_path .= "/*" if $remote_path !~ /\*/;
 989        my ($n) = ($switch =~ /^--(\w+)/);
 990        if (length $pfx && $pfx !~ m#/$#) {
 991                die "--prefix='$pfx' must have a trailing slash '/'\n";
 992        }
 993        command_noisy('config',
 994                      "svn-remote.$gs->{repo_id}.$n",
 995                      "$remote_path:refs/remotes/$pfx*" .
 996                        ('/*' x (($remote_path =~ tr/*/*/) - 1)) );
 997}
 998
 999sub verify_ref {
1000        my ($ref) = @_;
1001        eval { command_oneline([ 'rev-parse', '--verify', $ref ],
1002                               { STDERR => 0 }); };
1003}
1004
1005sub get_tree_from_treeish {
1006        my ($treeish) = @_;
1007        # $treeish can be a symbolic ref, too:
1008        my $type = command_oneline(qw/cat-file -t/, $treeish);
1009        my $expected;
1010        while ($type eq 'tag') {
1011                ($treeish, $type) = command(qw/cat-file tag/, $treeish);
1012        }
1013        if ($type eq 'commit') {
1014                $expected = (grep /^tree /, command(qw/cat-file commit/,
1015                                                    $treeish))[0];
1016                ($expected) = ($expected =~ /^tree ($sha1)$/o);
1017                die "Unable to get tree from $treeish\n" unless $expected;
1018        } elsif ($type eq 'tree') {
1019                $expected = $treeish;
1020        } else {
1021                die "$treeish is a $type, expected tree, tag or commit\n";
1022        }
1023        return $expected;
1024}
1025
1026sub get_commit_entry {
1027        my ($treeish) = shift;
1028        my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
1029        my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
1030        my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
1031        open my $log_fh, '>', $commit_editmsg or croak $!;
1032
1033        my $type = command_oneline(qw/cat-file -t/, $treeish);
1034        if ($type eq 'commit' || $type eq 'tag') {
1035                my ($msg_fh, $ctx) = command_output_pipe('cat-file',
1036                                                         $type, $treeish);
1037                my $in_msg = 0;
1038                my $author;
1039                my $saw_from = 0;
1040                my $msgbuf = "";
1041                while (<$msg_fh>) {
1042                        if (!$in_msg) {
1043                                $in_msg = 1 if (/^\s*$/);
1044                                $author = $1 if (/^author (.*>)/);
1045                        } elsif (/^git-svn-id: /) {
1046                                # skip this for now, we regenerate the
1047                                # correct one on re-fetch anyways
1048                                # TODO: set *:merge properties or like...
1049                        } else {
1050                                if (/^From:/ || /^Signed-off-by:/) {
1051                                        $saw_from = 1;
1052                                }
1053                                $msgbuf .= $_;
1054                        }
1055                }
1056                $msgbuf =~ s/\s+$//s;
1057                if ($Git::SVN::_add_author_from && defined($author)
1058                    && !$saw_from) {
1059                        $msgbuf .= "\n\nFrom: $author";
1060                }
1061                print $log_fh $msgbuf or croak $!;
1062                command_close_pipe($msg_fh, $ctx);
1063        }
1064        close $log_fh or croak $!;
1065
1066        if ($_edit || ($type eq 'tree')) {
1067                my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
1068                # TODO: strip out spaces, comments, like git-commit.sh
1069                system($editor, $commit_editmsg);
1070        }
1071        rename $commit_editmsg, $commit_msg or croak $!;
1072        open $log_fh, '<', $commit_msg or croak $!;
1073        { local $/; chomp($log_entry{log} = <$log_fh>); }
1074        close $log_fh or croak $!;
1075        unlink $commit_msg;
1076        \%log_entry;
1077}
1078
1079sub s_to_file {
1080        my ($str, $file, $mode) = @_;
1081        open my $fd,'>',$file or croak $!;
1082        print $fd $str,"\n" or croak $!;
1083        close $fd or croak $!;
1084        chmod ($mode &~ umask, $file) if (defined $mode);
1085}
1086
1087sub file_to_s {
1088        my $file = shift;
1089        open my $fd,'<',$file or croak "$!: file: $file\n";
1090        local $/;
1091        my $ret = <$fd>;
1092        close $fd or croak $!;
1093        $ret =~ s/\s*$//s;
1094        return $ret;
1095}
1096
1097# '<svn username> = real-name <email address>' mapping based on git-svnimport:
1098sub load_authors {
1099        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1100        my $log = $cmd eq 'log';
1101        while (<$authors>) {
1102                chomp;
1103                next unless /^(.+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
1104                my ($user, $name, $email) = ($1, $2, $3);
1105                if ($log) {
1106                        $Git::SVN::Log::rusers{"$name <$email>"} = $user;
1107                } else {
1108                        $users{$user} = [$name, $email];
1109                }
1110        }
1111        close $authors or croak $!;
1112}
1113
1114# convert GetOpt::Long specs for use by git-config
1115sub read_repo_config {
1116        return unless -d $ENV{GIT_DIR};
1117        my $opts = shift;
1118        my @config_only;
1119        foreach my $o (keys %$opts) {
1120                # if we have mixedCase and a long option-only, then
1121                # it's a config-only variable that we don't need for
1122                # the command-line.
1123                push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
1124                my $v = $opts->{$o};
1125                my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
1126                $key =~ s/-//g;
1127                my $arg = 'git-config';
1128                $arg .= ' --int' if ($o =~ /[:=]i$/);
1129                $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
1130                if (ref $v eq 'ARRAY') {
1131                        chomp(my @tmp = `$arg --get-all svn.$key`);
1132                        @$v = @tmp if @tmp;
1133                } else {
1134                        chomp(my $tmp = `$arg --get svn.$key`);
1135                        if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
1136                                $$v = $tmp;
1137                        }
1138                }
1139        }
1140        delete @$opts{@config_only} if @config_only;
1141}
1142
1143sub extract_metadata {
1144        my $id = shift or return (undef, undef, undef);
1145        my ($url, $rev, $uuid) = ($id =~ /^\s*git-svn-id:\s+(.*)\@(\d+)
1146                                                        \s([a-f\d\-]+)$/x);
1147        if (!defined $rev || !$uuid || !$url) {
1148                # some of the original repositories I made had
1149                # identifiers like this:
1150                ($rev, $uuid) = ($id =~/^\s*git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
1151        }
1152        return ($url, $rev, $uuid);
1153}
1154
1155sub cmt_metadata {
1156        return extract_metadata((grep(/^git-svn-id: /,
1157                command(qw/cat-file commit/, shift)))[-1]);
1158}
1159
1160sub working_head_info {
1161        my ($head, $refs) = @_;
1162        my @args = ('log', '--no-color', '--first-parent', '--pretty=medium');
1163        my ($fh, $ctx) = command_output_pipe(@args, $head);
1164        my $hash;
1165        my %max;
1166        while (<$fh>) {
1167                if ( m{^commit ($::sha1)$} ) {
1168                        unshift @$refs, $hash if $hash and $refs;
1169                        $hash = $1;
1170                        next;
1171                }
1172                next unless s{^\s*(git-svn-id:)}{$1};
1173                my ($url, $rev, $uuid) = extract_metadata($_);
1174                if (defined $url && defined $rev) {
1175                        next if $max{$url} and $max{$url} < $rev;
1176                        if (my $gs = Git::SVN->find_by_url($url)) {
1177                                my $c = $gs->rev_map_get($rev, $uuid);
1178                                if ($c && $c eq $hash) {
1179                                        close $fh; # break the pipe
1180                                        return ($url, $rev, $uuid, $gs);
1181                                } else {
1182                                        $max{$url} ||= $gs->rev_map_max;
1183                                }
1184                        }
1185                }
1186        }
1187        command_close_pipe($fh, $ctx);
1188        (undef, undef, undef, undef);
1189}
1190
1191sub read_commit_parents {
1192        my ($parents, $c) = @_;
1193        chomp(my $p = command_oneline(qw/rev-list --parents -1/, $c));
1194        $p =~ s/^($c)\s*// or die "rev-list --parents -1 $c failed!\n";
1195        @{$parents->{$c}} = split(/ /, $p);
1196}
1197
1198sub linearize_history {
1199        my ($gs, $refs) = @_;
1200        my %parents;
1201        foreach my $c (@$refs) {
1202                read_commit_parents(\%parents, $c);
1203        }
1204
1205        my @linear_refs;
1206        my %skip = ();
1207        my $last_svn_commit = $gs->last_commit;
1208        foreach my $c (reverse @$refs) {
1209                next if $c eq $last_svn_commit;
1210                last if $skip{$c};
1211
1212                unshift @linear_refs, $c;
1213                $skip{$c} = 1;
1214
1215                # we only want the first parent to diff against for linear
1216                # history, we save the rest to inject when we finalize the
1217                # svn commit
1218                my $fp_a = verify_ref("$c~1");
1219                my $fp_b = shift @{$parents{$c}} if $parents{$c};
1220                if (!$fp_a || !$fp_b) {
1221                        die "Commit $c\n",
1222                            "has no parent commit, and therefore ",
1223                            "nothing to diff against.\n",
1224                            "You should be working from a repository ",
1225                            "originally created by git-svn\n";
1226                }
1227                if ($fp_a ne $fp_b) {
1228                        die "$c~1 = $fp_a, however parsing commit $c ",
1229                            "revealed that:\n$c~1 = $fp_b\nBUG!\n";
1230                }
1231
1232                foreach my $p (@{$parents{$c}}) {
1233                        $skip{$p} = 1;
1234                }
1235        }
1236        (\@linear_refs, \%parents);
1237}
1238
1239sub find_file_type_and_diff_status {
1240        my ($path) = @_;
1241        return ('dir', '') if $path eq '';
1242
1243        my $diff_output =
1244            command_oneline(qw(diff --cached --name-status --), $path) || "";
1245        my $diff_status = (split(' ', $diff_output))[0] || "";
1246
1247        my $ls_tree = command_oneline(qw(ls-tree HEAD), $path) || "";
1248
1249        return (undef, undef) if !$diff_status && !$ls_tree;
1250
1251        if ($diff_status eq "A") {
1252                return ("link", $diff_status) if -l $path;
1253                return ("dir", $diff_status) if -d $path;
1254                return ("file", $diff_status);
1255        }
1256
1257        my $mode = (split(' ', $ls_tree))[0] || "";
1258
1259        return ("link", $diff_status) if $mode eq "120000";
1260        return ("dir", $diff_status) if $mode eq "040000";
1261        return ("file", $diff_status);
1262}
1263
1264sub md5sum {
1265        my $arg = shift;
1266        my $ref = ref $arg;
1267        my $md5 = Digest::MD5->new();
1268        if ($ref eq 'GLOB' || $ref eq 'IO::File' || $ref eq 'File::Temp') {
1269                $md5->addfile($arg) or croak $!;
1270        } elsif ($ref eq 'SCALAR') {
1271                $md5->add($$arg) or croak $!;
1272        } elsif (!$ref) {
1273                $md5->add($arg) or croak $!;
1274        } else {
1275                ::fatal "Can't provide MD5 hash for unknown ref type: '", $ref, "'";
1276        }
1277        return $md5->hexdigest();
1278}
1279
1280package Git::SVN;
1281use strict;
1282use warnings;
1283use Fcntl qw/:DEFAULT :seek/;
1284use constant rev_map_fmt => 'NH40';
1285use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
1286            $_repack $_repack_flags $_use_svm_props $_head
1287            $_use_svnsync_props $no_reuse_existing $_minimize_url
1288            $_use_log_author $_add_author_from/;
1289use Carp qw/croak/;
1290use File::Path qw/mkpath/;
1291use File::Copy qw/copy/;
1292use IPC::Open3;
1293
1294my ($_gc_nr, $_gc_period);
1295
1296# properties that we do not log:
1297my %SKIP_PROP;
1298BEGIN {
1299        %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
1300                                        svn:special svn:executable
1301                                        svn:entry:committed-rev
1302                                        svn:entry:last-author
1303                                        svn:entry:uuid
1304                                        svn:entry:committed-date/;
1305
1306        # some options are read globally, but can be overridden locally
1307        # per [svn-remote "..."] section.  Command-line options will *NOT*
1308        # override options set in an [svn-remote "..."] section
1309        no strict 'refs';
1310        for my $option (qw/follow_parent no_metadata use_svm_props
1311                           use_svnsync_props/) {
1312                my $key = $option;
1313                $key =~ tr/_//d;
1314                my $prop = "-$option";
1315                *$option = sub {
1316                        my ($self) = @_;
1317                        return $self->{$prop} if exists $self->{$prop};
1318                        my $k = "svn-remote.$self->{repo_id}.$key";
1319                        eval { command_oneline(qw/config --get/, $k) };
1320                        if ($@) {
1321                                $self->{$prop} = ${"Git::SVN::_$option"};
1322                        } else {
1323                                my $v = command_oneline(qw/config --bool/,$k);
1324                                $self->{$prop} = $v eq 'false' ? 0 : 1;
1325                        }
1326                        return $self->{$prop};
1327                }
1328        }
1329}
1330
1331
1332my (%LOCKFILES, %INDEX_FILES);
1333END {
1334        unlink keys %LOCKFILES if %LOCKFILES;
1335        unlink keys %INDEX_FILES if %INDEX_FILES;
1336}
1337
1338sub resolve_local_globs {
1339        my ($url, $fetch, $glob_spec) = @_;
1340        return unless defined $glob_spec;
1341        my $ref = $glob_spec->{ref};
1342        my $path = $glob_spec->{path};
1343        foreach (command(qw#for-each-ref --format=%(refname) refs/remotes#)) {
1344                next unless m#^refs/remotes/$ref->{regex}$#;
1345                my $p = $1;
1346                my $pathname = desanitize_refname($path->full_path($p));
1347                my $refname = desanitize_refname($ref->full_path($p));
1348                if (my $existing = $fetch->{$pathname}) {
1349                        if ($existing ne $refname) {
1350                                die "Refspec conflict:\n",
1351                                    "existing: refs/remotes/$existing\n",
1352                                    " globbed: refs/remotes/$refname\n";
1353                        }
1354                        my $u = (::cmt_metadata("refs/remotes/$refname"))[0];
1355                        $u =~ s!^\Q$url\E(/|$)!! or die
1356                          "refs/remotes/$refname: '$url' not found in '$u'\n";
1357                        if ($pathname ne $u) {
1358                                warn "W: Refspec glob conflict ",
1359                                     "(ref: refs/remotes/$refname):\n",
1360                                     "expected path: $pathname\n",
1361                                     "    real path: $u\n",
1362                                     "Continuing ahead with $u\n";
1363                                next;
1364                        }
1365                } else {
1366                        $fetch->{$pathname} = $refname;
1367                }
1368        }
1369}
1370
1371sub parse_revision_argument {
1372        my ($base, $head) = @_;
1373        if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
1374                return ($base, $head);
1375        }
1376        return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
1377        return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
1378        return ($head, $head) if ($::_revision eq 'HEAD');
1379        return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
1380        return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
1381        die "revision argument: $::_revision not understood by git-svn\n";
1382}
1383
1384sub fetch_all {
1385        my ($repo_id, $remotes) = @_;
1386        if (ref $repo_id) {
1387                my $gs = $repo_id;
1388                $repo_id = undef;
1389                $repo_id = $gs->{repo_id};
1390        }
1391        $remotes ||= read_all_remotes();
1392        my $remote = $remotes->{$repo_id} or
1393                     die "[svn-remote \"$repo_id\"] unknown\n";
1394        my $fetch = $remote->{fetch};
1395        my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
1396        my (@gs, @globs);
1397        my $ra = Git::SVN::Ra->new($url);
1398        my $uuid = $ra->get_uuid;
1399        my $head = $ra->get_latest_revnum;
1400        my $base = defined $fetch ? $head : 0;
1401
1402        # read the max revs for wildcard expansion (branches/*, tags/*)
1403        foreach my $t (qw/branches tags/) {
1404                defined $remote->{$t} or next;
1405                push @globs, $remote->{$t};
1406                my $max_rev = eval { tmp_config(qw/--int --get/,
1407                                         "svn-remote.$repo_id.${t}-maxRev") };
1408                if (defined $max_rev && ($max_rev < $base)) {
1409                        $base = $max_rev;
1410                } elsif (!defined $max_rev) {
1411                        $base = 0;
1412                }
1413        }
1414
1415        if ($fetch) {
1416                foreach my $p (sort keys %$fetch) {
1417                        my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
1418                        my $lr = $gs->rev_map_max;
1419                        if (defined $lr) {
1420                                $base = $lr if ($lr < $base);
1421                        }
1422                        push @gs, $gs;
1423                }
1424        }
1425
1426        ($base, $head) = parse_revision_argument($base, $head);
1427        $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
1428}
1429
1430sub read_all_remotes {
1431        my $r = {};
1432        my $use_svm_props = eval { command_oneline(qw/config --bool
1433            svn.useSvmProps/) };
1434        $use_svm_props = $use_svm_props eq 'true' if $use_svm_props;
1435        foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
1436                if (m!^(.+)\.fetch=\s*(.*)\s*:\s*(.+)\s*$!) {
1437                        my ($remote, $local_ref, $_remote_ref) = ($1, $2, $3);
1438                        die("svn-remote.$remote: remote ref '$_remote_ref' "
1439                            . "must start with 'refs/remotes/'\n")
1440                                unless $_remote_ref =~ m{^refs/remotes/(.+)};
1441                        my $remote_ref = $1;
1442                        $local_ref =~ s{^/}{};
1443                        $r->{$remote}->{fetch}->{$local_ref} = $remote_ref;
1444                        $r->{$remote}->{svm} = {} if $use_svm_props;
1445                } elsif (m!^(.+)\.usesvmprops=\s*(.*)\s*$!) {
1446                        $r->{$1}->{svm} = {};
1447                } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
1448                        $r->{$1}->{url} = $2;
1449                } elsif (m!^(.+)\.(branches|tags)=
1450                           (.*):refs/remotes/(.+)\s*$/!x) {
1451                        my ($p, $g) = ($3, $4);
1452                        my $rs = $r->{$1}->{$2} = {
1453                                          t => $2,
1454                                          remote => $1,
1455                                          path => Git::SVN::GlobSpec->new($p),
1456                                          ref => Git::SVN::GlobSpec->new($g) };
1457                        if (length($rs->{ref}->{right}) != 0) {
1458                                die "The '*' glob character must be the last ",
1459                                    "character of '$g'\n";
1460                        }
1461                }
1462        }
1463
1464        map {
1465                if (defined $r->{$_}->{svm}) {
1466                        my $svm;
1467                        eval {
1468                                my $section = "svn-remote.$_";
1469                                $svm = {
1470                                        source => tmp_config('--get',
1471                                            "$section.svm-source"),
1472                                        replace => tmp_config('--get',
1473                                            "$section.svm-replace"),
1474                                }
1475                        };
1476                        $r->{$_}->{svm} = $svm;
1477                }
1478        } keys %$r;
1479
1480        $r;
1481}
1482
1483sub init_vars {
1484        $_gc_nr = $_gc_period = 1000;
1485        if (defined $_repack || defined $_repack_flags) {
1486               warn "Repack options are obsolete; they have no effect.\n";
1487        }
1488}
1489
1490sub verify_remotes_sanity {
1491        return unless -d $ENV{GIT_DIR};
1492        my %seen;
1493        foreach (command(qw/config -l/)) {
1494                if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
1495                        if ($seen{$1}) {
1496                                die "Remote ref refs/remote/$1 is tracked by",
1497                                    "\n  \"$_\"\nand\n  \"$seen{$1}\"\n",
1498                                    "Please resolve this ambiguity in ",
1499                                    "your git configuration file before ",
1500                                    "continuing\n";
1501                        }
1502                        $seen{$1} = $_;
1503                }
1504        }
1505}
1506
1507sub find_existing_remote {
1508        my ($url, $remotes) = @_;
1509        return undef if $no_reuse_existing;
1510        my $existing;
1511        foreach my $repo_id (keys %$remotes) {
1512                my $u = $remotes->{$repo_id}->{url} or next;
1513                next if $u ne $url;
1514                $existing = $repo_id;
1515                last;
1516        }
1517        $existing;
1518}
1519
1520sub init_remote_config {
1521        my ($self, $url, $no_write) = @_;
1522        $url =~ s!/+$!!; # strip trailing slash
1523        my $r = read_all_remotes();
1524        my $existing = find_existing_remote($url, $r);
1525        if ($existing) {
1526                unless ($no_write) {
1527                        print STDERR "Using existing ",
1528                                     "[svn-remote \"$existing\"]\n";
1529                }
1530                $self->{repo_id} = $existing;
1531        } elsif ($_minimize_url) {
1532                my $min_url = Git::SVN::Ra->new($url)->minimize_url;
1533                $existing = find_existing_remote($min_url, $r);
1534                if ($existing) {
1535                        unless ($no_write) {
1536                                print STDERR "Using existing ",
1537                                             "[svn-remote \"$existing\"]\n";
1538                        }
1539                        $self->{repo_id} = $existing;
1540                }
1541                if ($min_url ne $url) {
1542                        unless ($no_write) {
1543                                print STDERR "Using higher level of URL: ",
1544                                             "$url => $min_url\n";
1545                        }
1546                        my $old_path = $self->{path};
1547                        $self->{path} = $url;
1548                        $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
1549                        if (length $old_path) {
1550                                $self->{path} .= "/$old_path";
1551                        }
1552                        $url = $min_url;
1553                }
1554        }
1555        my $orig_url;
1556        if (!$existing) {
1557                # verify that we aren't overwriting anything:
1558                $orig_url = eval {
1559                        command_oneline('config', '--get',
1560                                        "svn-remote.$self->{repo_id}.url")
1561                };
1562                if ($orig_url && ($orig_url ne $url)) {
1563                        die "svn-remote.$self->{repo_id}.url already set: ",
1564                            "$orig_url\nwanted to set to: $url\n";
1565                }
1566        }
1567        my ($xrepo_id, $xpath) = find_ref($self->refname);
1568        if (defined $xpath) {
1569                die "svn-remote.$xrepo_id.fetch already set to track ",
1570                    "$xpath:refs/remotes/", $self->refname, "\n";
1571        }
1572        unless ($no_write) {
1573                command_noisy('config',
1574                              "svn-remote.$self->{repo_id}.url", $url);
1575                $self->{path} =~ s{^/}{};
1576                command_noisy('config', '--add',
1577                              "svn-remote.$self->{repo_id}.fetch",
1578                              "$self->{path}:".$self->refname);
1579        }
1580        $self->{url} = $url;
1581}
1582
1583sub find_by_url { # repos_root and, path are optional
1584        my ($class, $full_url, $repos_root, $path) = @_;
1585
1586        return undef unless defined $full_url;
1587        remove_username($full_url);
1588        remove_username($repos_root) if defined $repos_root;
1589        my $remotes = read_all_remotes();
1590        if (defined $full_url && defined $repos_root && !defined $path) {
1591                $path = $full_url;
1592                $path =~ s#^\Q$repos_root\E(?:/|$)##;
1593        }
1594        foreach my $repo_id (keys %$remotes) {
1595                my $u = $remotes->{$repo_id}->{url} or next;
1596                remove_username($u);
1597                next if defined $repos_root && $repos_root ne $u;
1598
1599                my $fetch = $remotes->{$repo_id}->{fetch} || {};
1600                foreach (qw/branches tags/) {
1601                        resolve_local_globs($u, $fetch,
1602                                            $remotes->{$repo_id}->{$_});
1603                }
1604                my $p = $path;
1605                my $rwr = rewrite_root({repo_id => $repo_id});
1606                my $svm = $remotes->{$repo_id}->{svm}
1607                        if defined $remotes->{$repo_id}->{svm};
1608                unless (defined $p) {
1609                        $p = $full_url;
1610                        my $z = $u;
1611                        my $prefix = '';
1612                        if ($rwr) {
1613                                $z = $rwr;
1614                        } elsif (defined $svm) {
1615                                $z = $svm->{source};
1616                                $prefix = $svm->{replace};
1617                                $prefix =~ s#^\Q$u\E(?:/|$)##;
1618                                $prefix =~ s#/$##;
1619                        }
1620                        $p =~ s#^\Q$z\E(?:/|$)#$prefix# or next;
1621                }
1622                foreach my $f (keys %$fetch) {
1623                        next if $f ne $p;
1624                        return Git::SVN->new($fetch->{$f}, $repo_id, $f);
1625                }
1626        }
1627        undef;
1628}
1629
1630sub init {
1631        my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
1632        my $self = _new($class, $repo_id, $ref_id, $path);
1633        if (defined $url) {
1634                $self->init_remote_config($url, $no_write);
1635        }
1636        $self;
1637}
1638
1639sub find_ref {
1640        my ($ref_id) = @_;
1641        foreach (command(qw/config -l/)) {
1642                next unless m!^svn-remote\.(.+)\.fetch=
1643                              \s*(.*)\s*:\s*refs/remotes/(.+)\s*$!x;
1644                my ($repo_id, $path, $ref) = ($1, $2, $3);
1645                if ($ref eq $ref_id) {
1646                        $path = '' if ($path =~ m#^\./?#);
1647                        return ($repo_id, $path);
1648                }
1649        }
1650        (undef, undef, undef);
1651}
1652
1653sub new {
1654        my ($class, $ref_id, $repo_id, $path) = @_;
1655        if (defined $ref_id && !defined $repo_id && !defined $path) {
1656                ($repo_id, $path) = find_ref($ref_id);
1657                if (!defined $repo_id) {
1658                        die "Could not find a \"svn-remote.*.fetch\" key ",
1659                            "in the repository configuration matching: ",
1660                            "refs/remotes/$ref_id\n";
1661                }
1662        }
1663        my $self = _new($class, $repo_id, $ref_id, $path);
1664        if (!defined $self->{path} || !length $self->{path}) {
1665                my $fetch = command_oneline('config', '--get',
1666                                            "svn-remote.$repo_id.fetch",
1667                                            ":refs/remotes/$ref_id\$") or
1668                     die "Failed to read \"svn-remote.$repo_id.fetch\" ",
1669                         "\":refs/remotes/$ref_id\$\" in config\n";
1670                ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
1671        }
1672        $self->{url} = command_oneline('config', '--get',
1673                                       "svn-remote.$repo_id.url") or
1674                  die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
1675        $self->rebuild;
1676        $self;
1677}
1678
1679sub refname {
1680        my ($refname) = "refs/remotes/$_[0]->{ref_id}" ;
1681
1682        # It cannot end with a slash /, we'll throw up on this because
1683        # SVN can't have directories with a slash in their name, either:
1684        if ($refname =~ m{/$}) {
1685                die "ref: '$refname' ends with a trailing slash, this is ",
1686                    "not permitted by git nor Subversion\n";
1687        }
1688
1689        # It cannot have ASCII control character space, tilde ~, caret ^,
1690        # colon :, question-mark ?, asterisk *, space, or open bracket [
1691        # anywhere.
1692        #
1693        # Additionally, % must be escaped because it is used for escaping
1694        # and we want our escaped refname to be reversible
1695        $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
1696
1697        # no slash-separated component can begin with a dot .
1698        # /.* becomes /%2E*
1699        $refname =~ s{/\.}{/%2E}g;
1700
1701        # It cannot have two consecutive dots .. anywhere
1702        # .. becomes %2E%2E
1703        $refname =~ s{\.\.}{%2E%2E}g;
1704
1705        return $refname;
1706}
1707
1708sub desanitize_refname {
1709        my ($refname) = @_;
1710        $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
1711        return $refname;
1712}
1713
1714sub svm_uuid {
1715        my ($self) = @_;
1716        return $self->{svm}->{uuid} if $self->svm;
1717        $self->ra;
1718        unless ($self->{svm}) {
1719                die "SVM UUID not cached, and reading remotely failed\n";
1720        }
1721        $self->{svm}->{uuid};
1722}
1723
1724sub svm {
1725        my ($self) = @_;
1726        return $self->{svm} if $self->{svm};
1727        my $svm;
1728        # see if we have it in our config, first:
1729        eval {
1730                my $section = "svn-remote.$self->{repo_id}";
1731                $svm = {
1732                  source => tmp_config('--get', "$section.svm-source"),
1733                  uuid => tmp_config('--get', "$section.svm-uuid"),
1734                  replace => tmp_config('--get', "$section.svm-replace"),
1735                }
1736        };
1737        if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
1738                $self->{svm} = $svm;
1739        }
1740        $self->{svm};
1741}
1742
1743sub _set_svm_vars {
1744        my ($self, $ra) = @_;
1745        return $ra if $self->svm;
1746
1747        my @err = ( "useSvmProps set, but failed to read SVM properties\n",
1748                    "(svm:source, svm:uuid) ",
1749                    "from the following URLs:\n" );
1750        sub read_svm_props {
1751                my ($self, $ra, $path, $r) = @_;
1752                my $props = ($ra->get_dir($path, $r))[2];
1753                my $src = $props->{'svm:source'};
1754                my $uuid = $props->{'svm:uuid'};
1755                return undef if (!$src || !$uuid);
1756
1757                chomp($src, $uuid);
1758
1759                $uuid =~ m{^[0-9a-f\-]{30,}$}
1760                    or die "doesn't look right - svm:uuid is '$uuid'\n";
1761
1762                # the '!' is used to mark the repos_root!/relative/path
1763                $src =~ s{/?!/?}{/};
1764                $src =~ s{/+$}{}; # no trailing slashes please
1765                # username is of no interest
1766                $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
1767
1768                my $replace = $ra->{url};
1769                $replace .= "/$path" if length $path;
1770
1771                my $section = "svn-remote.$self->{repo_id}";
1772                tmp_config("$section.svm-source", $src);
1773                tmp_config("$section.svm-replace", $replace);
1774                tmp_config("$section.svm-uuid", $uuid);
1775                $self->{svm} = {
1776                        source => $src,
1777                        uuid => $uuid,
1778                        replace => $replace
1779                };
1780        }
1781
1782        my $r = $ra->get_latest_revnum;
1783        my $path = $self->{path};
1784        my %tried;
1785        while (length $path) {
1786                unless ($tried{"$self->{url}/$path"}) {
1787                        return $ra if $self->read_svm_props($ra, $path, $r);
1788                        $tried{"$self->{url}/$path"} = 1;
1789                }
1790                $path =~ s#/?[^/]+$##;
1791        }
1792        die "Path: '$path' should be ''\n" if $path ne '';
1793        return $ra if $self->read_svm_props($ra, $path, $r);
1794        $tried{"$self->{url}/$path"} = 1;
1795
1796        if ($ra->{repos_root} eq $self->{url}) {
1797                die @err, (map { "  $_\n" } keys %tried), "\n";
1798        }
1799
1800        # nope, make sure we're connected to the repository root:
1801        my $ok;
1802        my @tried_b;
1803        $path = $ra->{svn_path};
1804        $ra = Git::SVN::Ra->new($ra->{repos_root});
1805        while (length $path) {
1806                unless ($tried{"$ra->{url}/$path"}) {
1807                        $ok = $self->read_svm_props($ra, $path, $r);
1808                        last if $ok;
1809                        $tried{"$ra->{url}/$path"} = 1;
1810                }
1811                $path =~ s#/?[^/]+$##;
1812        }
1813        die "Path: '$path' should be ''\n" if $path ne '';
1814        $ok ||= $self->read_svm_props($ra, $path, $r);
1815        $tried{"$ra->{url}/$path"} = 1;
1816        if (!$ok) {
1817                die @err, (map { "  $_\n" } keys %tried), "\n";
1818        }
1819        Git::SVN::Ra->new($self->{url});
1820}
1821
1822sub svnsync {
1823        my ($self) = @_;
1824        return $self->{svnsync} if $self->{svnsync};
1825
1826        if ($self->no_metadata) {
1827                die "Can't have both 'noMetadata' and ",
1828                    "'useSvnsyncProps' options set!\n";
1829        }
1830        if ($self->rewrite_root) {
1831                die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
1832                    "options set!\n";
1833        }
1834
1835        my $svnsync;
1836        # see if we have it in our config, first:
1837        eval {
1838                my $section = "svn-remote.$self->{repo_id}";
1839
1840                my $url = tmp_config('--get', "$section.svnsync-url");
1841                ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
1842                   die "doesn't look right - svn:sync-from-url is '$url'\n";
1843
1844                my $uuid = tmp_config('--get', "$section.svnsync-uuid");
1845                ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}) or
1846                   die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
1847
1848                $svnsync = { url => $url, uuid => $uuid }
1849        };
1850        if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
1851                return $self->{svnsync} = $svnsync;
1852        }
1853
1854        my $err = "useSvnsyncProps set, but failed to read " .
1855                  "svnsync property: svn:sync-from-";
1856        my $rp = $self->ra->rev_proplist(0);
1857
1858        my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
1859        ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
1860                   die "doesn't look right - svn:sync-from-url is '$url'\n";
1861
1862        my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
1863        ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}) or
1864                   die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
1865
1866        my $section = "svn-remote.$self->{repo_id}";
1867        tmp_config('--add', "$section.svnsync-uuid", $uuid);
1868        tmp_config('--add', "$section.svnsync-url", $url);
1869        return $self->{svnsync} = { url => $url, uuid => $uuid };
1870}
1871
1872# this allows us to memoize our SVN::Ra UUID locally and avoid a
1873# remote lookup (useful for 'git svn log').
1874sub ra_uuid {
1875        my ($self) = @_;
1876        unless ($self->{ra_uuid}) {
1877                my $key = "svn-remote.$self->{repo_id}.uuid";
1878                my $uuid = eval { tmp_config('--get', $key) };
1879                if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/) {
1880                        $self->{ra_uuid} = $uuid;
1881                } else {
1882                        die "ra_uuid called without URL\n" unless $self->{url};
1883                        $self->{ra_uuid} = $self->ra->get_uuid;
1884                        tmp_config('--add', $key, $self->{ra_uuid});
1885                }
1886        }
1887        $self->{ra_uuid};
1888}
1889
1890sub _set_repos_root {
1891        my ($self, $repos_root) = @_;
1892        my $k = "svn-remote.$self->{repo_id}.reposRoot";
1893        $repos_root ||= $self->ra->{repos_root};
1894        tmp_config($k, $repos_root);
1895        $repos_root;
1896}
1897
1898sub repos_root {
1899        my ($self) = @_;
1900        my $k = "svn-remote.$self->{repo_id}.reposRoot";
1901        eval { tmp_config('--get', $k) } || $self->_set_repos_root;
1902}
1903
1904sub ra {
1905        my ($self) = shift;
1906        my $ra = Git::SVN::Ra->new($self->{url});
1907        $self->_set_repos_root($ra->{repos_root});
1908        if ($self->use_svm_props && !$self->{svm}) {
1909                if ($self->no_metadata) {
1910                        die "Can't have both 'noMetadata' and ",
1911                            "'useSvmProps' options set!\n";
1912                } elsif ($self->use_svnsync_props) {
1913                        die "Can't have both 'useSvnsyncProps' and ",
1914                            "'useSvmProps' options set!\n";
1915                }
1916                $ra = $self->_set_svm_vars($ra);
1917                $self->{-want_revprops} = 1;
1918        }
1919        $ra;
1920}
1921
1922sub rel_path {
1923        my ($self) = @_;
1924        my $repos_root = $self->ra->{repos_root};
1925        return $self->{path} if ($self->{url} eq $repos_root);
1926        my $url = $self->{url} .
1927                  (length $self->{path} ? "/$self->{path}" : $self->{path});
1928        $url =~ s!^\Q$repos_root\E(?:/+|$)!!g;
1929        $url;
1930}
1931
1932# prop_walk(PATH, REV, SUB)
1933# -------------------------
1934# Recursively traverse PATH at revision REV and invoke SUB for each
1935# directory that contains a SVN property.  SUB will be invoked as
1936# follows:  &SUB(gs, path, props);  where `gs' is this instance of
1937# Git::SVN, `path' the path to the directory where the properties
1938# `props' were found.  The `path' will be relative to point of checkout,
1939# that is, if url://repo/trunk is the current Git branch, and that
1940# directory contains a sub-directory `d', SUB will be invoked with `/d/'
1941# as `path' (note the trailing `/').
1942sub prop_walk {
1943        my ($self, $path, $rev, $sub) = @_;
1944
1945        $path =~ s#^/##;
1946        my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev);
1947        $path =~ s#^/*#/#g;
1948        my $p = $path;
1949        # Strip the irrelevant part of the path.
1950        $p =~ s#^/+\Q$self->{path}\E(/|$)#/#;
1951        # Ensure the path is terminated by a `/'.
1952        $p =~ s#/*$#/#;
1953
1954        # The properties contain all the internal SVN stuff nobody
1955        # (usually) cares about.
1956        my $interesting_props = 0;
1957        foreach (keys %{$props}) {
1958                # If it doesn't start with `svn:', it must be a
1959                # user-defined property.
1960                ++$interesting_props and next if $_ !~ /^svn:/;
1961                # FIXME: Fragile, if SVN adds new public properties,
1962                # this needs to be updated.
1963                ++$interesting_props if /^svn:(?:ignore|keywords|executable
1964                                                 |eol-style|mime-type
1965                                                 |externals|needs-lock)$/x;
1966        }
1967        &$sub($self, $p, $props) if $interesting_props;
1968
1969        foreach (sort keys %$dirent) {
1970                next if $dirent->{$_}->{kind} != $SVN::Node::dir;
1971                $self->prop_walk($self->{path} . $p . $_, $rev, $sub);
1972        }
1973}
1974
1975sub last_rev { ($_[0]->last_rev_commit)[0] }
1976sub last_commit { ($_[0]->last_rev_commit)[1] }
1977
1978# returns the newest SVN revision number and newest commit SHA1
1979sub last_rev_commit {
1980        my ($self) = @_;
1981        if (defined $self->{last_rev} && defined $self->{last_commit}) {
1982                return ($self->{last_rev}, $self->{last_commit});
1983        }
1984        my $c = ::verify_ref($self->refname.'^0');
1985        if ($c && !$self->use_svm_props && !$self->no_metadata) {
1986                my $rev = (::cmt_metadata($c))[1];
1987                if (defined $rev) {
1988                        ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
1989                        return ($rev, $c);
1990                }
1991        }
1992        my $map_path = $self->map_path;
1993        unless (-e $map_path) {
1994                ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
1995                return (undef, undef);
1996        }
1997        my ($rev, $commit) = $self->rev_map_max(1);
1998        ($self->{last_rev}, $self->{last_commit}) = ($rev, $commit);
1999        return ($rev, $commit);
2000}
2001
2002sub get_fetch_range {
2003        my ($self, $min, $max) = @_;
2004        $max ||= $self->ra->get_latest_revnum;
2005        $min ||= $self->rev_map_max;
2006        (++$min, $max);
2007}
2008
2009sub tmp_config {
2010        my (@args) = @_;
2011        my $old_def_config = "$ENV{GIT_DIR}/svn/config";
2012        my $config = "$ENV{GIT_DIR}/svn/.metadata";
2013        if (! -f $config && -f $old_def_config) {
2014                rename $old_def_config, $config or
2015                       die "Failed rename $old_def_config => $config: $!\n";
2016        }
2017        my $old_config = $ENV{GIT_CONFIG};
2018        $ENV{GIT_CONFIG} = $config;
2019        $@ = undef;
2020        my @ret = eval {
2021                unless (-f $config) {
2022                        mkfile($config);
2023                        open my $fh, '>', $config or
2024                            die "Can't open $config: $!\n";
2025                        print $fh "; This file is used internally by ",
2026                                  "git-svn\n" or die
2027                                  "Couldn't write to $config: $!\n";
2028                        print $fh "; You should not have to edit it\n" or
2029                              die "Couldn't write to $config: $!\n";
2030                        close $fh or die "Couldn't close $config: $!\n";
2031                }
2032                command('config', @args);
2033        };
2034        my $err = $@;
2035        if (defined $old_config) {
2036                $ENV{GIT_CONFIG} = $old_config;
2037        } else {
2038                delete $ENV{GIT_CONFIG};
2039        }
2040        die $err if $err;
2041        wantarray ? @ret : $ret[0];
2042}
2043
2044sub tmp_index_do {
2045        my ($self, $sub) = @_;
2046        my $old_index = $ENV{GIT_INDEX_FILE};
2047        $ENV{GIT_INDEX_FILE} = $self->{index};
2048        $@ = undef;
2049        my @ret = eval {
2050                my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
2051                mkpath([$dir]) unless -d $dir;
2052                &$sub;
2053        };
2054        my $err = $@;
2055        if (defined $old_index) {
2056                $ENV{GIT_INDEX_FILE} = $old_index;
2057        } else {
2058                delete $ENV{GIT_INDEX_FILE};
2059        }
2060        die $err if $err;
2061        wantarray ? @ret : $ret[0];
2062}
2063
2064sub assert_index_clean {
2065        my ($self, $treeish) = @_;
2066
2067        $self->tmp_index_do(sub {
2068                command_noisy('read-tree', $treeish) unless -e $self->{index};
2069                my $x = command_oneline('write-tree');
2070                my ($y) = (command(qw/cat-file commit/, $treeish) =~
2071                           /^tree ($::sha1)/mo);
2072                return if $y eq $x;
2073
2074                warn "Index mismatch: $y != $x\nrereading $treeish\n";
2075                unlink $self->{index} or die "unlink $self->{index}: $!\n";
2076                command_noisy('read-tree', $treeish);
2077                $x = command_oneline('write-tree');
2078                if ($y ne $x) {
2079                        ::fatal "trees ($treeish) $y != $x\n",
2080                                "Something is seriously wrong...";
2081                }
2082        });
2083}
2084
2085sub get_commit_parents {
2086        my ($self, $log_entry) = @_;
2087        my (%seen, @ret, @tmp);
2088        # legacy support for 'set-tree'; this is only used by set_tree_cb:
2089        if (my $ip = $self->{inject_parents}) {
2090                if (my $commit = delete $ip->{$log_entry->{revision}}) {
2091                        push @tmp, $commit;
2092                }
2093        }
2094        if (my $cur = ::verify_ref($self->refname.'^0')) {
2095                push @tmp, $cur;
2096        }
2097        if (my $ipd = $self->{inject_parents_dcommit}) {
2098                if (my $commit = delete $ipd->{$log_entry->{revision}}) {
2099                        push @tmp, @$commit;
2100                }
2101        }
2102        push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
2103        while (my $p = shift @tmp) {
2104                next if $seen{$p};
2105                $seen{$p} = 1;
2106                push @ret, $p;
2107                # MAXPARENT is defined to 16 in commit-tree.c:
2108                last if @ret >= 16;
2109        }
2110        if (@tmp) {
2111                die "r$log_entry->{revision}: No room for parents:\n\t",
2112                    join("\n\t", @tmp), "\n";
2113        }
2114        @ret;
2115}
2116
2117sub rewrite_root {
2118        my ($self) = @_;
2119        return $self->{-rewrite_root} if exists $self->{-rewrite_root};
2120        my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
2121        my $rwr = eval { command_oneline(qw/config --get/, $k) };
2122        if ($rwr) {
2123                $rwr =~ s#/+$##;
2124                if ($rwr !~ m#^[a-z\+]+://#) {
2125                        die "$rwr is not a valid URL (key: $k)\n";
2126                }
2127        }
2128        $self->{-rewrite_root} = $rwr;
2129}
2130
2131sub metadata_url {
2132        my ($self) = @_;
2133        ($self->rewrite_root || $self->{url}) .
2134           (length $self->{path} ? '/' . $self->{path} : '');
2135}
2136
2137sub full_url {
2138        my ($self) = @_;
2139        $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
2140}
2141
2142
2143sub set_commit_header_env {
2144        my ($log_entry) = @_;
2145        my %env;
2146        foreach my $ned (qw/NAME EMAIL DATE/) {
2147                foreach my $ac (qw/AUTHOR COMMITTER/) {
2148                        $env{"GIT_${ac}_${ned}"} = $ENV{"GIT_${ac}_${ned}"};
2149                }
2150        }
2151
2152        $ENV{GIT_AUTHOR_NAME} = $log_entry->{name};
2153        $ENV{GIT_AUTHOR_EMAIL} = $log_entry->{email};
2154        $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
2155
2156        $ENV{GIT_COMMITTER_NAME} = (defined $log_entry->{commit_name})
2157                                                ? $log_entry->{commit_name}
2158                                                : $log_entry->{name};
2159        $ENV{GIT_COMMITTER_EMAIL} = (defined $log_entry->{commit_email})
2160                                                ? $log_entry->{commit_email}
2161                                                : $log_entry->{email};
2162        \%env;
2163}
2164
2165sub restore_commit_header_env {
2166        my ($env) = @_;
2167        foreach my $ned (qw/NAME EMAIL DATE/) {
2168                foreach my $ac (qw/AUTHOR COMMITTER/) {
2169                        my $k = "GIT_${ac}_${ned}";
2170                        if (defined $env->{$k}) {
2171                                $ENV{$k} = $env->{$k};
2172                        } else {
2173                                delete $ENV{$k};
2174                        }
2175                }
2176        }
2177}
2178
2179sub gc {
2180        command_noisy('gc', '--auto');
2181};
2182
2183sub do_git_commit {
2184        my ($self, $log_entry) = @_;
2185        my $lr = $self->last_rev;
2186        if (defined $lr && $lr >= $log_entry->{revision}) {
2187                die "Last fetched revision of ", $self->refname,
2188                    " was r$lr, but we are about to fetch: ",
2189                    "r$log_entry->{revision}!\n";
2190        }
2191        if (my $c = $self->rev_map_get($log_entry->{revision})) {
2192                croak "$log_entry->{revision} = $c already exists! ",
2193                      "Why are we refetching it?\n";
2194        }
2195        my $old_env = set_commit_header_env($log_entry);
2196        my $tree = $log_entry->{tree};
2197        if (!defined $tree) {
2198                $tree = $self->tmp_index_do(sub {
2199                                            command_oneline('write-tree') });
2200        }
2201        die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
2202
2203        my @exec = ('git-commit-tree', $tree);
2204        foreach ($self->get_commit_parents($log_entry)) {
2205                push @exec, '-p', $_;
2206        }
2207        defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
2208                                                                   or croak $!;
2209        print $msg_fh $log_entry->{log} or croak $!;
2210        restore_commit_header_env($old_env);
2211        unless ($self->no_metadata) {
2212                print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
2213                              or croak $!;
2214        }
2215        $msg_fh->flush == 0 or croak $!;
2216        close $msg_fh or croak $!;
2217        chomp(my $commit = do { local $/; <$out_fh> });
2218        close $out_fh or croak $!;
2219        waitpid $pid, 0;
2220        croak $? if $?;
2221        if ($commit !~ /^$::sha1$/o) {
2222                die "Failed to commit, invalid sha1: $commit\n";
2223        }
2224
2225        $self->rev_map_set($log_entry->{revision}, $commit, 1);
2226
2227        $self->{last_rev} = $log_entry->{revision};
2228        $self->{last_commit} = $commit;
2229        print "r$log_entry->{revision}";
2230        if (defined $log_entry->{svm_revision}) {
2231                 print " (\@$log_entry->{svm_revision})";
2232                 $self->rev_map_set($log_entry->{svm_revision}, $commit,
2233                                   0, $self->svm_uuid);
2234        }
2235        print " = $commit ($self->{ref_id})\n";
2236        if (--$_gc_nr == 0) {
2237                $_gc_nr = $_gc_period;
2238                gc();
2239        }
2240        return $commit;
2241}
2242
2243sub match_paths {
2244        my ($self, $paths, $r) = @_;
2245        return 1 if $self->{path} eq '';
2246        if (my $path = $paths->{"/$self->{path}"}) {
2247                return ($path->{action} eq 'D') ? 0 : 1;
2248        }
2249        $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
2250        if (grep /$self->{path_regex}/, keys %$paths) {
2251                return 1;
2252        }
2253        my $c = '';
2254        foreach (split m#/#, $self->{path}) {
2255                $c .= "/$_";
2256                next unless ($paths->{$c} &&
2257                             ($paths->{$c}->{action} =~ /^[AR]$/));
2258                if ($self->ra->check_path($self->{path}, $r) ==
2259                    $SVN::Node::dir) {
2260                        return 1;
2261                }
2262        }
2263        return 0;
2264}
2265
2266sub find_parent_branch {
2267        my ($self, $paths, $rev) = @_;
2268        return undef unless $self->follow_parent;
2269        unless (defined $paths) {
2270                my $err_handler = $SVN::Error::handler;
2271                $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
2272                $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1, sub {
2273                                   $paths =
2274                                      Git::SVN::Ra::dup_changed_paths($_[0]) });
2275                $SVN::Error::handler = $err_handler;
2276        }
2277        return undef unless defined $paths;
2278
2279        # look for a parent from another branch:
2280        my @b_path_components = split m#/#, $self->rel_path;
2281        my @a_path_components;
2282        my $i;
2283        while (@b_path_components) {
2284                $i = $paths->{'/'.join('/', @b_path_components)};
2285                last if $i && defined $i->{copyfrom_path};
2286                unshift(@a_path_components, pop(@b_path_components));
2287        }
2288        return undef unless defined $i && defined $i->{copyfrom_path};
2289        my $branch_from = $i->{copyfrom_path};
2290        if (@a_path_components) {
2291                print STDERR "branch_from: $branch_from => ";
2292                $branch_from .= '/'.join('/', @a_path_components);
2293                print STDERR $branch_from, "\n";
2294        }
2295        my $r = $i->{copyfrom_rev};
2296        my $repos_root = $self->ra->{repos_root};
2297        my $url = $self->ra->{url};
2298        my $new_url = $repos_root . $branch_from;
2299        print STDERR  "Found possible branch point: ",
2300                      "$new_url => ", $self->full_url, ", $r\n";
2301        $branch_from =~ s#^/##;
2302        my $gs = Git::SVN->find_by_url($new_url, $repos_root, $branch_from);
2303        unless ($gs) {
2304                my $ref_id = $self->{ref_id};
2305                $ref_id =~ s/\@\d+$//;
2306                $ref_id .= "\@$r";
2307                # just grow a tail if we're not unique enough :x
2308                $ref_id .= '-' while find_ref($ref_id);
2309                print STDERR "Initializing parent: $ref_id\n";
2310                my ($u, $p, $repo_id) = ($new_url, '', $ref_id);
2311                if ($u =~ s#^\Q$url\E(/|$)##) {
2312                        $p = $u;
2313                        $u = $url;
2314                        $repo_id = $self->{repo_id};
2315                }
2316                $gs = Git::SVN->init($u, $p, $repo_id, $ref_id, 1);
2317        }
2318        my ($r0, $parent) = $gs->find_rev_before($r, 1);
2319        if (!defined $r0 || !defined $parent) {
2320                my ($base, $head) = parse_revision_argument(0, $r);
2321                if ($base <= $r) {
2322                        $gs->fetch($base, $r);
2323                }
2324                ($r0, $parent) = $gs->last_rev_commit;
2325        }
2326        if (defined $r0 && defined $parent) {
2327                print STDERR "Found branch parent: ($self->{ref_id}) $parent\n";
2328                my $ed;
2329                if ($self->ra->can_do_switch) {
2330                        $self->assert_index_clean($parent);
2331                        print STDERR "Following parent with do_switch\n";
2332                        # do_switch works with svn/trunk >= r22312, but that
2333                        # is not included with SVN 1.4.3 (the latest version
2334                        # at the moment), so we can't rely on it
2335                        $self->{last_commit} = $parent;
2336                        $ed = SVN::Git::Fetcher->new($self);
2337                        $gs->ra->gs_do_switch($r0, $rev, $gs,
2338                                              $self->full_url, $ed)
2339                          or die "SVN connection failed somewhere...\n";
2340                } elsif ($self->ra->trees_match($new_url, $r0,
2341                                                $self->full_url, $rev)) {
2342                        print STDERR "Trees match:\n",
2343                                     "  $new_url\@$r0\n",
2344                                     "  ${\$self->full_url}\@$rev\n",
2345                                     "Following parent with no changes\n";
2346                        $self->tmp_index_do(sub {
2347                            command_noisy('read-tree', $parent);
2348                        });
2349                        $self->{last_commit} = $parent;
2350                } else {
2351                        print STDERR "Following parent with do_update\n";
2352                        $ed = SVN::Git::Fetcher->new($self);
2353                        $self->ra->gs_do_update($rev, $rev, $self, $ed)
2354                          or die "SVN connection failed somewhere...\n";
2355                }
2356                print STDERR "Successfully followed parent\n";
2357                return $self->make_log_entry($rev, [$parent], $ed);
2358        }
2359        return undef;
2360}
2361
2362sub do_fetch {
2363        my ($self, $paths, $rev) = @_;
2364        my $ed;
2365        my ($last_rev, @parents);
2366        if (my $lc = $self->last_commit) {
2367                # we can have a branch that was deleted, then re-added
2368                # under the same name but copied from another path, in
2369                # which case we'll have multiple parents (we don't
2370                # want to break the original ref, nor lose copypath info):
2371                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2372                        push @{$log_entry->{parents}}, $lc;
2373                        return $log_entry;
2374                }
2375                $ed = SVN::Git::Fetcher->new($self);
2376                $last_rev = $self->{last_rev};
2377                $ed->{c} = $lc;
2378                @parents = ($lc);
2379        } else {
2380                $last_rev = $rev;
2381                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2382                        return $log_entry;
2383                }
2384                $ed = SVN::Git::Fetcher->new($self);
2385        }
2386        unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
2387                die "SVN connection failed somewhere...\n";
2388        }
2389        $self->make_log_entry($rev, \@parents, $ed);
2390}
2391
2392sub get_untracked {
2393        my ($self, $ed) = @_;
2394        my @out;
2395        my $h = $ed->{empty};
2396        foreach (sort keys %$h) {
2397                my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
2398                push @out, "  $act: " . uri_encode($_);
2399                warn "W: $act: $_\n";
2400        }
2401        foreach my $t (qw/dir_prop file_prop/) {
2402                $h = $ed->{$t} or next;
2403                foreach my $path (sort keys %$h) {
2404                        my $ppath = $path eq '' ? '.' : $path;
2405                        foreach my $prop (sort keys %{$h->{$path}}) {
2406                                next if $SKIP_PROP{$prop};
2407                                my $v = $h->{$path}->{$prop};
2408                                my $t_ppath_prop = "$t: " .
2409                                                    uri_encode($ppath) . ' ' .
2410                                                    uri_encode($prop);
2411                                if (defined $v) {
2412                                        push @out, "  +$t_ppath_prop " .
2413                                                   uri_encode($v);
2414                                } else {
2415                                        push @out, "  -$t_ppath_prop";
2416                                }
2417                        }
2418                }
2419        }
2420        foreach my $t (qw/absent_file absent_directory/) {
2421                $h = $ed->{$t} or next;
2422                foreach my $parent (sort keys %$h) {
2423                        foreach my $path (sort @{$h->{$parent}}) {
2424                                push @out, "  $t: " .
2425                                           uri_encode("$parent/$path");
2426                                warn "W: $t: $parent/$path ",
2427                                     "Insufficient permissions?\n";
2428                        }
2429                }
2430        }
2431        \@out;
2432}
2433
2434sub parse_svn_date {
2435        my $date = shift || return '+0000 1970-01-01 00:00:00';
2436        my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2437                                            (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x) or
2438                                         croak "Unable to parse date: $date\n";
2439        "+0000 $Y-$m-$d $H:$M:$S";
2440}
2441
2442sub check_author {
2443        my ($author) = @_;
2444        if (!defined $author || length $author == 0) {
2445                $author = '(no author)';
2446        } elsif (defined $::_authors && ! defined $::users{$author}) {
2447                die "Author: $author not defined in $::_authors file\n";
2448        }
2449        $author;
2450}
2451
2452sub make_log_entry {
2453        my ($self, $rev, $parents, $ed) = @_;
2454        my $untracked = $self->get_untracked($ed);
2455
2456        open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
2457        print $un "r$rev\n" or croak $!;
2458        print $un $_, "\n" foreach @$untracked;
2459        my %log_entry = ( parents => $parents || [], revision => $rev,
2460                          log => '');
2461
2462        my $headrev;
2463        my $logged = delete $self->{logged_rev_props};
2464        if (!$logged || $self->{-want_revprops}) {
2465                my $rp = $self->ra->rev_proplist($rev);
2466                foreach (sort keys %$rp) {
2467                        my $v = $rp->{$_};
2468                        if (/^svn:(author|date|log)$/) {
2469                                $log_entry{$1} = $v;
2470                        } elsif ($_ eq 'svm:headrev') {
2471                                $headrev = $v;
2472                        } else {
2473                                print $un "  rev_prop: ", uri_encode($_), ' ',
2474                                          uri_encode($v), "\n";
2475                        }
2476                }
2477        } else {
2478                map { $log_entry{$_} = $logged->{$_} } keys %$logged;
2479        }
2480        close $un or croak $!;
2481
2482        $log_entry{date} = parse_svn_date($log_entry{date});
2483        $log_entry{log} .= "\n";
2484        my $author = $log_entry{author} = check_author($log_entry{author});
2485        my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
2486                                                       : ($author, undef);
2487
2488        my ($commit_name, $commit_email) = ($name, $email);
2489        if ($_use_log_author) {
2490                my $name_field;
2491                if ($log_entry{log} =~ /From:\s+(.*\S)\s*\n/i) {
2492                        $name_field = $1;
2493                } elsif ($log_entry{log} =~ /Signed-off-by:\s+(.*\S)\s*\n/i) {
2494                        $name_field = $1;
2495                }
2496                if (!defined $name_field) {
2497                        if (!defined $email) {
2498                                $email = $name;
2499                        }
2500                } elsif ($name_field =~ /(.*?)\s+<(.*)>/) {
2501                        ($name, $email) = ($1, $2);
2502                } elsif ($name_field =~ /(.*)@/) {
2503                        ($name, $email) = ($1, $name_field);
2504                } else {
2505                        ($name, $email) = ($name_field, $name_field);
2506                }
2507        }
2508        if (defined $headrev && $self->use_svm_props) {
2509                if ($self->rewrite_root) {
2510                        die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
2511                            "options set!\n";
2512                }
2513                my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$};
2514                # we don't want "SVM: initializing mirror for junk" ...
2515                return undef if $r == 0;
2516                my $svm = $self->svm;
2517                if ($uuid ne $svm->{uuid}) {
2518                        die "UUID mismatch on SVM path:\n",
2519                            "expected: $svm->{uuid}\n",
2520                            "     got: $uuid\n";
2521                }
2522                my $full_url = $self->full_url;
2523                $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
2524                             die "Failed to replace '$svm->{replace}' with ",
2525                                 "'$svm->{source}' in $full_url\n";
2526                # throw away username for storing in records
2527                remove_username($full_url);
2528                $log_entry{metadata} = "$full_url\@$r $uuid";
2529                $log_entry{svm_revision} = $r;
2530                $email ||= "$author\@$uuid";
2531                $commit_email ||= "$author\@$uuid";
2532        } elsif ($self->use_svnsync_props) {
2533                my $full_url = $self->svnsync->{url};
2534                $full_url .= "/$self->{path}" if length $self->{path};
2535                remove_username($full_url);
2536                my $uuid = $self->svnsync->{uuid};
2537                $log_entry{metadata} = "$full_url\@$rev $uuid";
2538                $email ||= "$author\@$uuid";
2539                $commit_email ||= "$author\@$uuid";
2540        } else {
2541                my $url = $self->metadata_url;
2542                remove_username($url);
2543                $log_entry{metadata} = "$url\@$rev " .
2544                                       $self->ra->get_uuid;
2545                $email ||= "$author\@" . $self->ra->get_uuid;
2546                $commit_email ||= "$author\@" . $self->ra->get_uuid;
2547        }
2548        $log_entry{name} = $name;
2549        $log_entry{email} = $email;
2550        $log_entry{commit_name} = $commit_name;
2551        $log_entry{commit_email} = $commit_email;
2552        \%log_entry;
2553}
2554
2555sub fetch {
2556        my ($self, $min_rev, $max_rev, @parents) = @_;
2557        my ($last_rev, $last_commit) = $self->last_rev_commit;
2558        my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
2559        $self->ra->gs_fetch_loop_common($base, $head, [$self]);
2560}
2561
2562sub set_tree_cb {
2563        my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
2564        $self->{inject_parents} = { $rev => $tree };
2565        $self->fetch(undef, undef);
2566}
2567
2568sub set_tree {
2569        my ($self, $tree) = (shift, shift);
2570        my $log_entry = ::get_commit_entry($tree);
2571        unless ($self->{last_rev}) {
2572                fatal("Must have an existing revision to commit");
2573        }
2574        my %ed_opts = ( r => $self->{last_rev},
2575                        log => $log_entry->{log},
2576                        ra => $self->ra,
2577                        tree_a => $self->{last_commit},
2578                        tree_b => $tree,
2579                        editor_cb => sub {
2580                               $self->set_tree_cb($log_entry, $tree, @_) },
2581                        svn_path => $self->{path} );
2582        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
2583                print "No changes\nr$self->{last_rev} = $tree\n";
2584        }
2585}
2586
2587sub rebuild_from_rev_db {
2588        my ($self, $path) = @_;
2589        my $r = -1;
2590        open my $fh, '<', $path or croak "open: $!";
2591        binmode $fh or croak "binmode: $!";
2592        while (<$fh>) {
2593                length($_) == 41 or croak "inconsistent size in ($_) != 41";
2594                chomp($_);
2595                ++$r;
2596                next if $_ eq ('0' x 40);
2597                $self->rev_map_set($r, $_);
2598                print "r$r = $_\n";
2599        }
2600        close $fh or croak "close: $!";
2601        unlink $path or croak "unlink: $!";
2602}
2603
2604sub rebuild {
2605        my ($self) = @_;
2606        my $map_path = $self->map_path;
2607        return if (-e $map_path && ! -z $map_path);
2608        return unless ::verify_ref($self->refname.'^0');
2609        if ($self->use_svm_props || $self->no_metadata) {
2610                my $rev_db = $self->rev_db_path;
2611                $self->rebuild_from_rev_db($rev_db);
2612                if ($self->use_svm_props) {
2613                        my $svm_rev_db = $self->rev_db_path($self->svm_uuid);
2614                        $self->rebuild_from_rev_db($svm_rev_db);
2615                }
2616                $self->unlink_rev_db_symlink;
2617                return;
2618        }
2619        print "Rebuilding $map_path ...\n";
2620        my ($log, $ctx) =
2621            command_output_pipe(qw/rev-list --pretty=raw --no-color --reverse/,
2622                                $self->refname, '--');
2623        my $metadata_url = $self->metadata_url;
2624        remove_username($metadata_url);
2625        my $svn_uuid = $self->ra_uuid;
2626        my $c;
2627        while (<$log>) {
2628                if ( m{^commit ($::sha1)$} ) {
2629                        $c = $1;
2630                        next;
2631                }
2632                next unless s{^\s*(git-svn-id:)}{$1};
2633                my ($url, $rev, $uuid) = ::extract_metadata($_);
2634                remove_username($url);
2635
2636                # ignore merges (from set-tree)
2637                next if (!defined $rev || !$uuid);
2638
2639                # if we merged or otherwise started elsewhere, this is
2640                # how we break out of it
2641                if (($uuid ne $svn_uuid) ||
2642                    ($metadata_url && $url && ($url ne $metadata_url))) {
2643                        next;
2644                }
2645
2646                $self->rev_map_set($rev, $c);
2647                print "r$rev = $c\n";
2648        }
2649        command_close_pipe($log, $ctx);
2650        print "Done rebuilding $map_path\n";
2651        my $rev_db_path = $self->rev_db_path;
2652        if (-f $self->rev_db_path) {
2653                unlink $self->rev_db_path or croak "unlink: $!";
2654        }
2655        $self->unlink_rev_db_symlink;
2656}
2657
2658# rev_map:
2659# Tie::File seems to be prone to offset errors if revisions get sparse,
2660# it's not that fast, either.  Tie::File is also not in Perl 5.6.  So
2661# one of my favorite modules is out :<  Next up would be one of the DBM
2662# modules, but I'm not sure which is most portable...
2663#
2664# This is the replacement for the rev_db format, which was too big
2665# and inefficient for large repositories with a lot of sparse history
2666# (mainly tags)
2667#
2668# The format is this:
2669#   - 24 bytes for every record,
2670#     * 4 bytes for the integer representing an SVN revision number
2671#     * 20 bytes representing the sha1 of a git commit
2672#   - No empty padding records like the old format
2673#     (except the last record, which can be overwritten)
2674#   - new records are written append-only since SVN revision numbers
2675#     increase monotonically
2676#   - lookups on SVN revision number are done via a binary search
2677#   - Piping the file to xxd -c24 is a good way of dumping it for
2678#     viewing or editing (piped back through xxd -r), should the need
2679#     ever arise.
2680#   - The last record can be padding revision with an all-zero sha1
2681#     This is used to optimize fetch performance when using multiple
2682#     "fetch" directives in .git/config
2683#
2684# These files are disposable unless noMetadata or useSvmProps is set
2685
2686sub _rev_map_set {
2687        my ($fh, $rev, $commit) = @_;
2688
2689        binmode $fh or croak "binmode: $!";
2690        my $size = (stat($fh))[7];
2691        ($size % 24) == 0 or croak "inconsistent size: $size";
2692
2693        my $wr_offset = 0;
2694        if ($size > 0) {
2695                sysseek($fh, -24, SEEK_END) or croak "seek: $!";
2696                my $read = sysread($fh, my $buf, 24) or croak "read: $!";
2697                $read == 24 or croak "read only $read bytes (!= 24)";
2698                my ($last_rev, $last_commit) = unpack(rev_map_fmt, $buf);
2699                if ($last_commit eq ('0' x40)) {
2700                        if ($size >= 48) {
2701                                sysseek($fh, -48, SEEK_END) or croak "seek: $!";
2702                                $read = sysread($fh, $buf, 24) or
2703                                    croak "read: $!";
2704                                $read == 24 or
2705                                    croak "read only $read bytes (!= 24)";
2706                                ($last_rev, $last_commit) =
2707                                    unpack(rev_map_fmt, $buf);
2708                                if ($last_commit eq ('0' x40)) {
2709                                        croak "inconsistent .rev_map\n";
2710                                }
2711                        }
2712                        if ($last_rev >= $rev) {
2713                                croak "last_rev is higher!: $last_rev >= $rev";
2714                        }
2715                        $wr_offset = -24;
2716                }
2717        }
2718        sysseek($fh, $wr_offset, SEEK_END) or croak "seek: $!";
2719        syswrite($fh, pack(rev_map_fmt, $rev, $commit), 24) == 24 or
2720          croak "write: $!";
2721}
2722
2723sub mkfile {
2724        my ($path) = @_;
2725        unless (-e $path) {
2726                my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
2727                mkpath([$dir]) unless -d $dir;
2728                open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
2729                close $fh or die "Couldn't close (create) $path: $!\n";
2730        }
2731}
2732
2733sub rev_map_set {
2734        my ($self, $rev, $commit, $update_ref, $uuid) = @_;
2735        length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
2736        my $db = $self->map_path($uuid);
2737        my $db_lock = "$db.lock";
2738        my $sig;
2739        if ($update_ref) {
2740                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
2741                            $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
2742        }
2743        mkfile($db);
2744
2745        $LOCKFILES{$db_lock} = 1;
2746        my $sync;
2747        # both of these options make our .rev_db file very, very important
2748        # and we can't afford to lose it because rebuild() won't work
2749        if ($self->use_svm_props || $self->no_metadata) {
2750                $sync = 1;
2751                copy($db, $db_lock) or die "rev_map_set(@_): ",
2752                                           "Failed to copy: ",
2753                                           "$db => $db_lock ($!)\n";
2754        } else {
2755                rename $db, $db_lock or die "rev_map_set(@_): ",
2756                                            "Failed to rename: ",
2757                                            "$db => $db_lock ($!)\n";
2758        }
2759
2760        sysopen(my $fh, $db_lock, O_RDWR | O_CREAT)
2761             or croak "Couldn't open $db_lock: $!\n";
2762        _rev_map_set($fh, $rev, $commit);
2763        if ($sync) {
2764                $fh->flush or die "Couldn't flush $db_lock: $!\n";
2765                $fh->sync or die "Couldn't sync $db_lock: $!\n";
2766        }
2767        close $fh or croak $!;
2768        if ($update_ref) {
2769                $_head = $self;
2770                command_noisy('update-ref', '-m', "r$rev",
2771                              $self->refname, $commit);
2772        }
2773        rename $db_lock, $db or die "rev_map_set(@_): ", "Failed to rename: ",
2774                                    "$db_lock => $db ($!)\n";
2775        delete $LOCKFILES{$db_lock};
2776        if ($update_ref) {
2777                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
2778                            $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
2779                kill $sig, $$ if defined $sig;
2780        }
2781}
2782
2783# If want_commit, this will return an array of (rev, commit) where
2784# commit _must_ be a valid commit in the archive.
2785# Otherwise, it'll return the max revision (whether or not the
2786# commit is valid or just a 0x40 placeholder).
2787sub rev_map_max {
2788        my ($self, $want_commit) = @_;
2789        $self->rebuild;
2790        my $map_path = $self->map_path;
2791        stat $map_path or return $want_commit ? (0, undef) : 0;
2792        sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
2793        binmode $fh or croak "binmode: $!";
2794        my $size = (stat($fh))[7];
2795        ($size % 24) == 0 or croak "inconsistent size: $size";
2796
2797        if ($size == 0) {
2798                close $fh or croak "close: $!";
2799                return $want_commit ? (0, undef) : 0;
2800        }
2801
2802        sysseek($fh, -24, SEEK_END) or croak "seek: $!";
2803        sysread($fh, my $buf, 24) == 24 or croak "read: $!";
2804        my ($r, $c) = unpack(rev_map_fmt, $buf);
2805        if ($want_commit && $c eq ('0' x40)) {
2806                if ($size < 48) {
2807                        return $want_commit ? (0, undef) : 0;
2808                }
2809                sysseek($fh, -48, SEEK_END) or croak "seek: $!";
2810                sysread($fh, $buf, 24) == 24 or croak "read: $!";
2811                ($r, $c) = unpack(rev_map_fmt, $buf);
2812                if ($c eq ('0'x40)) {
2813                        croak "Penultimate record is all-zeroes in $map_path";
2814                }
2815        }
2816        close $fh or croak "close: $!";
2817        $want_commit ? ($r, $c) : $r;
2818}
2819
2820sub rev_map_get {
2821        my ($self, $rev, $uuid) = @_;
2822        my $map_path = $self->map_path($uuid);
2823        return undef unless -e $map_path;
2824
2825        sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
2826        binmode $fh or croak "binmode: $!";
2827        my $size = (stat($fh))[7];
2828        ($size % 24) == 0 or croak "inconsistent size: $size";
2829
2830        if ($size == 0) {
2831                close $fh or croak "close: $fh";
2832                return undef;
2833        }
2834
2835        my ($l, $u) = (0, $size - 24);
2836        my ($r, $c, $buf);
2837
2838        while ($l <= $u) {
2839                my $i = int(($l/24 + $u/24) / 2) * 24;
2840                sysseek($fh, $i, SEEK_SET) or croak "seek: $!";
2841                sysread($fh, my $buf, 24) == 24 or croak "read: $!";
2842                my ($r, $c) = unpack('NH40', $buf);
2843
2844                if ($r < $rev) {
2845                        $l = $i + 24;
2846                } elsif ($r > $rev) {
2847                        $u = $i - 24;
2848                } else { # $r == $rev
2849                        close($fh) or croak "close: $!";
2850                        return $c eq ('0' x 40) ? undef : $c;
2851                }
2852        }
2853        close($fh) or croak "close: $!";
2854        undef;
2855}
2856
2857# Finds the first svn revision that exists on (if $eq_ok is true) or
2858# before $rev for the current branch.  It will not search any lower
2859# than $min_rev.  Returns the git commit hash and svn revision number
2860# if found, else (undef, undef).
2861sub find_rev_before {
2862        my ($self, $rev, $eq_ok, $min_rev) = @_;
2863        --$rev unless $eq_ok;
2864        $min_rev ||= 1;
2865        while ($rev >= $min_rev) {
2866                if (my $c = $self->rev_map_get($rev)) {
2867                        return ($rev, $c);
2868                }
2869                --$rev;
2870        }
2871        return (undef, undef);
2872}
2873
2874# Finds the first svn revision that exists on (if $eq_ok is true) or
2875# after $rev for the current branch.  It will not search any higher
2876# than $max_rev.  Returns the git commit hash and svn revision number
2877# if found, else (undef, undef).
2878sub find_rev_after {
2879        my ($self, $rev, $eq_ok, $max_rev) = @_;
2880        ++$rev unless $eq_ok;
2881        $max_rev ||= $self->rev_map_max;
2882        while ($rev <= $max_rev) {
2883                if (my $c = $self->rev_map_get($rev)) {
2884                        return ($rev, $c);
2885                }
2886                ++$rev;
2887        }
2888        return (undef, undef);
2889}
2890
2891sub _new {
2892        my ($class, $repo_id, $ref_id, $path) = @_;
2893        unless (defined $repo_id && length $repo_id) {
2894                $repo_id = $Git::SVN::default_repo_id;
2895        }
2896        unless (defined $ref_id && length $ref_id) {
2897                $_[2] = $ref_id = $Git::SVN::default_ref_id;
2898        }
2899        $_[1] = $repo_id;
2900        my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
2901        $_[3] = $path = '' unless (defined $path);
2902        mkpath(["$ENV{GIT_DIR}/svn"]);
2903        bless {
2904                ref_id => $ref_id, dir => $dir, index => "$dir/index",
2905                path => $path, config => "$ENV{GIT_DIR}/svn/config",
2906                map_root => "$dir/.rev_map", repo_id => $repo_id }, $class;
2907}
2908
2909# for read-only access of old .rev_db formats
2910sub unlink_rev_db_symlink {
2911        my ($self) = @_;
2912        my $link = $self->rev_db_path;
2913        $link =~ s/\.[\w-]+$// or croak "missing UUID at the end of $link";
2914        if (-l $link) {
2915                unlink $link or croak "unlink: $link failed!";
2916        }
2917}
2918
2919sub rev_db_path {
2920        my ($self, $uuid) = @_;
2921        my $db_path = $self->map_path($uuid);
2922        $db_path =~ s{/\.rev_map\.}{/\.rev_db\.}
2923            or croak "map_path: $db_path does not contain '/.rev_map.' !";
2924        $db_path;
2925}
2926
2927# the new replacement for .rev_db
2928sub map_path {
2929        my ($self, $uuid) = @_;
2930        $uuid ||= $self->ra_uuid;
2931        "$self->{map_root}.$uuid";
2932}
2933
2934sub uri_encode {
2935        my ($f) = @_;
2936        $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
2937        $f
2938}
2939
2940sub remove_username {
2941        $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
2942}
2943
2944package Git::SVN::Prompt;
2945use strict;
2946use warnings;
2947require SVN::Core;
2948use vars qw/$_no_auth_cache $_username/;
2949
2950sub simple {
2951        my ($cred, $realm, $default_username, $may_save, $pool) = @_;
2952        $may_save = undef if $_no_auth_cache;
2953        $default_username = $_username if defined $_username;
2954        if (defined $default_username && length $default_username) {
2955                if (defined $realm && length $realm) {
2956                        print STDERR "Authentication realm: $realm\n";
2957                        STDERR->flush;
2958                }
2959                $cred->username($default_username);
2960        } else {
2961                username($cred, $realm, $may_save, $pool);
2962        }
2963        $cred->password(_read_password("Password for '" .
2964                                       $cred->username . "': ", $realm));
2965        $cred->may_save($may_save);
2966        $SVN::_Core::SVN_NO_ERROR;
2967}
2968
2969sub ssl_server_trust {
2970        my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
2971        $may_save = undef if $_no_auth_cache;
2972        print STDERR "Error validating server certificate for '$realm':\n";
2973        {
2974                no warnings 'once';
2975                # All variables SVN::Auth::SSL::* are used only once,
2976                # so we're shutting up Perl warnings about this.
2977                if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
2978                        print STDERR " - The certificate is not issued ",
2979                            "by a trusted authority. Use the\n",
2980                            "   fingerprint to validate ",
2981                            "the certificate manually!\n";
2982                }
2983                if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
2984                        print STDERR " - The certificate hostname ",
2985                            "does not match.\n";
2986                }
2987                if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
2988                        print STDERR " - The certificate is not yet valid.\n";
2989                }
2990                if ($failures & $SVN::Auth::SSL::EXPIRED) {
2991                        print STDERR " - The certificate has expired.\n";
2992                }
2993                if ($failures & $SVN::Auth::SSL::OTHER) {
2994                        print STDERR " - The certificate has ",
2995                            "an unknown error.\n";
2996                }
2997        } # no warnings 'once'
2998        printf STDERR
2999                "Certificate information:\n".
3000                " - Hostname: %s\n".
3001                " - Valid: from %s until %s\n".
3002                " - Issuer: %s\n".
3003                " - Fingerprint: %s\n",
3004                map $cert_info->$_, qw(hostname valid_from valid_until
3005                                       issuer_dname fingerprint);
3006        my $choice;
3007prompt:
3008        print STDERR $may_save ?
3009              "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
3010              "(R)eject or accept (t)emporarily? ";
3011        STDERR->flush;
3012        $choice = lc(substr(<STDIN> || 'R', 0, 1));
3013        if ($choice =~ /^t$/i) {
3014                $cred->may_save(undef);
3015        } elsif ($choice =~ /^r$/i) {
3016                return -1;
3017        } elsif ($may_save && $choice =~ /^p$/i) {
3018                $cred->may_save($may_save);
3019        } else {
3020                goto prompt;
3021        }
3022        $cred->accepted_failures($failures);
3023        $SVN::_Core::SVN_NO_ERROR;
3024}
3025
3026sub ssl_client_cert {
3027        my ($cred, $realm, $may_save, $pool) = @_;
3028        $may_save = undef if $_no_auth_cache;
3029        print STDERR "Client certificate filename: ";
3030        STDERR->flush;
3031        chomp(my $filename = <STDIN>);
3032        $cred->cert_file($filename);
3033        $cred->may_save($may_save);
3034        $SVN::_Core::SVN_NO_ERROR;
3035}
3036
3037sub ssl_client_cert_pw {
3038        my ($cred, $realm, $may_save, $pool) = @_;
3039        $may_save = undef if $_no_auth_cache;
3040        $cred->password(_read_password("Password: ", $realm));
3041        $cred->may_save($may_save);
3042        $SVN::_Core::SVN_NO_ERROR;
3043}
3044
3045sub username {
3046        my ($cred, $realm, $may_save, $pool) = @_;
3047        $may_save = undef if $_no_auth_cache;
3048        if (defined $realm && length $realm) {
3049                print STDERR "Authentication realm: $realm\n";
3050        }
3051        my $username;
3052        if (defined $_username) {
3053                $username = $_username;
3054        } else {
3055                print STDERR "Username: ";
3056                STDERR->flush;
3057                chomp($username = <STDIN>);
3058        }
3059        $cred->username($username);
3060        $cred->may_save($may_save);
3061        $SVN::_Core::SVN_NO_ERROR;
3062}
3063
3064sub _read_password {
3065        my ($prompt, $realm) = @_;
3066        print STDERR $prompt;
3067        STDERR->flush;
3068        require Term::ReadKey;
3069        Term::ReadKey::ReadMode('noecho');
3070        my $password = '';
3071        while (defined(my $key = Term::ReadKey::ReadKey(0))) {
3072                last if $key =~ /[\012\015]/; # \n\r
3073                $password .= $key;
3074        }
3075        Term::ReadKey::ReadMode('restore');
3076        print STDERR "\n";
3077        STDERR->flush;
3078        $password;
3079}
3080
3081package SVN::Git::Fetcher;
3082use vars qw/@ISA/;
3083use strict;
3084use warnings;
3085use Carp qw/croak/;
3086use File::Temp qw/tempfile/;
3087use IO::File qw//;
3088
3089# file baton members: path, mode_a, mode_b, pool, fh, blob, base
3090sub new {
3091        my ($class, $git_svn) = @_;
3092        my $self = SVN::Delta::Editor->new;
3093        bless $self, $class;
3094        $self->{c} = $git_svn->{last_commit} if exists $git_svn->{last_commit};
3095        $self->{empty} = {};
3096        $self->{dir_prop} = {};
3097        $self->{file_prop} = {};
3098        $self->{absent_dir} = {};
3099        $self->{absent_file} = {};
3100        $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
3101        $self;
3102}
3103
3104sub set_path_strip {
3105        my ($self, $path) = @_;
3106        $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
3107}
3108
3109sub open_root {
3110        { path => '' };
3111}
3112
3113sub open_directory {
3114        my ($self, $path, $pb, $rev) = @_;
3115        { path => $path };
3116}
3117
3118sub git_path {
3119        my ($self, $path) = @_;
3120        if ($self->{path_strip}) {
3121                $path =~ s!$self->{path_strip}!! or
3122                  die "Failed to strip path '$path' ($self->{path_strip})\n";
3123        }
3124        $path;
3125}
3126
3127sub delete_entry {
3128        my ($self, $path, $rev, $pb) = @_;
3129
3130        my $gpath = $self->git_path($path);
3131        return undef if ($gpath eq '');
3132
3133        # remove entire directories.
3134        if (command('ls-tree', $self->{c}, '--', $gpath) =~ /^040000 tree/) {
3135                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3136                                                     -r --name-only -z/,
3137                                                     $self->{c}, '--', $gpath);
3138                local $/ = "\0";
3139                while (<$ls>) {
3140                        chomp;
3141                        $self->{gii}->remove($_);
3142                        print "\tD\t$_\n" unless $::_q;
3143                }
3144                print "\tD\t$gpath/\n" unless $::_q;
3145                command_close_pipe($ls, $ctx);
3146                $self->{empty}->{$path} = 0
3147        } else {
3148                $self->{gii}->remove($gpath);
3149                print "\tD\t$gpath\n" unless $::_q;
3150        }
3151        undef;
3152}
3153
3154sub open_file {
3155        my ($self, $path, $pb, $rev) = @_;
3156        my $gpath = $self->git_path($path);
3157        my ($mode, $blob) = (command('ls-tree', $self->{c}, '--', $gpath)
3158                             =~ /^(\d{6}) blob ([a-f\d]{40})\t/);
3159        unless (defined $mode && defined $blob) {
3160                die "$path was not found in commit $self->{c} (r$rev)\n";
3161        }
3162        { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
3163          pool => SVN::Pool->new, action => 'M' };
3164}
3165
3166sub add_file {
3167        my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
3168        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3169        delete $self->{empty}->{$dir};
3170        { path => $path, mode_a => 100644, mode_b => 100644,
3171          pool => SVN::Pool->new, action => 'A' };
3172}
3173
3174sub add_directory {
3175        my ($self, $path, $cp_path, $cp_rev) = @_;
3176        my $gpath = $self->git_path($path);
3177        if ($gpath eq '') {
3178                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3179                                                     -r --name-only -z/,
3180                                                     $self->{c});
3181                local $/ = "\0";
3182                while (<$ls>) {
3183                        chomp;
3184                        $self->{gii}->remove($_);
3185                        print "\tD\t$_\n" unless $::_q;
3186                }
3187                command_close_pipe($ls, $ctx);
3188                $self->{empty}->{$path} = 0;
3189        }
3190        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3191        delete $self->{empty}->{$dir};
3192        $self->{empty}->{$path} = 1;
3193        { path => $path };
3194}
3195
3196sub change_dir_prop {
3197        my ($self, $db, $prop, $value) = @_;
3198        $self->{dir_prop}->{$db->{path}} ||= {};
3199        $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
3200        undef;
3201}
3202
3203sub absent_directory {
3204        my ($self, $path, $pb) = @_;
3205        $self->{absent_dir}->{$pb->{path}} ||= [];
3206        push @{$self->{absent_dir}->{$pb->{path}}}, $path;
3207        undef;
3208}
3209
3210sub absent_file {
3211        my ($self, $path, $pb) = @_;
3212        $self->{absent_file}->{$pb->{path}} ||= [];
3213        push @{$self->{absent_file}->{$pb->{path}}}, $path;
3214        undef;
3215}
3216
3217sub change_file_prop {
3218        my ($self, $fb, $prop, $value) = @_;
3219        if ($prop eq 'svn:executable') {
3220                if ($fb->{mode_b} != 120000) {
3221                        $fb->{mode_b} = defined $value ? 100755 : 100644;
3222                }
3223        } elsif ($prop eq 'svn:special') {
3224                $fb->{mode_b} = defined $value ? 120000 : 100644;
3225        } else {
3226                $self->{file_prop}->{$fb->{path}} ||= {};
3227                $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
3228        }
3229        undef;
3230}
3231
3232sub apply_textdelta {
3233        my ($self, $fb, $exp) = @_;
3234        my $fh = Git::temp_acquire('svn_delta');
3235        # $fh gets auto-closed() by SVN::TxDelta::apply(),
3236        # (but $base does not,) so dup() it for reading in close_file
3237        open my $dup, '<&', $fh or croak $!;
3238        my $base = Git::temp_acquire('git_blob');
3239        if ($fb->{blob}) {
3240                print $base 'link ' if ($fb->{mode_a} == 120000);
3241                my $size = $::_repository->cat_blob($fb->{blob}, $base);
3242                die "Failed to read object $fb->{blob}" if ($size < 0);
3243
3244                if (defined $exp) {
3245                        seek $base, 0, 0 or croak $!;
3246                        my $got = ::md5sum($base);
3247                        die "Checksum mismatch: $fb->{path} $fb->{blob}\n",
3248                            "expected: $exp\n",
3249                            "     got: $got\n" if ($got ne $exp);
3250                }
3251        }
3252        seek $base, 0, 0 or croak $!;
3253        $fb->{fh} = $fh;
3254        $fb->{base} = $base;
3255        [ SVN::TxDelta::apply($base, $dup, undef, $fb->{path}, $fb->{pool}) ];
3256}
3257
3258sub close_file {
3259        my ($self, $fb, $exp) = @_;
3260        my $hash;
3261        my $path = $self->git_path($fb->{path});
3262        if (my $fh = $fb->{fh}) {
3263                if (defined $exp) {
3264                        seek($fh, 0, 0) or croak $!;
3265                        my $got = ::md5sum($fh);
3266                        if ($got ne $exp) {
3267                                die "Checksum mismatch: $path\n",
3268                                    "expected: $exp\n    got: $got\n";
3269                        }
3270                }
3271                if ($fb->{mode_b} == 120000) {
3272                        sysseek($fh, 0, 0) or croak $!;
3273                        sysread($fh, my $buf, 5) == 5 or croak $!;
3274
3275                        unless ($buf eq 'link ') {
3276                                warn "$path has mode 120000",
3277                                                " but is not a link\n";
3278                        } else {
3279                                my $tmp_fh = Git::temp_acquire('svn_hash');
3280                                my $res;
3281                                while ($res = sysread($fh, my $str, 1024)) {
3282                                        my $out = syswrite($tmp_fh, $str, $res);
3283                                        defined($out) && $out == $res
3284                                                or croak("write ",
3285                                                        $tmp_fh->filename,
3286                                                        ": $!\n");
3287                                }
3288                                defined $res or croak $!;
3289
3290                                ($fh, $tmp_fh) = ($tmp_fh, $fh);
3291                                Git::temp_release($tmp_fh, 1);
3292                        }
3293                }
3294
3295                $hash = $::_repository->hash_and_insert_object(
3296                                $fh->filename);
3297                $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
3298
3299                Git::temp_release($fb->{base}, 1);
3300                Git::temp_release($fh, 1);
3301        } else {
3302                $hash = $fb->{blob} or die "no blob information\n";
3303        }
3304        $fb->{pool}->clear;
3305        $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
3306        print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
3307        undef;
3308}
3309
3310sub abort_edit {
3311        my $self = shift;
3312        $self->{nr} = $self->{gii}->{nr};
3313        delete $self->{gii};
3314        $self->SUPER::abort_edit(@_);
3315}
3316
3317sub close_edit {
3318        my $self = shift;
3319        $self->{git_commit_ok} = 1;
3320        $self->{nr} = $self->{gii}->{nr};
3321        delete $self->{gii};
3322        $self->SUPER::close_edit(@_);
3323}
3324
3325package SVN::Git::Editor;
3326use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
3327use strict;
3328use warnings;
3329use Carp qw/croak/;
3330use IO::File;
3331
3332sub new {
3333        my ($class, $opts) = @_;
3334        foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
3335                die "$_ required!\n" unless (defined $opts->{$_});
3336        }
3337
3338        my $pool = SVN::Pool->new;
3339        my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
3340        my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
3341                                     $opts->{r}, $mods);
3342
3343        # $opts->{ra} functions should not be used after this:
3344        my @ce  = $opts->{ra}->get_commit_editor($opts->{log},
3345                                                $opts->{editor_cb}, $pool);
3346        my $self = SVN::Delta::Editor->new(@ce, $pool);
3347        bless $self, $class;
3348        foreach (qw/svn_path r tree_a tree_b/) {
3349                $self->{$_} = $opts->{$_};
3350        }
3351        $self->{url} = $opts->{ra}->{url};
3352        $self->{mods} = $mods;
3353        $self->{types} = $types;
3354        $self->{pool} = $pool;
3355        $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
3356        $self->{rm} = { };
3357        $self->{path_prefix} = length $self->{svn_path} ?
3358                               "$self->{svn_path}/" : '';
3359        $self->{config} = $opts->{config};
3360        return $self;
3361}
3362
3363sub generate_diff {
3364        my ($tree_a, $tree_b) = @_;
3365        my @diff_tree = qw(diff-tree -z -r);
3366        if ($_cp_similarity) {
3367                push @diff_tree, "-C$_cp_similarity";
3368        } else {
3369                push @diff_tree, '-C';
3370        }
3371        push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
3372        push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
3373        push @diff_tree, $tree_a, $tree_b;
3374        my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
3375        local $/ = "\0";
3376        my $state = 'meta';
3377        my @mods;
3378        while (<$diff_fh>) {
3379                chomp $_; # this gets rid of the trailing "\0"
3380                if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
3381                                        $::sha1\s($::sha1)\s
3382                                        ([MTCRAD])\d*$/xo) {
3383                        push @mods, {   mode_a => $1, mode_b => $2,
3384                                        sha1_b => $3, chg => $4 };
3385                        if ($4 =~ /^(?:C|R)$/) {
3386                                $state = 'file_a';
3387                        } else {
3388                                $state = 'file_b';
3389                        }
3390                } elsif ($state eq 'file_a') {
3391                        my $x = $mods[$#mods] or croak "Empty array\n";
3392                        if ($x->{chg} !~ /^(?:C|R)$/) {
3393                                croak "Error parsing $_, $x->{chg}\n";
3394                        }
3395                        $x->{file_a} = $_;
3396                        $state = 'file_b';
3397                } elsif ($state eq 'file_b') {
3398                        my $x = $mods[$#mods] or croak "Empty array\n";
3399                        if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
3400                                croak "Error parsing $_, $x->{chg}\n";
3401                        }
3402                        if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
3403                                croak "Error parsing $_, $x->{chg}\n";
3404                        }
3405                        $x->{file_b} = $_;
3406                        $state = 'meta';
3407                } else {
3408                        croak "Error parsing $_\n";
3409                }
3410        }
3411        command_close_pipe($diff_fh, $ctx);
3412        \@mods;
3413}
3414
3415sub check_diff_paths {
3416        my ($ra, $pfx, $rev, $mods) = @_;
3417        my %types;
3418        $pfx .= '/' if length $pfx;
3419
3420        sub type_diff_paths {
3421                my ($ra, $types, $path, $rev) = @_;
3422                my @p = split m#/+#, $path;
3423                my $c = shift @p;
3424                unless (defined $types->{$c}) {
3425                        $types->{$c} = $ra->check_path($c, $rev);
3426                }
3427                while (@p) {
3428                        $c .= '/' . shift @p;
3429                        next if defined $types->{$c};
3430                        $types->{$c} = $ra->check_path($c, $rev);
3431                }
3432        }
3433
3434        foreach my $m (@$mods) {
3435                foreach my $f (qw/file_a file_b/) {
3436                        next unless defined $m->{$f};
3437                        my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
3438                        if (length $pfx.$dir && ! defined $types{$dir}) {
3439                                type_diff_paths($ra, \%types, $pfx.$dir, $rev);
3440                        }
3441                }
3442        }
3443        \%types;
3444}
3445
3446sub split_path {
3447        return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
3448}
3449
3450sub repo_path {
3451        my ($self, $path) = @_;
3452        $self->{path_prefix}.(defined $path ? $path : '');
3453}
3454
3455sub url_path {
3456        my ($self, $path) = @_;
3457        if ($self->{url} =~ m#^https?://#) {
3458                $path =~ s/([^a-zA-Z0-9_.-])/uc sprintf("%%%02x",ord($1))/eg;
3459        }
3460        $self->{url} . '/' . $self->repo_path($path);
3461}
3462
3463sub rmdirs {
3464        my ($self) = @_;
3465        my $rm = $self->{rm};
3466        delete $rm->{''}; # we never delete the url we're tracking
3467        return unless %$rm;
3468
3469        foreach (keys %$rm) {
3470                my @d = split m#/#, $_;
3471                my $c = shift @d;
3472                $rm->{$c} = 1;
3473                while (@d) {
3474                        $c .= '/' . shift @d;
3475                        $rm->{$c} = 1;
3476                }
3477        }
3478        delete $rm->{$self->{svn_path}};
3479        delete $rm->{''}; # we never delete the url we're tracking
3480        return unless %$rm;
3481
3482        my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
3483                                             $self->{tree_b});
3484        local $/ = "\0";
3485        while (<$fh>) {
3486                chomp;
3487                my @dn = split m#/#, $_;
3488                while (pop @dn) {
3489                        delete $rm->{join '/', @dn};
3490                }
3491                unless (%$rm) {
3492                        close $fh;
3493                        return;
3494                }
3495        }
3496        command_close_pipe($fh, $ctx);
3497
3498        my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
3499        foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
3500                $self->close_directory($bat->{$d}, $p);
3501                my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
3502                print "\tD+\t$d/\n" unless $::_q;
3503                $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
3504                delete $bat->{$d};
3505        }
3506}
3507
3508sub open_or_add_dir {
3509        my ($self, $full_path, $baton) = @_;
3510        my $t = $self->{types}->{$full_path};
3511        if (!defined $t) {
3512                die "$full_path not known in r$self->{r} or we have a bug!\n";
3513        }
3514        {
3515                no warnings 'once';
3516                # SVN::Node::none and SVN::Node::file are used only once,
3517                # so we're shutting up Perl's warnings about them.
3518                if ($t == $SVN::Node::none) {
3519                        return $self->add_directory($full_path, $baton,
3520                            undef, -1, $self->{pool});
3521                } elsif ($t == $SVN::Node::dir) {
3522                        return $self->open_directory($full_path, $baton,
3523                            $self->{r}, $self->{pool});
3524                } # no warnings 'once'
3525                print STDERR "$full_path already exists in repository at ",
3526                    "r$self->{r} and it is not a directory (",
3527                    ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
3528        } # no warnings 'once'
3529        exit 1;
3530}
3531
3532sub ensure_path {
3533        my ($self, $path) = @_;
3534        my $bat = $self->{bat};
3535        my $repo_path = $self->repo_path($path);
3536        return $bat->{''} unless (length $repo_path);
3537        my @p = split m#/+#, $repo_path;
3538        my $c = shift @p;
3539        $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
3540        while (@p) {
3541                my $c0 = $c;
3542                $c .= '/' . shift @p;
3543                $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
3544        }
3545        return $bat->{$c};
3546}
3547
3548# Subroutine to convert a globbing pattern to a regular expression.
3549# From perl cookbook.
3550sub glob2pat {
3551        my $globstr = shift;
3552        my %patmap = ('*' => '.*', '?' => '.', '[' => '[', ']' => ']');
3553        $globstr =~ s{(.)} { $patmap{$1} || "\Q$1" }ge;
3554        return '^' . $globstr . '$';
3555}
3556
3557sub check_autoprop {
3558        my ($self, $pattern, $properties, $file, $fbat) = @_;
3559        # Convert the globbing pattern to a regular expression.
3560        my $regex = glob2pat($pattern);
3561        # Check if the pattern matches the file name.
3562        if($file =~ m/($regex)/) {
3563                # Parse the list of properties to set.
3564                my @props = split(/;/, $properties);
3565                foreach my $prop (@props) {
3566                        # Parse 'name=value' syntax and set the property.
3567                        if ($prop =~ /([^=]+)=(.*)/) {
3568                                my ($n,$v) = ($1,$2);
3569                                for ($n, $v) {
3570                                        s/^\s+//; s/\s+$//;
3571                                }
3572                                $self->change_file_prop($fbat, $n, $v);
3573                        }
3574                }
3575        }
3576}
3577
3578sub apply_autoprops {
3579        my ($self, $file, $fbat) = @_;
3580        my $conf_t = ${$self->{config}}{'config'};
3581        no warnings 'once';
3582        # Check [miscellany]/enable-auto-props in svn configuration.
3583        if (SVN::_Core::svn_config_get_bool(
3584                $conf_t,
3585                $SVN::_Core::SVN_CONFIG_SECTION_MISCELLANY,
3586                $SVN::_Core::SVN_CONFIG_OPTION_ENABLE_AUTO_PROPS,
3587                0)) {
3588                # Auto-props are enabled.  Enumerate them to look for matches.
3589                my $callback = sub {
3590                        $self->check_autoprop($_[0], $_[1], $file, $fbat);
3591                };
3592                SVN::_Core::svn_config_enumerate(
3593                        $conf_t,
3594                        $SVN::_Core::SVN_CONFIG_SECTION_AUTO_PROPS,
3595                        $callback);
3596        }
3597}
3598
3599sub A {
3600        my ($self, $m) = @_;
3601        my ($dir, $file) = split_path($m->{file_b});
3602        my $pbat = $self->ensure_path($dir);
3603        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3604                                        undef, -1);
3605        print "\tA\t$m->{file_b}\n" unless $::_q;
3606        $self->apply_autoprops($file, $fbat);
3607        $self->chg_file($fbat, $m);
3608        $self->close_file($fbat,undef,$self->{pool});
3609}
3610
3611sub C {
3612        my ($self, $m) = @_;
3613        my ($dir, $file) = split_path($m->{file_b});
3614        my $pbat = $self->ensure_path($dir);
3615        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3616                                $self->url_path($m->{file_a}), $self->{r});
3617        print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
3618        $self->chg_file($fbat, $m);
3619        $self->close_file($fbat,undef,$self->{pool});
3620}
3621
3622sub delete_entry {
3623        my ($self, $path, $pbat) = @_;
3624        my $rpath = $self->repo_path($path);
3625        my ($dir, $file) = split_path($rpath);
3626        $self->{rm}->{$dir} = 1;
3627        $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
3628}
3629
3630sub R {
3631        my ($self, $m) = @_;
3632        my ($dir, $file) = split_path($m->{file_b});
3633        my $pbat = $self->ensure_path($dir);
3634        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3635                                $self->url_path($m->{file_a}), $self->{r});
3636        print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
3637        $self->chg_file($fbat, $m);
3638        $self->close_file($fbat,undef,$self->{pool});
3639
3640        ($dir, $file) = split_path($m->{file_a});
3641        $pbat = $self->ensure_path($dir);
3642        $self->delete_entry($m->{file_a}, $pbat);
3643}
3644
3645sub M {
3646        my ($self, $m) = @_;
3647        my ($dir, $file) = split_path($m->{file_b});
3648        my $pbat = $self->ensure_path($dir);
3649        my $fbat = $self->open_file($self->repo_path($m->{file_b}),
3650                                $pbat,$self->{r},$self->{pool});
3651        print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
3652        $self->chg_file($fbat, $m);
3653        $self->close_file($fbat,undef,$self->{pool});
3654}
3655
3656sub T { shift->M(@_) }
3657
3658sub change_file_prop {
3659        my ($self, $fbat, $pname, $pval) = @_;
3660        $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
3661}
3662
3663sub chg_file {
3664        my ($self, $fbat, $m) = @_;
3665        if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
3666                $self->change_file_prop($fbat,'svn:executable','*');
3667        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
3668                $self->change_file_prop($fbat,'svn:executable',undef);
3669        }
3670        my $fh = Git::temp_acquire('git_blob');
3671        if ($m->{mode_b} =~ /^120/) {
3672                print $fh 'link ' or croak $!;
3673                $self->change_file_prop($fbat,'svn:special','*');
3674        } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
3675                $self->change_file_prop($fbat,'svn:special',undef);
3676        }
3677        my $size = $::_repository->cat_blob($m->{sha1_b}, $fh);
3678        croak "Failed to read object $m->{sha1_b}" if ($size < 0);
3679        $fh->flush == 0 or croak $!;
3680        seek $fh, 0, 0 or croak $!;
3681
3682        my $exp = ::md5sum($fh);
3683        seek $fh, 0, 0 or croak $!;
3684
3685        my $pool = SVN::Pool->new;
3686        my $atd = $self->apply_textdelta($fbat, undef, $pool);
3687        my $got = SVN::TxDelta::send_stream($fh, @$atd, $pool);
3688        die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
3689        Git::temp_release($fh, 1);
3690        $pool->clear;
3691}
3692
3693sub D {
3694        my ($self, $m) = @_;
3695        my ($dir, $file) = split_path($m->{file_b});
3696        my $pbat = $self->ensure_path($dir);
3697        print "\tD\t$m->{file_b}\n" unless $::_q;
3698        $self->delete_entry($m->{file_b}, $pbat);
3699}
3700
3701sub close_edit {
3702        my ($self) = @_;
3703        my ($p,$bat) = ($self->{pool}, $self->{bat});
3704        foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
3705                next if $_ eq '';
3706                $self->close_directory($bat->{$_}, $p);
3707        }
3708        $self->close_directory($bat->{''}, $p);
3709        $self->SUPER::close_edit($p);
3710        $p->clear;
3711}
3712
3713sub abort_edit {
3714        my ($self) = @_;
3715        $self->SUPER::abort_edit($self->{pool});
3716}
3717
3718sub DESTROY {
3719        my $self = shift;
3720        $self->SUPER::DESTROY(@_);
3721        $self->{pool}->clear;
3722}
3723
3724# this drives the editor
3725sub apply_diff {
3726        my ($self) = @_;
3727        my $mods = $self->{mods};
3728        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
3729        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
3730                my $f = $m->{chg};
3731                if (defined $o{$f}) {
3732                        $self->$f($m);
3733                } else {
3734                        fatal("Invalid change type: $f");
3735                }
3736        }
3737        $self->rmdirs if $_rmdir;
3738        if (@$mods == 0) {
3739                $self->abort_edit;
3740        } else {
3741                $self->close_edit;
3742        }
3743        return scalar @$mods;
3744}
3745
3746package Git::SVN::Ra;
3747use vars qw/@ISA $config_dir $_log_window_size/;
3748use strict;
3749use warnings;
3750my ($ra_invalid, $can_do_switch, %ignored_err, $RA);
3751
3752BEGIN {
3753        # enforce temporary pool usage for some simple functions
3754        no strict 'refs';
3755        for my $f (qw/rev_proplist get_latest_revnum get_uuid get_repos_root/) {
3756                my $SUPER = "SUPER::$f";
3757                *$f = sub {
3758                        my $self = shift;
3759                        my $pool = SVN::Pool->new;
3760                        my @ret = $self->$SUPER(@_,$pool);
3761                        $pool->clear;
3762                        wantarray ? @ret : $ret[0];
3763                };
3764        }
3765}
3766
3767sub _auth_providers () {
3768        [
3769          SVN::Client::get_simple_provider(),
3770          SVN::Client::get_ssl_server_trust_file_provider(),
3771          SVN::Client::get_simple_prompt_provider(
3772            \&Git::SVN::Prompt::simple, 2),
3773          SVN::Client::get_ssl_client_cert_file_provider(),
3774          SVN::Client::get_ssl_client_cert_prompt_provider(
3775            \&Git::SVN::Prompt::ssl_client_cert, 2),
3776          SVN::Client::get_ssl_client_cert_pw_file_provider(),
3777          SVN::Client::get_ssl_client_cert_pw_prompt_provider(
3778            \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
3779          SVN::Client::get_username_provider(),
3780          SVN::Client::get_ssl_server_trust_prompt_provider(
3781            \&Git::SVN::Prompt::ssl_server_trust),
3782          SVN::Client::get_username_prompt_provider(
3783            \&Git::SVN::Prompt::username, 2)
3784        ]
3785}
3786
3787sub escape_uri_only {
3788        my ($uri) = @_;
3789        my @tmp;
3790        foreach (split m{/}, $uri) {
3791                s/([^\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
3792                push @tmp, $_;
3793        }
3794        join('/', @tmp);
3795}
3796
3797sub escape_url {
3798        my ($url) = @_;
3799        if ($url =~ m#^(https?)://([^/]+)(.*)$#) {
3800                my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
3801                $url = "$scheme://$domain$uri";
3802        }
3803        $url;
3804}
3805
3806sub new {
3807        my ($class, $url) = @_;
3808        $url =~ s!/+$!!;
3809        return $RA if ($RA && $RA->{url} eq $url);
3810
3811        SVN::_Core::svn_config_ensure($config_dir, undef);
3812        my ($baton, $callbacks) = SVN::Core::auth_open_helper(_auth_providers);
3813        my $config = SVN::Core::config_get_config($config_dir);
3814        $RA = undef;
3815        my $dont_store_passwords = 1;
3816        my $conf_t = ${$config}{'config'};
3817        {
3818                no warnings 'once';
3819                # The usage of $SVN::_Core::SVN_CONFIG_* variables
3820                # produces warnings that variables are used only once.
3821                # I had not found the better way to shut them up, so
3822                # the warnings of type 'once' are disabled in this block.
3823                if (SVN::_Core::svn_config_get_bool($conf_t,
3824                    $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
3825                    $SVN::_Core::SVN_CONFIG_OPTION_STORE_PASSWORDS,
3826                    1) == 0) {
3827                        SVN::_Core::svn_auth_set_parameter($baton,
3828                            $SVN::_Core::SVN_AUTH_PARAM_DONT_STORE_PASSWORDS,
3829                            bless (\$dont_store_passwords, "_p_void"));
3830                }
3831                if (SVN::_Core::svn_config_get_bool($conf_t,
3832                    $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
3833                    $SVN::_Core::SVN_CONFIG_OPTION_STORE_AUTH_CREDS,
3834                    1) == 0) {
3835                        $Git::SVN::Prompt::_no_auth_cache = 1;
3836                }
3837        } # no warnings 'once'
3838        my $self = SVN::Ra->new(url => escape_url($url), auth => $baton,
3839                              config => $config,
3840                              pool => SVN::Pool->new,
3841                              auth_provider_callbacks => $callbacks);
3842        $self->{url} = $url;
3843        $self->{svn_path} = $url;
3844        $self->{repos_root} = $self->get_repos_root;
3845        $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
3846        $self->{cache} = { check_path => { r => 0, data => {} },
3847                           get_dir => { r => 0, data => {} } };
3848        $RA = bless $self, $class;
3849}
3850
3851sub check_path {
3852        my ($self, $path, $r) = @_;
3853        my $cache = $self->{cache}->{check_path};
3854        if ($r == $cache->{r} && exists $cache->{data}->{$path}) {
3855                return $cache->{data}->{$path};
3856        }
3857        my $pool = SVN::Pool->new;
3858        my $t = $self->SUPER::check_path($path, $r, $pool);
3859        $pool->clear;
3860        if ($r != $cache->{r}) {
3861                %{$cache->{data}} = ();
3862                $cache->{r} = $r;
3863        }
3864        $cache->{data}->{$path} = $t;
3865}
3866
3867sub get_dir {
3868        my ($self, $dir, $r) = @_;
3869        my $cache = $self->{cache}->{get_dir};
3870        if ($r == $cache->{r}) {
3871                if (my $x = $cache->{data}->{$dir}) {
3872                        return wantarray ? @$x : $x->[0];
3873                }
3874        }
3875        my $pool = SVN::Pool->new;
3876        my ($d, undef, $props) = $self->SUPER::get_dir($dir, $r, $pool);
3877        my %dirents = map { $_ => { kind => $d->{$_}->kind } } keys %$d;
3878        $pool->clear;
3879        if ($r != $cache->{r}) {
3880                %{$cache->{data}} = ();
3881                $cache->{r} = $r;
3882        }
3883        $cache->{data}->{$dir} = [ \%dirents, $r, $props ];
3884        wantarray ? (\%dirents, $r, $props) : \%dirents;
3885}
3886
3887sub DESTROY {
3888        # do not call the real DESTROY since we store ourselves in $RA
3889}
3890
3891sub get_log {
3892        my ($self, @args) = @_;
3893        my $pool = SVN::Pool->new;
3894        splice(@args, 3, 1) if ($SVN::Core::VERSION le '1.2.0');
3895        my $ret = $self->SUPER::get_log(@args, $pool);
3896        $pool->clear;
3897        $ret;
3898}
3899
3900sub trees_match {
3901        my ($self, $url1, $rev1, $url2, $rev2) = @_;
3902        my $ctx = SVN::Client->new(auth => _auth_providers);
3903        my $out = IO::File->new_tmpfile;
3904
3905        # older SVN (1.1.x) doesn't take $pool as the last parameter for
3906        # $ctx->diff(), so we'll create a default one
3907        my $pool = SVN::Pool->new_default_sub;
3908
3909        $ra_invalid = 1; # this will open a new SVN::Ra connection to $url1
3910        $ctx->diff([], $url1, $rev1, $url2, $rev2, 1, 1, 0, $out, $out);
3911        $out->flush;
3912        my $ret = (($out->stat)[7] == 0);
3913        close $out or croak $!;
3914
3915        $ret;
3916}
3917
3918sub get_commit_editor {
3919        my ($self, $log, $cb, $pool) = @_;
3920        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
3921        $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
3922}
3923
3924sub gs_do_update {
3925        my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
3926        my $new = ($rev_a == $rev_b);
3927        my $path = $gs->{path};
3928
3929        if ($new && -e $gs->{index}) {
3930                unlink $gs->{index} or die
3931                  "Couldn't unlink index: $gs->{index}: $!\n";
3932        }
3933        my $pool = SVN::Pool->new;
3934        $editor->set_path_strip($path);
3935        my (@pc) = split m#/#, $path;
3936        my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
3937                                        1, $editor, $pool);
3938        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
3939
3940        # Since we can't rely on svn_ra_reparent being available, we'll
3941        # just have to do some magic with set_path to make it so
3942        # we only want a partial path.
3943        my $sp = '';
3944        my $final = join('/', @pc);
3945        while (@pc) {
3946                $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
3947                $sp .= '/' if length $sp;
3948                $sp .= shift @pc;
3949        }
3950        die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
3951
3952        $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
3953
3954        $reporter->finish_report($pool);
3955        $pool->clear;
3956        $editor->{git_commit_ok};
3957}
3958
3959# this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
3960# svn_ra_reparent didn't work before 1.4)
3961sub gs_do_switch {
3962        my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
3963        my $path = $gs->{path};
3964        my $pool = SVN::Pool->new;
3965
3966        my $full_url = $self->{url};
3967        my $old_url = $full_url;
3968        $full_url .= '/' . escape_uri_only($path) if length $path;
3969        my ($ra, $reparented);
3970        if ($old_url ne $full_url) {
3971                if ($old_url !~ m#^svn(\+ssh)?://#) {
3972                        SVN::_Ra::svn_ra_reparent($self->{session}, $full_url,
3973                                                  $pool);
3974                        $self->{url} = $full_url;
3975                        $reparented = 1;
3976                } else {
3977                        $_[0] = undef;
3978                        $self = undef;
3979                        $RA = undef;
3980                        $ra = Git::SVN::Ra->new($full_url);
3981                        $ra_invalid = 1;
3982                }
3983        }
3984        $ra ||= $self;
3985        my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
3986        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
3987        $reporter->set_path('', $rev_a, 0, @lock, $pool);
3988        $reporter->finish_report($pool);
3989
3990        if ($reparented) {
3991                SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
3992                $self->{url} = $old_url;
3993        }
3994
3995        $pool->clear;
3996        $editor->{git_commit_ok};
3997}
3998
3999sub longest_common_path {
4000        my ($gsv, $globs) = @_;
4001        my %common;
4002        my $common_max = scalar @$gsv;
4003
4004        foreach my $gs (@$gsv) {
4005                my @tmp = split m#/#, $gs->{path};
4006                my $p = '';
4007                foreach (@tmp) {
4008                        $p .= length($p) ? "/$_" : $_;
4009                        $common{$p} ||= 0;
4010                        $common{$p}++;
4011                }
4012        }
4013        $globs ||= [];
4014        $common_max += scalar @$globs;
4015        foreach my $glob (@$globs) {
4016                my @tmp = split m#/#, $glob->{path}->{left};
4017                my $p = '';
4018                foreach (@tmp) {
4019                        $p .= length($p) ? "/$_" : $_;
4020                        $common{$p} ||= 0;
4021                        $common{$p}++;
4022                }
4023        }
4024
4025        my $longest_path = '';
4026        foreach (sort {length $b <=> length $a} keys %common) {
4027                if ($common{$_} == $common_max) {
4028                        $longest_path = $_;
4029                        last;
4030                }
4031        }
4032        $longest_path;
4033}
4034
4035sub gs_fetch_loop_common {
4036        my ($self, $base, $head, $gsv, $globs) = @_;
4037        return if ($base > $head);
4038        my $inc = $_log_window_size;
4039        my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
4040        my $longest_path = longest_common_path($gsv, $globs);
4041        my $ra_url = $self->{url};
4042        while (1) {
4043                my %revs;
4044                my $err;
4045                my $err_handler = $SVN::Error::handler;
4046                $SVN::Error::handler = sub {
4047                        ($err) = @_;
4048                        skip_unknown_revs($err);
4049                };
4050                sub _cb {
4051                        my ($paths, $r, $author, $date, $log) = @_;
4052                        [ dup_changed_paths($paths),
4053                          { author => $author, date => $date, log => $log } ];
4054                }
4055                $self->get_log([$longest_path], $min, $max, 0, 1, 1,
4056                               sub { $revs{$_[1]} = _cb(@_) });
4057                if ($err && $max >= $head) {
4058                        print STDERR "Path '$longest_path' ",
4059                                     "was probably deleted:\n",
4060                                     $err->expanded_message,
4061                                     "\nWill attempt to follow ",
4062                                     "revisions r$min .. r$max ",
4063                                     "committed before the deletion\n";
4064                        my $hi = $max;
4065                        while (--$hi >= $min) {
4066                                my $ok;
4067                                $self->get_log([$longest_path], $min, $hi,
4068                                               0, 1, 1, sub {
4069                                               $ok ||= $_[1];
4070                                               $revs{$_[1]} = _cb(@_) });
4071                                if ($ok) {
4072                                        print STDERR "r$min .. r$ok OK\n";
4073                                        last;
4074                                }
4075                        }
4076                }
4077                $SVN::Error::handler = $err_handler;
4078
4079                my %exists = map { $_->{path} => $_ } @$gsv;
4080                foreach my $r (sort {$a <=> $b} keys %revs) {
4081                        my ($paths, $logged) = @{$revs{$r}};
4082
4083                        foreach my $gs ($self->match_globs(\%exists, $paths,
4084                                                           $globs, $r)) {
4085                                if ($gs->rev_map_max >= $r) {
4086                                        next;
4087                                }
4088                                next unless $gs->match_paths($paths, $r);
4089                                $gs->{logged_rev_props} = $logged;
4090                                if (my $last_commit = $gs->last_commit) {
4091                                        $gs->assert_index_clean($last_commit);
4092                                }
4093                                my $log_entry = $gs->do_fetch($paths, $r);
4094                                if ($log_entry) {
4095                                        $gs->do_git_commit($log_entry);
4096                                }
4097                                $INDEX_FILES{$gs->{index}} = 1;
4098                        }
4099                        foreach my $g (@$globs) {
4100                                my $k = "svn-remote.$g->{remote}." .
4101                                        "$g->{t}-maxRev";
4102                                Git::SVN::tmp_config($k, $r);
4103                        }
4104                        if ($ra_invalid) {
4105                                $_[0] = undef;
4106                                $self = undef;
4107                                $RA = undef;
4108                                $self = Git::SVN::Ra->new($ra_url);
4109                                $ra_invalid = undef;
4110                        }
4111                }
4112                # pre-fill the .rev_db since it'll eventually get filled in
4113                # with '0' x40 if something new gets committed
4114                foreach my $gs (@$gsv) {
4115                        next if $gs->rev_map_max >= $max;
4116                        next if defined $gs->rev_map_get($max);
4117                        $gs->rev_map_set($max, 0 x40);
4118                }
4119                foreach my $g (@$globs) {
4120                        my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
4121                        Git::SVN::tmp_config($k, $max);
4122                }
4123                last if $max >= $head;
4124                $min = $max + 1;
4125                $max += $inc;
4126                $max = $head if ($max > $head);
4127        }
4128        Git::SVN::gc();
4129}
4130
4131sub get_dir_globbed {
4132        my ($self, $left, $depth, $r) = @_;
4133
4134        my @x = eval { $self->get_dir($left, $r) };
4135        return unless scalar @x == 3;
4136        my $dirents = $x[0];
4137        my @finalents;
4138        foreach my $de (keys %$dirents) {
4139                next if $dirents->{$de}->{kind} != $SVN::Node::dir;
4140                if ($depth > 1) {
4141                        my @args = ("$left/$de", $depth - 1, $r);
4142                        foreach my $dir ($self->get_dir_globbed(@args)) {
4143                                push @finalents, "$de/$dir";
4144                        }
4145                } else {
4146                        push @finalents, $de;
4147                }
4148        }
4149        @finalents;
4150}
4151
4152sub match_globs {
4153        my ($self, $exists, $paths, $globs, $r) = @_;
4154
4155        sub get_dir_check {
4156                my ($self, $exists, $g, $r) = @_;
4157
4158                my @dirs = $self->get_dir_globbed($g->{path}->{left},
4159                                                  $g->{path}->{depth},
4160                                                  $r);
4161
4162                foreach my $de (@dirs) {
4163                        my $p = $g->{path}->full_path($de);
4164                        next if $exists->{$p};
4165                        next if (length $g->{path}->{right} &&
4166                                 ($self->check_path($p, $r) !=
4167                                  $SVN::Node::dir));
4168                        $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
4169                                         $g->{ref}->full_path($de), 1);
4170                }
4171        }
4172        foreach my $g (@$globs) {
4173                if (my $path = $paths->{"/$g->{path}->{left}"}) {
4174                        if ($path->{action} =~ /^[AR]$/) {
4175                                get_dir_check($self, $exists, $g, $r);
4176                        }
4177                }
4178                foreach (keys %$paths) {
4179                        if (/$g->{path}->{left_regex}/ &&
4180                            !/$g->{path}->{regex}/) {
4181                                next if $paths->{$_}->{action} !~ /^[AR]$/;
4182                                get_dir_check($self, $exists, $g, $r);
4183                        }
4184                        next unless /$g->{path}->{regex}/;
4185                        my $p = $1;
4186                        my $pathname = $g->{path}->full_path($p);
4187                        next if $exists->{$pathname};
4188                        next if ($self->check_path($pathname, $r) !=
4189                                 $SVN::Node::dir);
4190                        $exists->{$pathname} = Git::SVN->init(
4191                                              $self->{url}, $pathname, undef,
4192                                              $g->{ref}->full_path($p), 1);
4193                }
4194                my $c = '';
4195                foreach (split m#/#, $g->{path}->{left}) {
4196                        $c .= "/$_";
4197                        next unless ($paths->{$c} &&
4198                                     ($paths->{$c}->{action} =~ /^[AR]$/));
4199                        get_dir_check($self, $exists, $g, $r);
4200                }
4201        }
4202        values %$exists;
4203}
4204
4205sub minimize_url {
4206        my ($self) = @_;
4207        return $self->{url} if ($self->{url} eq $self->{repos_root});
4208        my $url = $self->{repos_root};
4209        my @components = split(m!/!, $self->{svn_path});
4210        my $c = '';
4211        do {
4212                $url .= "/$c" if length $c;
4213                eval { (ref $self)->new($url)->get_latest_revnum };
4214        } while ($@ && ($c = shift @components));
4215        $url;
4216}
4217
4218sub can_do_switch {
4219        my $self = shift;
4220        unless (defined $can_do_switch) {
4221                my $pool = SVN::Pool->new;
4222                my $rep = eval {
4223                        $self->do_switch(1, '', 0, $self->{url},
4224                                         SVN::Delta::Editor->new, $pool);
4225                };
4226                if ($@) {
4227                        $can_do_switch = 0;
4228                } else {
4229                        $rep->abort_report($pool);
4230                        $can_do_switch = 1;
4231                }
4232                $pool->clear;
4233        }
4234        $can_do_switch;
4235}
4236
4237sub skip_unknown_revs {
4238        my ($err) = @_;
4239        my $errno = $err->apr_err();
4240        # Maybe the branch we're tracking didn't
4241        # exist when the repo started, so it's
4242        # not an error if it doesn't, just continue
4243        #
4244        # Wonderfully consistent library, eh?
4245        # 160013 - svn:// and file://
4246        # 175002 - http(s)://
4247        # 175007 - http(s):// (this repo required authorization, too...)
4248        #   More codes may be discovered later...
4249        if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
4250                my $err_key = $err->expanded_message;
4251                # revision numbers change every time, filter them out
4252                $err_key =~ s/\d+/\0/g;
4253                $err_key = "$errno\0$err_key";
4254                unless ($ignored_err{$err_key}) {
4255                        warn "W: Ignoring error from SVN, path probably ",
4256                             "does not exist: ($errno): ",
4257                             $err->expanded_message,"\n";
4258                        warn "W: Do not be alarmed at the above message ",
4259                             "git-svn is just searching aggressively for ",
4260                             "old history.\n",
4261                             "This may take a while on large repositories\n";
4262                        $ignored_err{$err_key} = 1;
4263                }
4264                return;
4265        }
4266        die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
4267}
4268
4269# svn_log_changed_path_t objects passed to get_log are likely to be
4270# overwritten even if only the refs are copied to an external variable,
4271# so we should dup the structures in their entirety.  Using an externally
4272# passed pool (instead of our temporary and quickly cleared pool in
4273# Git::SVN::Ra) does not help matters at all...
4274sub dup_changed_paths {
4275        my ($paths) = @_;
4276        return undef unless $paths;
4277        my %ret;
4278        foreach my $p (keys %$paths) {
4279                my $i = $paths->{$p};
4280                my %s = map { $_ => $i->$_ }
4281                              qw/copyfrom_path copyfrom_rev action/;
4282                $ret{$p} = \%s;
4283        }
4284        \%ret;
4285}
4286
4287package Git::SVN::Log;
4288use strict;
4289use warnings;
4290use POSIX qw/strftime/;
4291use constant commit_log_separator => ('-' x 72) . "\n";
4292use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
4293            %rusers $show_commit $incremental/;
4294my $l_fmt;
4295
4296sub cmt_showable {
4297        my ($c) = @_;
4298        return 1 if defined $c->{r};
4299
4300        # big commit message got truncated by the 16k pretty buffer in rev-list
4301        if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
4302                                $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
4303                @{$c->{l}} = ();
4304                my @log = command(qw/cat-file commit/, $c->{c});
4305
4306                # shift off the headers
4307                shift @log while ($log[0] ne '');
4308                shift @log;
4309
4310                # TODO: make $c->{l} not have a trailing newline in the future
4311                @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
4312
4313                (undef, $c->{r}, undef) = ::extract_metadata(
4314                                (grep(/^git-svn-id: /, @log))[-1]);
4315        }
4316        return defined $c->{r};
4317}
4318
4319sub log_use_color {
4320        return $color || Git->repository->get_colorbool('color.diff');
4321}
4322
4323sub git_svn_log_cmd {
4324        my ($r_min, $r_max, @args) = @_;
4325        my $head = 'HEAD';
4326        my (@files, @log_opts);
4327        foreach my $x (@args) {
4328                if ($x eq '--' || @files) {
4329                        push @files, $x;
4330                } else {
4331                        if (::verify_ref("$x^0")) {
4332                                $head = $x;
4333                        } else {
4334                                push @log_opts, $x;
4335                        }
4336                }
4337        }
4338
4339        my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
4340        $gs ||= Git::SVN->_new;
4341        my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
4342                   $gs->refname);
4343        push @cmd, '-r' unless $non_recursive;
4344        push @cmd, qw/--raw --name-status/ if $verbose;
4345        push @cmd, '--color' if log_use_color();
4346        push @cmd, @log_opts;
4347        if (defined $r_max && $r_max == $r_min) {
4348                push @cmd, '--max-count=1';
4349                if (my $c = $gs->rev_map_get($r_max)) {
4350                        push @cmd, $c;
4351                }
4352        } elsif (defined $r_max) {
4353                if ($r_max < $r_min) {
4354                        ($r_min, $r_max) = ($r_max, $r_min);
4355                }
4356                my (undef, $c_max) = $gs->find_rev_before($r_max, 1, $r_min);
4357                my (undef, $c_min) = $gs->find_rev_after($r_min, 1, $r_max);
4358                # If there are no commits in the range, both $c_max and $c_min
4359                # will be undefined.  If there is at least 1 commit in the
4360                # range, both will be defined.
4361                return () if !defined $c_min || !defined $c_max;
4362                if ($c_min eq $c_max) {
4363                        push @cmd, '--max-count=1', $c_min;
4364                } else {
4365                        push @cmd, '--boundary', "$c_min..$c_max";
4366                }
4367        }
4368        return (@cmd, @files);
4369}
4370
4371# adapted from pager.c
4372sub config_pager {
4373        $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
4374        if (!defined $pager) {
4375                $pager = 'less';
4376        } elsif (length $pager == 0 || $pager eq 'cat') {
4377                $pager = undef;
4378        }
4379        $ENV{GIT_PAGER_IN_USE} = defined($pager);
4380}
4381
4382sub run_pager {
4383        return unless -t *STDOUT && defined $pager;
4384        pipe my $rfd, my $wfd or return;
4385        defined(my $pid = fork) or ::fatal "Can't fork: $!";
4386        if (!$pid) {
4387                open STDOUT, '>&', $wfd or
4388                                     ::fatal "Can't redirect to stdout: $!";
4389                return;
4390        }
4391        open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!";
4392        $ENV{LESS} ||= 'FRSX';
4393        exec $pager or ::fatal "Can't run pager: $! ($pager)";
4394}
4395
4396sub format_svn_date {
4397        return strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)", localtime(shift));
4398}
4399
4400sub parse_git_date {
4401        my ($t, $tz) = @_;
4402        # Date::Parse isn't in the standard Perl distro :(
4403        if ($tz =~ s/^\+//) {
4404                $t += tz_to_s_offset($tz);
4405        } elsif ($tz =~ s/^\-//) {
4406                $t -= tz_to_s_offset($tz);
4407        }
4408        return $t;
4409}
4410
4411sub set_local_timezone {
4412        if (defined $TZ) {
4413                $ENV{TZ} = $TZ;
4414        } else {
4415                delete $ENV{TZ};
4416        }
4417}
4418
4419sub tz_to_s_offset {
4420        my ($tz) = @_;
4421        $tz =~ s/(\d\d)$//;
4422        return ($1 * 60) + ($tz * 3600);
4423}
4424
4425sub get_author_info {
4426        my ($dest, $author, $t, $tz) = @_;
4427        $author =~ s/(?:^\s*|\s*$)//g;
4428        $dest->{a_raw} = $author;
4429        my $au;
4430        if ($::_authors) {
4431                $au = $rusers{$author} || undef;
4432        }
4433        if (!$au) {
4434                ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
4435        }
4436        $dest->{t} = $t;
4437        $dest->{tz} = $tz;
4438        $dest->{a} = $au;
4439        $dest->{t_utc} = parse_git_date($t, $tz);
4440}
4441
4442sub process_commit {
4443        my ($c, $r_min, $r_max, $defer) = @_;
4444        if (defined $r_min && defined $r_max) {
4445                if ($r_min == $c->{r} && $r_min == $r_max) {
4446                        show_commit($c);
4447                        return 0;
4448                }
4449                return 1 if $r_min == $r_max;
4450                if ($r_min < $r_max) {
4451                        # we need to reverse the print order
4452                        return 0 if (defined $limit && --$limit < 0);
4453                        push @$defer, $c;
4454                        return 1;
4455                }
4456                if ($r_min != $r_max) {
4457                        return 1 if ($r_min < $c->{r});
4458                        return 1 if ($r_max > $c->{r});
4459                }
4460        }
4461        return 0 if (defined $limit && --$limit < 0);
4462        show_commit($c);
4463        return 1;
4464}
4465
4466sub show_commit {
4467        my $c = shift;
4468        if ($oneline) {
4469                my $x = "\n";
4470                if (my $l = $c->{l}) {
4471                        while ($l->[0] =~ /^\s*$/) { shift @$l }
4472                        $x = $l->[0];
4473                }
4474                $l_fmt ||= 'A' . length($c->{r});
4475                print 'r',pack($l_fmt, $c->{r}),' | ';
4476                print "$c->{c} | " if $show_commit;
4477                print $x;
4478        } else {
4479                show_commit_normal($c);
4480        }
4481}
4482
4483sub show_commit_changed_paths {
4484        my ($c) = @_;
4485        return unless $c->{changed};
4486        print "Changed paths:\n", @{$c->{changed}};
4487}
4488
4489sub show_commit_normal {
4490        my ($c) = @_;
4491        print commit_log_separator, "r$c->{r} | ";
4492        print "$c->{c} | " if $show_commit;
4493        print "$c->{a} | ", format_svn_date($c->{t_utc}), ' | ';
4494        my $nr_line = 0;
4495
4496        if (my $l = $c->{l}) {
4497                while ($l->[$#$l] eq "\n" && $#$l > 0
4498                                          && $l->[($#$l - 1)] eq "\n") {
4499                        pop @$l;
4500                }
4501                $nr_line = scalar @$l;
4502                if (!$nr_line) {
4503                        print "1 line\n\n\n";
4504                } else {
4505                        if ($nr_line == 1) {
4506                                $nr_line = '1 line';
4507                        } else {
4508                                $nr_line .= ' lines';
4509                        }
4510                        print $nr_line, "\n";
4511                        show_commit_changed_paths($c);
4512                        print "\n";
4513                        print $_ foreach @$l;
4514                }
4515        } else {
4516                print "1 line\n";
4517                show_commit_changed_paths($c);
4518                print "\n";
4519
4520        }
4521        foreach my $x (qw/raw stat diff/) {
4522                if ($c->{$x}) {
4523                        print "\n";
4524                        print $_ foreach @{$c->{$x}}
4525                }
4526        }
4527}
4528
4529sub cmd_show_log {
4530        my (@args) = @_;
4531        my ($r_min, $r_max);
4532        my $r_last = -1; # prevent dupes
4533        set_local_timezone();
4534        if (defined $::_revision) {
4535                if ($::_revision =~ /^(\d+):(\d+)$/) {
4536                        ($r_min, $r_max) = ($1, $2);
4537                } elsif ($::_revision =~ /^\d+$/) {
4538                        $r_min = $r_max = $::_revision;
4539                } else {
4540                        ::fatal "-r$::_revision is not supported, use ",
4541                                "standard 'git log' arguments instead";
4542                }
4543        }
4544
4545        config_pager();
4546        @args = git_svn_log_cmd($r_min, $r_max, @args);
4547        if (!@args) {
4548                print commit_log_separator unless $incremental || $oneline;
4549                return;
4550        }
4551        my $log = command_output_pipe(@args);
4552        run_pager();
4553        my (@k, $c, $d, $stat);
4554        my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
4555        while (<$log>) {
4556                if (/^${esc_color}commit -?($::sha1_short)/o) {
4557                        my $cmt = $1;
4558                        if ($c && cmt_showable($c) && $c->{r} != $r_last) {
4559                                $r_last = $c->{r};
4560                                process_commit($c, $r_min, $r_max, \@k) or
4561                                                                goto out;
4562                        }
4563                        $d = undef;
4564                        $c = { c => $cmt };
4565                } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
4566                        get_author_info($c, $1, $2, $3);
4567                } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
4568                        # ignore
4569                } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
4570                        push @{$c->{raw}}, $_;
4571                } elsif (/^${esc_color}[ACRMDT]\t/) {
4572                        # we could add $SVN->{svn_path} here, but that requires
4573                        # remote access at the moment (repo_path_split)...
4574                        s#^(${esc_color})([ACRMDT])\t#$1   $2 #o;
4575                        push @{$c->{changed}}, $_;
4576                } elsif (/^${esc_color}diff /o) {
4577                        $d = 1;
4578                        push @{$c->{diff}}, $_;
4579                } elsif ($d) {
4580                        push @{$c->{diff}}, $_;
4581                } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
4582                          $esc_color*[\+\-]*$esc_color$/x) {
4583                        $stat = 1;
4584                        push @{$c->{stat}}, $_;
4585                } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
4586                        push @{$c->{stat}}, $_;
4587                        $stat = undef;
4588                } elsif (/^${esc_color}    (git-svn-id:.+)$/o) {
4589                        ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
4590                } elsif (s/^${esc_color}    //o) {
4591                        push @{$c->{l}}, $_;
4592                }
4593        }
4594        if ($c && defined $c->{r} && $c->{r} != $r_last) {
4595                $r_last = $c->{r};
4596                process_commit($c, $r_min, $r_max, \@k);
4597        }
4598        if (@k) {
4599                ($r_min, $r_max) = ($r_max, $r_min);
4600                process_commit($_, $r_min, $r_max) foreach reverse @k;
4601        }
4602out:
4603        close $log;
4604        print commit_log_separator unless $incremental || $oneline;
4605}
4606
4607sub cmd_blame {
4608        my $path = pop;
4609
4610        config_pager();
4611        run_pager();
4612
4613        my ($fh, $ctx, $rev);
4614
4615        if ($_git_format) {
4616                ($fh, $ctx) = command_output_pipe('blame', @_, $path);
4617                while (my $line = <$fh>) {
4618                        if ($line =~ /^\^?([[:xdigit:]]+)\s/) {
4619                                # Uncommitted edits show up as a rev ID of
4620                                # all zeros, which we can't look up with
4621                                # cmt_metadata
4622                                if ($1 !~ /^0+$/) {
4623                                        (undef, $rev, undef) =
4624                                                ::cmt_metadata($1);
4625                                        $rev = '0' if (!$rev);
4626                                } else {
4627                                        $rev = '0';
4628                                }
4629                                $rev = sprintf('%-10s', $rev);
4630                                $line =~ s/^\^?[[:xdigit:]]+(\s)/$rev$1/;
4631                        }
4632                        print $line;
4633                }
4634        } else {
4635                ($fh, $ctx) = command_output_pipe('blame', '-p', @_, 'HEAD',
4636                                                  '--', $path);
4637                my ($sha1);
4638                my %authors;
4639                while (my $line = <$fh>) {
4640                        if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
4641                                $sha1 = $1;
4642                                (undef, $rev, undef) = ::cmt_metadata($1);
4643                                $rev = '0' if (!$rev);
4644                        }
4645                        elsif ($line =~ /^author (.*)/) {
4646                                $authors{$rev} = $1;
4647                                $authors{$rev} =~ s/\s/_/g;
4648                        }
4649                        elsif ($line =~ /^\t(.*)$/) {
4650                                printf("%6s %10s %s\n", $rev, $authors{$rev}, $1);
4651                        }
4652                }
4653        }
4654        command_close_pipe($fh, $ctx);
4655}
4656
4657package Git::SVN::Migration;
4658# these version numbers do NOT correspond to actual version numbers
4659# of git nor git-svn.  They are just relative.
4660#
4661# v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
4662#
4663# v1 layout: .git/$id/info/url, refs/remotes/$id
4664#
4665# v2 layout: .git/svn/$id/info/url, refs/remotes/$id
4666#
4667# v3 layout: .git/svn/$id, refs/remotes/$id
4668#            - info/url may remain for backwards compatibility
4669#            - this is what we migrate up to this layout automatically,
4670#            - this will be used by git svn init on single branches
4671# v3.1 layout (auto migrated):
4672#            - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
4673#              for backwards compatibility
4674#
4675# v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
4676#            - this is only created for newly multi-init-ed
4677#              repositories.  Similar in spirit to the
4678#              --use-separate-remotes option in git-clone (now default)
4679#            - we do not automatically migrate to this (following
4680#              the example set by core git)
4681#
4682# v5 layout: .rev_db.$UUID => .rev_map.$UUID
4683#            - newer, more-efficient format that uses 24-bytes per record
4684#              with no filler space.
4685#            - use xxd -c24 < .rev_map.$UUID to view and debug
4686#            - This is a one-way migration, repositories updated to the
4687#              new format will not be able to use old git-svn without
4688#              rebuilding the .rev_db.  Rebuilding the rev_db is not
4689#              possible if noMetadata or useSvmProps are set; but should
4690#              be no problem for users that use the (sensible) defaults.
4691use strict;
4692use warnings;
4693use Carp qw/croak/;
4694use File::Path qw/mkpath/;
4695use File::Basename qw/dirname basename/;
4696use vars qw/$_minimize/;
4697
4698sub migrate_from_v0 {
4699        my $git_dir = $ENV{GIT_DIR};
4700        return undef unless -d $git_dir;
4701        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
4702        my $migrated = 0;
4703        while (<$fh>) {
4704                chomp;
4705                my ($id, $orig_ref) = ($_, $_);
4706                next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
4707                next unless -f "$git_dir/$id/info/url";
4708                my $new_ref = "refs/remotes/$id";
4709                if (::verify_ref("$new_ref^0")) {
4710                        print STDERR "W: $orig_ref is probably an old ",
4711                                     "branch used by an ancient version of ",
4712                                     "git-svn.\n",
4713                                     "However, $new_ref also exists.\n",
4714                                     "We will not be able ",
4715                                     "to use this branch until this ",
4716                                     "ambiguity is resolved.\n";
4717                        next;
4718                }
4719                print STDERR "Migrating from v0 layout...\n" if !$migrated;
4720                print STDERR "Renaming ref: $orig_ref => $new_ref\n";
4721                command_noisy('update-ref', $new_ref, $orig_ref);
4722                command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
4723                $migrated++;
4724        }
4725        command_close_pipe($fh, $ctx);
4726        print STDERR "Done migrating from v0 layout...\n" if $migrated;
4727        $migrated;
4728}
4729
4730sub migrate_from_v1 {
4731        my $git_dir = $ENV{GIT_DIR};
4732        my $migrated = 0;
4733        return $migrated unless -d $git_dir;
4734        my $svn_dir = "$git_dir/svn";
4735
4736        # just in case somebody used 'svn' as their $id at some point...
4737        return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
4738
4739        print STDERR "Migrating from a git-svn v1 layout...\n";
4740        mkpath([$svn_dir]);
4741        print STDERR "Data from a previous version of git-svn exists, but\n\t",
4742                     "$svn_dir\n\t(required for this version ",
4743                     "($::VERSION) of git-svn) does not exist.\n";
4744        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
4745        while (<$fh>) {
4746                my $x = $_;
4747                next unless $x =~ s#^refs/remotes/##;
4748                chomp $x;
4749                next unless -f "$git_dir/$x/info/url";
4750                my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
4751                next unless $u;
4752                my $dn = dirname("$git_dir/svn/$x");
4753                mkpath([$dn]) unless -d $dn;
4754                if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
4755                        mkpath(["$git_dir/svn/svn"]);
4756                        print STDERR " - $git_dir/$x/info => ",
4757                                        "$git_dir/svn/$x/info\n";
4758                        rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
4759                               croak "$!: $x";
4760                        # don't worry too much about these, they probably
4761                        # don't exist with repos this old (save for index,
4762                        # and we can easily regenerate that)
4763                        foreach my $f (qw/unhandled.log index .rev_db/) {
4764                                rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
4765                        }
4766                } else {
4767                        print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
4768                        rename "$git_dir/$x", "$git_dir/svn/$x" or
4769                               croak "$!: $x";
4770                }
4771                $migrated++;
4772        }
4773        command_close_pipe($fh, $ctx);
4774        print STDERR "Done migrating from a git-svn v1 layout\n";
4775        $migrated;
4776}
4777
4778sub read_old_urls {
4779        my ($l_map, $pfx, $path) = @_;
4780        my @dir;
4781        foreach (<$path/*>) {
4782                if (-r "$_/info/url") {
4783                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
4784                        my $ref_id = $pfx . basename $_;
4785                        my $url = ::file_to_s("$_/info/url");
4786                        $l_map->{$ref_id} = $url;
4787                } elsif (-d $_) {
4788                        push @dir, $_;
4789                }
4790        }
4791        foreach (@dir) {
4792                my $x = $_;
4793                $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
4794                read_old_urls($l_map, $x, $_);
4795        }
4796}
4797
4798sub migrate_from_v2 {
4799        my @cfg = command(qw/config -l/);
4800        return if grep /^svn-remote\..+\.url=/, @cfg;
4801        my %l_map;
4802        read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
4803        my $migrated = 0;
4804
4805        foreach my $ref_id (sort keys %l_map) {
4806                eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
4807                if ($@) {
4808                        Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
4809                }
4810                $migrated++;
4811        }
4812        $migrated;
4813}
4814
4815sub minimize_connections {
4816        my $r = Git::SVN::read_all_remotes();
4817        my $new_urls = {};
4818        my $root_repos = {};
4819        foreach my $repo_id (keys %$r) {
4820                my $url = $r->{$repo_id}->{url} or next;
4821                my $fetch = $r->{$repo_id}->{fetch} or next;
4822                my $ra = Git::SVN::Ra->new($url);
4823
4824                # skip existing cases where we already connect to the root
4825                if (($ra->{url} eq $ra->{repos_root}) ||
4826                    ($ra->{repos_root} eq $repo_id)) {
4827                        $root_repos->{$ra->{url}} = $repo_id;
4828                        next;
4829                }
4830
4831                my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
4832                my $root_path = $ra->{url};
4833                $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
4834                foreach my $path (keys %$fetch) {
4835                        my $ref_id = $fetch->{$path};
4836                        my $gs = Git::SVN->new($ref_id, $repo_id, $path);
4837
4838                        # make sure we can read when connecting to
4839                        # a higher level of a repository
4840                        my ($last_rev, undef) = $gs->last_rev_commit;
4841                        if (!defined $last_rev) {
4842                                $last_rev = eval {
4843                                        $root_ra->get_latest_revnum;
4844                                };
4845                                next if $@;
4846                        }
4847                        my $new = $root_path;
4848                        $new .= length $path ? "/$path" : '';
4849                        eval {
4850                                $root_ra->get_log([$new], $last_rev, $last_rev,
4851                                                  0, 0, 1, sub { });
4852                        };
4853                        next if $@;
4854                        $new_urls->{$ra->{repos_root}}->{$new} =
4855                                { ref_id => $ref_id,
4856                                  old_repo_id => $repo_id,
4857                                  old_path => $path };
4858                }
4859        }
4860
4861        my @emptied;
4862        foreach my $url (keys %$new_urls) {
4863                # see if we can re-use an existing [svn-remote "repo_id"]
4864                # instead of creating a(n ugly) new section:
4865                my $repo_id = $root_repos->{$url} || $url;
4866
4867                my $fetch = $new_urls->{$url};
4868                foreach my $path (keys %$fetch) {
4869                        my $x = $fetch->{$path};
4870                        Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
4871                        my $pfx = "svn-remote.$x->{old_repo_id}";
4872
4873                        my $old_fetch = quotemeta("$x->{old_path}:".
4874                                                  "refs/remotes/$x->{ref_id}");
4875                        command_noisy(qw/config --unset/,
4876                                      "$pfx.fetch", '^'. $old_fetch . '$');
4877                        delete $r->{$x->{old_repo_id}}->
4878                               {fetch}->{$x->{old_path}};
4879                        if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
4880                                command_noisy(qw/config --unset/,
4881                                              "$pfx.url");
4882                                push @emptied, $x->{old_repo_id}
4883                        }
4884                }
4885        }
4886        if (@emptied) {
4887                my $file = $ENV{GIT_CONFIG} || $ENV{GIT_CONFIG_LOCAL} ||
4888                           "$ENV{GIT_DIR}/config";
4889                print STDERR <<EOF;
4890The following [svn-remote] sections in your config file ($file) are empty
4891and can be safely removed:
4892EOF
4893                print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
4894        }
4895}
4896
4897sub migration_check {
4898        migrate_from_v0();
4899        migrate_from_v1();
4900        migrate_from_v2();
4901        minimize_connections() if $_minimize;
4902}
4903
4904package Git::IndexInfo;
4905use strict;
4906use warnings;
4907use Git qw/command_input_pipe command_close_pipe/;
4908
4909sub new {
4910        my ($class) = @_;
4911        my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
4912        bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
4913}
4914
4915sub remove {
4916        my ($self, $path) = @_;
4917        if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
4918                return ++$self->{nr};
4919        }
4920        undef;
4921}
4922
4923sub update {
4924        my ($self, $mode, $hash, $path) = @_;
4925        if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
4926                return ++$self->{nr};
4927        }
4928        undef;
4929}
4930
4931sub DESTROY {
4932        my ($self) = @_;
4933        command_close_pipe($self->{gui}, $self->{ctx});
4934}
4935
4936package Git::SVN::GlobSpec;
4937use strict;
4938use warnings;
4939
4940sub new {
4941        my ($class, $glob) = @_;
4942        my $re = $glob;
4943        $re =~ s!/+$!!g; # no need for trailing slashes
4944        $re =~ m!^([^*]*)(\*(?:/\*)*)([^*]*)$!;
4945        my $temp = $re;
4946        my ($left, $right) = ($1, $3);
4947        $re = $2;
4948        my $depth = $re =~ tr/*/*/;
4949        if ($depth != $temp =~ tr/*/*/) {
4950                die "Only one set of wildcard directories " .
4951                        "(e.g. '*' or '*/*/*') is supported: '$glob'\n";
4952        }
4953        if ($depth == 0) {
4954                die "One '*' is needed for glob: '$glob'\n";
4955        }
4956        $re =~ s!\*!\[^/\]*!g;
4957        $re = quotemeta($left) . "($re)" . quotemeta($right);
4958        if (length $left && !($left =~ s!/+$!!g)) {
4959                die "Missing trailing '/' on left side of: '$glob' ($left)\n";
4960        }
4961        if (length $right && !($right =~ s!^/+!!g)) {
4962                die "Missing leading '/' on right side of: '$glob' ($right)\n";
4963        }
4964        my $left_re = qr/^\/\Q$left\E(\/|$)/;
4965        bless { left => $left, right => $right, left_regex => $left_re,
4966                regex => qr/$re/, glob => $glob, depth => $depth }, $class;
4967}
4968
4969sub full_path {
4970        my ($self, $path) = @_;
4971        return (length $self->{left} ? "$self->{left}/" : '') .
4972               $path . (length $self->{right} ? "/$self->{right}" : '');
4973}
4974
4975__END__
4976
4977Data structures:
4978
4979
4980$remotes = { # returned by read_all_remotes()
4981        'svn' => {
4982                # svn-remote.svn.url=https://svn.musicpd.org
4983                url => 'https://svn.musicpd.org',
4984                # svn-remote.svn.fetch=mpd/trunk:trunk
4985                fetch => {
4986                        'mpd/trunk' => 'trunk',
4987                },
4988                # svn-remote.svn.tags=mpd/tags/*:tags/*
4989                tags => {
4990                        path => {
4991                                left => 'mpd/tags',
4992                                right => '',
4993                                regex => qr!mpd/tags/([^/]+)$!,
4994                                glob => 'tags/*',
4995                        },
4996                        ref => {
4997                                left => 'tags',
4998                                right => '',
4999                                regex => qr!tags/([^/]+)$!,
5000                                glob => 'tags/*',
5001                        },
5002                }
5003        }
5004};
5005
5006$log_entry hashref as returned by libsvn_log_entry()
5007{
5008        log => 'whitespace-formatted log entry
5009',                                              # trailing newline is preserved
5010        revision => '8',                        # integer
5011        date => '2004-02-24T17:01:44.108345Z',  # commit date
5012        author => 'committer name'
5013};
5014
5015
5016# this is generated by generate_diff();
5017@mods = array of diff-index line hashes, each element represents one line
5018        of diff-index output
5019
5020diff-index line ($m hash)
5021{
5022        mode_a => first column of diff-index output, no leading ':',
5023        mode_b => second column of diff-index output,
5024        sha1_b => sha1sum of the final blob,
5025        chg => change type [MCRADT],
5026        file_a => original file name of a file (iff chg is 'C' or 'R')
5027        file_b => new/current file name of a file (any chg)
5028}
5029;
5030
5031# retval of read_url_paths{,_all}();
5032$l_map = {
5033        # repository root url
5034        'https://svn.musicpd.org' => {
5035                # repository path               # GIT_SVN_ID
5036                'mpd/trunk'             =>      'trunk',
5037                'mpd/tags/0.11.5'       =>      'tags/0.11.5',
5038        },
5039}
5040
5041Notes:
5042        I don't trust the each() function on unless I created %hash myself
5043        because the internal iterator may not have started at base.