ee00e417d2091366e273878046b6fd3df318c29a
   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, "Translate between SVN revision numbers and tree-ish",
 175                        {} ],
 176        'rebase' => [ \&cmd_rebase, "Fetch and rebase your working directory",
 177                        { 'merge|m|M' => \$_merge,
 178                          'verbose|v' => \$_verbose,
 179                          'strategy|s=s' => \$_strategy,
 180                          'local|l' => \$_local,
 181                          'fetch-all|all' => \$_fetch_all,
 182                          'dry-run|n' => \$_dry_run,
 183                          %fc_opts } ],
 184        'commit-diff' => [ \&cmd_commit_diff,
 185                           'Commit a diff between two trees',
 186                        { 'message|m=s' => \$_message,
 187                          'file|F=s' => \$_file,
 188                          'revision|r=s' => \$_revision,
 189                        %cmt_opts } ],
 190        'info' => [ \&cmd_info,
 191                    "Show info about the latest SVN revision
 192                     on the current branch",
 193                    { 'url' => \$_url, } ],
 194        'blame' => [ \&Git::SVN::Log::cmd_blame,
 195                    "Show what revision and author last modified each line of a file",
 196                    { 'git-format' => \$_git_format } ],
 197);
 198
 199my $cmd;
 200for (my $i = 0; $i < @ARGV; $i++) {
 201        if (defined $cmd{$ARGV[$i]}) {
 202                $cmd = $ARGV[$i];
 203                splice @ARGV, $i, 1;
 204                last;
 205        }
 206};
 207
 208# make sure we're always running at the top-level working directory
 209unless ($cmd && $cmd =~ /(?:clone|init|multi-init)$/) {
 210        unless (-d $ENV{GIT_DIR}) {
 211                if ($git_dir_user_set) {
 212                        die "GIT_DIR=$ENV{GIT_DIR} explicitly set, ",
 213                            "but it is not a directory\n";
 214                }
 215                my $git_dir = delete $ENV{GIT_DIR};
 216                chomp(my $cdup = command_oneline(qw/rev-parse --show-cdup/));
 217                unless (length $cdup) {
 218                        die "Already at toplevel, but $git_dir ",
 219                            "not found '$cdup'\n";
 220                }
 221                chdir $cdup or die "Unable to chdir up to '$cdup'\n";
 222                unless (-d $git_dir) {
 223                        die "$git_dir still not found after going to ",
 224                            "'$cdup'\n";
 225                }
 226                $ENV{GIT_DIR} = $git_dir;
 227        }
 228        $_repository = Git->repository(Repository => $ENV{GIT_DIR});
 229}
 230
 231my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
 232
 233read_repo_config(\%opts);
 234Getopt::Long::Configure('pass_through') if ($cmd && ($cmd eq 'log' || $cmd eq 'blame'));
 235my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
 236                    'minimize-connections' => \$Git::SVN::Migration::_minimize,
 237                    'id|i=s' => \$Git::SVN::default_ref_id,
 238                    'svn-remote|remote|R=s' => sub {
 239                       $Git::SVN::no_reuse_existing = 1;
 240                       $Git::SVN::default_repo_id = $_[1] });
 241exit 1 if (!$rv && $cmd && $cmd ne 'log');
 242
 243usage(0) if $_help;
 244version() if $_version;
 245usage(1) unless defined $cmd;
 246load_authors() if $_authors;
 247
 248unless ($cmd =~ /^(?:clone|init|multi-init|commit-diff)$/) {
 249        Git::SVN::Migration::migration_check();
 250}
 251Git::SVN::init_vars();
 252eval {
 253        Git::SVN::verify_remotes_sanity();
 254        $cmd{$cmd}->[0]->(@ARGV);
 255};
 256fatal $@ if $@;
 257post_fetch_checkout();
 258exit 0;
 259
 260####################### primary functions ######################
 261sub usage {
 262        my $exit = shift || 0;
 263        my $fd = $exit ? \*STDERR : \*STDOUT;
 264        print $fd <<"";
 265git-svn - bidirectional operations between a single Subversion tree and git
 266Usage: git svn <command> [options] [arguments]\n
 267
 268        print $fd "Available commands:\n" unless $cmd;
 269
 270        foreach (sort keys %cmd) {
 271                next if $cmd && $cmd ne $_;
 272                next if /^multi-/; # don't show deprecated commands
 273                print $fd '  ',pack('A17',$_),$cmd{$_}->[1],"\n";
 274                foreach (sort keys %{$cmd{$_}->[2]}) {
 275                        # mixed-case options are for .git/config only
 276                        next if /[A-Z]/ && /^[a-z]+$/i;
 277                        # prints out arguments as they should be passed:
 278                        my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
 279                        print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
 280                                                        "--$_" : "-$_" }
 281                                                split /\|/,$_)," $x\n";
 282                }
 283        }
 284        print $fd <<"";
 285\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
 286arbitrary identifier if you're tracking multiple SVN branches/repositories in
 287one git repository and want to keep them separate.  See git-svn(1) for more
 288information.
 289
 290        exit $exit;
 291}
 292
 293sub version {
 294        print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
 295        exit 0;
 296}
 297
 298sub do_git_init_db {
 299        unless (-d $ENV{GIT_DIR}) {
 300                my @init_db = ('init');
 301                push @init_db, "--template=$_template" if defined $_template;
 302                if (defined $_shared) {
 303                        if ($_shared =~ /[a-z]/) {
 304                                push @init_db, "--shared=$_shared";
 305                        } else {
 306                                push @init_db, "--shared";
 307                        }
 308                }
 309                command_noisy(@init_db);
 310                $_repository = Git->repository(Repository => ".git");
 311        }
 312        my $set;
 313        my $pfx = "svn-remote.$Git::SVN::default_repo_id";
 314        foreach my $i (keys %icv) {
 315                die "'$set' and '$i' cannot both be set\n" if $set;
 316                next unless defined $icv{$i};
 317                command_noisy('config', "$pfx.$i", $icv{$i});
 318                $set = $i;
 319        }
 320}
 321
 322sub init_subdir {
 323        my $repo_path = shift or return;
 324        mkpath([$repo_path]) unless -d $repo_path;
 325        chdir $repo_path or die "Couldn't chdir to $repo_path: $!\n";
 326        $ENV{GIT_DIR} = '.git';
 327        $_repository = Git->repository(Repository => $ENV{GIT_DIR});
 328}
 329
 330sub cmd_clone {
 331        my ($url, $path) = @_;
 332        if (!defined $path &&
 333            (defined $_trunk || defined $_branches || defined $_tags ||
 334             defined $_stdlayout) &&
 335            $url !~ m#^[a-z\+]+://#) {
 336                $path = $url;
 337        }
 338        $path = basename($url) if !defined $path || !length $path;
 339        cmd_init($url, $path);
 340        Git::SVN::fetch_all($Git::SVN::default_repo_id);
 341}
 342
 343sub cmd_init {
 344        if (defined $_stdlayout) {
 345                $_trunk = 'trunk' if (!defined $_trunk);
 346                $_tags = 'tags' if (!defined $_tags);
 347                $_branches = 'branches' if (!defined $_branches);
 348        }
 349        if (defined $_trunk || defined $_branches || defined $_tags) {
 350                return cmd_multi_init(@_);
 351        }
 352        my $url = shift or die "SVN repository location required ",
 353                               "as a command-line argument\n";
 354        init_subdir(@_);
 355        do_git_init_db();
 356
 357        Git::SVN->init($url);
 358}
 359
 360sub cmd_fetch {
 361        if (grep /^\d+=./, @_) {
 362                die "'<rev>=<commit>' fetch arguments are ",
 363                    "no longer supported.\n";
 364        }
 365        my ($remote) = @_;
 366        if (@_ > 1) {
 367                die "Usage: $0 fetch [--all] [svn-remote]\n";
 368        }
 369        $remote ||= $Git::SVN::default_repo_id;
 370        if ($_fetch_all) {
 371                cmd_multi_fetch();
 372        } else {
 373                Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
 374        }
 375}
 376
 377sub cmd_set_tree {
 378        my (@commits) = @_;
 379        if ($_stdin || !@commits) {
 380                print "Reading from stdin...\n";
 381                @commits = ();
 382                while (<STDIN>) {
 383                        if (/\b($sha1_short)\b/o) {
 384                                unshift @commits, $1;
 385                        }
 386                }
 387        }
 388        my @revs;
 389        foreach my $c (@commits) {
 390                my @tmp = command('rev-parse',$c);
 391                if (scalar @tmp == 1) {
 392                        push @revs, $tmp[0];
 393                } elsif (scalar @tmp > 1) {
 394                        push @revs, reverse(command('rev-list',@tmp));
 395                } else {
 396                        fatal "Failed to rev-parse $c";
 397                }
 398        }
 399        my $gs = Git::SVN->new;
 400        my ($r_last, $cmt_last) = $gs->last_rev_commit;
 401        $gs->fetch;
 402        if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
 403                fatal "There are new revisions that were fetched ",
 404                      "and need to be merged (or acknowledged) ",
 405                      "before committing.\nlast rev: $r_last\n",
 406                      " current: $gs->{last_rev}";
 407        }
 408        $gs->set_tree($_) foreach @revs;
 409        print "Done committing ",scalar @revs," revisions to SVN\n";
 410        unlink $gs->{index};
 411}
 412
 413sub cmd_dcommit {
 414        my $head = shift;
 415        git_cmd_try { command_oneline(qw/diff-index --quiet HEAD/) }
 416                'Cannot dcommit with a dirty index.  Commit your changes first, '
 417                . "or stash them with `git stash'.\n";
 418        $head ||= 'HEAD';
 419        my @refs;
 420        my ($url, $rev, $uuid, $gs) = working_head_info($head, \@refs);
 421        $url = $_commit_url if defined $_commit_url;
 422        my $last_rev = $_revision if defined $_revision;
 423        if ($url) {
 424                print "Committing to $url ...\n";
 425        }
 426        unless ($gs) {
 427                die "Unable to determine upstream SVN information from ",
 428                    "$head history.\nPerhaps the repository is empty.";
 429        }
 430        my ($linear_refs, $parents) = linearize_history($gs, \@refs);
 431        if ($_no_rebase && scalar(@$linear_refs) > 1) {
 432                warn "Attempting to commit more than one change while ",
 433                     "--no-rebase is enabled.\n",
 434                     "If these changes depend on each other, re-running ",
 435                     "without --no-rebase may be required."
 436        }
 437        while (1) {
 438                my $d = shift @$linear_refs or last;
 439                unless (defined $last_rev) {
 440                        (undef, $last_rev, undef) = cmt_metadata("$d~1");
 441                        unless (defined $last_rev) {
 442                                fatal "Unable to extract revision information ",
 443                                      "from commit $d~1";
 444                        }
 445                }
 446                if ($_dry_run) {
 447                        print "diff-tree $d~1 $d\n";
 448                } else {
 449                        my $cmt_rev;
 450                        my %ed_opts = ( r => $last_rev,
 451                                        log => get_commit_entry($d)->{log},
 452                                        ra => Git::SVN::Ra->new($url),
 453                                        config => SVN::Core::config_get_config(
 454                                                $Git::SVN::Ra::config_dir
 455                                        ),
 456                                        tree_a => "$d~1",
 457                                        tree_b => $d,
 458                                        editor_cb => sub {
 459                                               print "Committed r$_[0]\n";
 460                                               $cmt_rev = $_[0];
 461                                        },
 462                                        svn_path => '');
 463                        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
 464                                print "No changes\n$d~1 == $d\n";
 465                        } elsif ($parents->{$d} && @{$parents->{$d}}) {
 466                                $gs->{inject_parents_dcommit}->{$cmt_rev} =
 467                                                               $parents->{$d};
 468                        }
 469                        $_fetch_all ? $gs->fetch_all : $gs->fetch;
 470                        $last_rev = $cmt_rev;
 471                        next if $_no_rebase;
 472
 473                        # we always want to rebase against the current HEAD,
 474                        # not any head that was passed to us
 475                        my @diff = command('diff-tree', $d,
 476                                           $gs->refname, '--');
 477                        my @finish;
 478                        if (@diff) {
 479                                @finish = rebase_cmd();
 480                                print STDERR "W: $d and ", $gs->refname,
 481                                             " differ, using @finish:\n",
 482                                             join("\n", @diff), "\n";
 483                        } else {
 484                                print "No changes between current HEAD and ",
 485                                      $gs->refname,
 486                                      "\nResetting to the latest ",
 487                                      $gs->refname, "\n";
 488                                @finish = qw/reset --mixed/;
 489                        }
 490                        command_noisy(@finish, $gs->refname);
 491                        if (@diff) {
 492                                @refs = ();
 493                                my ($url_, $rev_, $uuid_, $gs_) =
 494                                              working_head_info($head, \@refs);
 495                                my ($linear_refs_, $parents_) =
 496                                              linearize_history($gs_, \@refs);
 497                                if (scalar(@$linear_refs) !=
 498                                    scalar(@$linear_refs_)) {
 499                                        fatal "# of revisions changed ",
 500                                          "\nbefore:\n",
 501                                          join("\n", @$linear_refs),
 502                                          "\n\nafter:\n",
 503                                          join("\n", @$linear_refs_), "\n",
 504                                          'If you are attempting to commit ',
 505                                          "merges, try running:\n\t",
 506                                          'git rebase --interactive',
 507                                          '--preserve-merges ',
 508                                          $gs->refname,
 509                                          "\nBefore dcommitting";
 510                                }
 511                                if ($url_ ne $url) {
 512                                        fatal "URL mismatch after rebase: ",
 513                                              "$url_ != $url";
 514                                }
 515                                if ($uuid_ ne $uuid) {
 516                                        fatal "uuid mismatch after rebase: ",
 517                                              "$uuid_ != $uuid";
 518                                }
 519                                # remap parents
 520                                my (%p, @l, $i);
 521                                for ($i = 0; $i < scalar @$linear_refs; $i++) {
 522                                        my $new = $linear_refs_->[$i] or next;
 523                                        $p{$new} =
 524                                                $parents->{$linear_refs->[$i]};
 525                                        push @l, $new;
 526                                }
 527                                $parents = \%p;
 528                                $linear_refs = \@l;
 529                        }
 530                }
 531        }
 532        unlink $gs->{index};
 533}
 534
 535sub cmd_find_rev {
 536        my $revision_or_hash = shift or die "SVN or git revision required ",
 537                                            "as a command-line argument\n";
 538        my $result;
 539        if ($revision_or_hash =~ /^r\d+$/) {
 540                my $head = shift;
 541                $head ||= 'HEAD';
 542                my @refs;
 543                my (undef, undef, $uuid, $gs) = working_head_info($head, \@refs);
 544                unless ($gs) {
 545                        die "Unable to determine upstream SVN information from ",
 546                            "$head history\n";
 547                }
 548                my $desired_revision = substr($revision_or_hash, 1);
 549                $result = $gs->rev_map_get($desired_revision, $uuid);
 550        } else {
 551                my (undef, $rev, undef) = cmt_metadata($revision_or_hash);
 552                $result = $rev;
 553        }
 554        print "$result\n" if $result;
 555}
 556
 557sub cmd_rebase {
 558        command_noisy(qw/update-index --refresh/);
 559        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 560        unless ($gs) {
 561                die "Unable to determine upstream SVN information from ",
 562                    "working tree history\n";
 563        }
 564        if ($_dry_run) {
 565                print "Remote Branch: " . $gs->refname . "\n";
 566                print "SVN URL: " . $url . "\n";
 567                return;
 568        }
 569        if (command(qw/diff-index HEAD --/)) {
 570                print STDERR "Cannot rebase with uncommited changes:\n";
 571                command_noisy('status');
 572                exit 1;
 573        }
 574        unless ($_local) {
 575                # rebase will checkout for us, so no need to do it explicitly
 576                $_no_checkout = 'true';
 577                $_fetch_all ? $gs->fetch_all : $gs->fetch;
 578        }
 579        command_noisy(rebase_cmd(), $gs->refname);
 580}
 581
 582sub cmd_show_ignore {
 583        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 584        $gs ||= Git::SVN->new;
 585        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 586        $gs->prop_walk($gs->{path}, $r, sub {
 587                my ($gs, $path, $props) = @_;
 588                print STDOUT "\n# $path\n";
 589                my $s = $props->{'svn:ignore'} or return;
 590                $s =~ s/[\r\n]+/\n/g;
 591                chomp $s;
 592                $s =~ s#^#$path#gm;
 593                print STDOUT "$s\n";
 594        });
 595}
 596
 597sub cmd_show_externals {
 598        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 599        $gs ||= Git::SVN->new;
 600        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 601        $gs->prop_walk($gs->{path}, $r, sub {
 602                my ($gs, $path, $props) = @_;
 603                print STDOUT "\n# $path\n";
 604                my $s = $props->{'svn:externals'} or return;
 605                $s =~ s/[\r\n]+/\n/g;
 606                chomp $s;
 607                $s =~ s#^#$path#gm;
 608                print STDOUT "$s\n";
 609        });
 610}
 611
 612sub cmd_create_ignore {
 613        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 614        $gs ||= Git::SVN->new;
 615        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 616        $gs->prop_walk($gs->{path}, $r, sub {
 617                my ($gs, $path, $props) = @_;
 618                # $path is of the form /path/to/dir/
 619                my $ignore = '.' . $path . '.gitignore';
 620                my $s = $props->{'svn:ignore'} or return;
 621                open(GITIGNORE, '>', $ignore)
 622                  or fatal("Failed to open `$ignore' for writing: $!");
 623                $s =~ s/[\r\n]+/\n/g;
 624                chomp $s;
 625                # Prefix all patterns so that the ignore doesn't apply
 626                # to sub-directories.
 627                $s =~ s#^#/#gm;
 628                print GITIGNORE "$s\n";
 629                close(GITIGNORE)
 630                  or fatal("Failed to close `$ignore': $!");
 631                command_noisy('add', '-f', $ignore);
 632        });
 633}
 634
 635sub canonicalize_path {
 636        my ($path) = @_;
 637        my $dot_slash_added = 0;
 638        if (substr($path, 0, 1) ne "/") {
 639                $path = "./" . $path;
 640                $dot_slash_added = 1;
 641        }
 642        # File::Spec->canonpath doesn't collapse x/../y into y (for a
 643        # good reason), so let's do this manually.
 644        $path =~ s#/+#/#g;
 645        $path =~ s#/\.(?:/|$)#/#g;
 646        $path =~ s#/[^/]+/\.\.##g;
 647        $path =~ s#/$##g;
 648        $path =~ s#^\./## if $dot_slash_added;
 649        $path =~ s#^/##;
 650        $path =~ s#^\.$##;
 651        return $path;
 652}
 653
 654# get_svnprops(PATH)
 655# ------------------
 656# Helper for cmd_propget and cmd_proplist below.
 657sub get_svnprops {
 658        my $path = shift;
 659        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 660        $gs ||= Git::SVN->new;
 661
 662        # prefix THE PATH by the sub-directory from which the user
 663        # invoked us.
 664        $path = $cmd_dir_prefix . $path;
 665        fatal("No such file or directory: $path") unless -e $path;
 666        my $is_dir = -d $path ? 1 : 0;
 667        $path = $gs->{path} . '/' . $path;
 668
 669        # canonicalize the path (otherwise libsvn will abort or fail to
 670        # find the file)
 671        $path = canonicalize_path($path);
 672
 673        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 674        my $props;
 675        if ($is_dir) {
 676                (undef, undef, $props) = $gs->ra->get_dir($path, $r);
 677        }
 678        else {
 679                (undef, $props) = $gs->ra->get_file($path, $r, undef);
 680        }
 681        return $props;
 682}
 683
 684# cmd_propget (PROP, PATH)
 685# ------------------------
 686# Print the SVN property PROP for PATH.
 687sub cmd_propget {
 688        my ($prop, $path) = @_;
 689        $path = '.' if not defined $path;
 690        usage(1) if not defined $prop;
 691        my $props = get_svnprops($path);
 692        if (not defined $props->{$prop}) {
 693                fatal("`$path' does not have a `$prop' SVN property.");
 694        }
 695        print $props->{$prop} . "\n";
 696}
 697
 698# cmd_proplist (PATH)
 699# -------------------
 700# Print the list of SVN properties for PATH.
 701sub cmd_proplist {
 702        my $path = shift;
 703        $path = '.' if not defined $path;
 704        my $props = get_svnprops($path);
 705        print "Properties on '$path':\n";
 706        foreach (sort keys %{$props}) {
 707                print "  $_\n";
 708        }
 709}
 710
 711sub cmd_multi_init {
 712        my $url = shift;
 713        unless (defined $_trunk || defined $_branches || defined $_tags) {
 714                usage(1);
 715        }
 716
 717        # there are currently some bugs that prevent multi-init/multi-fetch
 718        # setups from working well without this.
 719        $Git::SVN::_minimize_url = 1;
 720
 721        $_prefix = '' unless defined $_prefix;
 722        if (defined $url) {
 723                $url =~ s#/+$##;
 724                init_subdir(@_);
 725        }
 726        do_git_init_db();
 727        if (defined $_trunk) {
 728                my $trunk_ref = $_prefix . 'trunk';
 729                # try both old-style and new-style lookups:
 730                my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
 731                unless ($gs_trunk) {
 732                        my ($trunk_url, $trunk_path) =
 733                                              complete_svn_url($url, $_trunk);
 734                        $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
 735                                                   undef, $trunk_ref);
 736                }
 737        }
 738        return unless defined $_branches || defined $_tags;
 739        my $ra = $url ? Git::SVN::Ra->new($url) : undef;
 740        complete_url_ls_init($ra, $_branches, '--branches/-b', $_prefix);
 741        complete_url_ls_init($ra, $_tags, '--tags/-t', $_prefix . 'tags/');
 742}
 743
 744sub cmd_multi_fetch {
 745        my $remotes = Git::SVN::read_all_remotes();
 746        foreach my $repo_id (sort keys %$remotes) {
 747                if ($remotes->{$repo_id}->{url}) {
 748                        Git::SVN::fetch_all($repo_id, $remotes);
 749                }
 750        }
 751}
 752
 753# this command is special because it requires no metadata
 754sub cmd_commit_diff {
 755        my ($ta, $tb, $url) = @_;
 756        my $usage = "Usage: $0 commit-diff -r<revision> ".
 757                    "<tree-ish> <tree-ish> [<URL>]";
 758        fatal($usage) if (!defined $ta || !defined $tb);
 759        my $svn_path = '';
 760        if (!defined $url) {
 761                my $gs = eval { Git::SVN->new };
 762                if (!$gs) {
 763                        fatal("Needed URL or usable git-svn --id in ",
 764                              "the command-line\n", $usage);
 765                }
 766                $url = $gs->{url};
 767                $svn_path = $gs->{path};
 768        }
 769        unless (defined $_revision) {
 770                fatal("-r|--revision is a required argument\n", $usage);
 771        }
 772        if (defined $_message && defined $_file) {
 773                fatal("Both --message/-m and --file/-F specified ",
 774                      "for the commit message.\n",
 775                      "I have no idea what you mean");
 776        }
 777        if (defined $_file) {
 778                $_message = file_to_s($_file);
 779        } else {
 780                $_message ||= get_commit_entry($tb)->{log};
 781        }
 782        my $ra ||= Git::SVN::Ra->new($url);
 783        my $r = $_revision;
 784        if ($r eq 'HEAD') {
 785                $r = $ra->get_latest_revnum;
 786        } elsif ($r !~ /^\d+$/) {
 787                die "revision argument: $r not understood by git-svn\n";
 788        }
 789        my %ed_opts = ( r => $r,
 790                        log => $_message,
 791                        ra => $ra,
 792                        tree_a => $ta,
 793                        tree_b => $tb,
 794                        editor_cb => sub { print "Committed r$_[0]\n" },
 795                        svn_path => $svn_path );
 796        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
 797                print "No changes\n$ta == $tb\n";
 798        }
 799}
 800
 801sub cmd_info {
 802        my $path = canonicalize_path(defined($_[0]) ? $_[0] : ".");
 803        if (exists $_[1]) {
 804                die "Too many arguments specified\n";
 805        }
 806
 807        my ($file_type, $diff_status) = find_file_type_and_diff_status($path);
 808
 809        if (!$file_type && !$diff_status) {
 810                print STDERR "$path:  (Not a versioned resource)\n\n";
 811                return;
 812        }
 813
 814        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 815        unless ($gs) {
 816                die "Unable to determine upstream SVN information from ",
 817                    "working tree history\n";
 818        }
 819
 820        # canonicalize_path() will return "" to make libsvn 1.5.x happy,
 821        $path = "." if $path eq "";
 822
 823        my $full_url = $url . ($path eq "." ? "" : "/$path");
 824
 825        if ($_url) {
 826                print $full_url, "\n";
 827                return;
 828        }
 829
 830        my $result = "Path: $path\n";
 831        $result .= "Name: " . basename($path) . "\n" if $file_type ne "dir";
 832        $result .= "URL: " . $full_url . "\n";
 833
 834        eval {
 835                my $repos_root = $gs->repos_root;
 836                Git::SVN::remove_username($repos_root);
 837                $result .= "Repository Root: $repos_root\n";
 838        };
 839        if ($@) {
 840                $result .= "Repository Root: (offline)\n";
 841        }
 842        $result .= "Repository UUID: $uuid\n" unless $diff_status eq "A";
 843        $result .= "Revision: " . ($diff_status eq "A" ? 0 : $rev) . "\n";
 844
 845        $result .= "Node Kind: " .
 846                   ($file_type eq "dir" ? "directory" : "file") . "\n";
 847
 848        my $schedule = $diff_status eq "A"
 849                       ? "add"
 850                       : ($diff_status eq "D" ? "delete" : "normal");
 851        $result .= "Schedule: $schedule\n";
 852
 853        if ($diff_status eq "A") {
 854                print $result, "\n";
 855                return;
 856        }
 857
 858        my ($lc_author, $lc_rev, $lc_date_utc);
 859        my @args = Git::SVN::Log::git_svn_log_cmd($rev, $rev, "--", $path);
 860        my $log = command_output_pipe(@args);
 861        my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
 862        while (<$log>) {
 863                if (/^${esc_color}author (.+) <[^>]+> (\d+) ([\-\+]?\d+)$/o) {
 864                        $lc_author = $1;
 865                        $lc_date_utc = Git::SVN::Log::parse_git_date($2, $3);
 866                } elsif (/^${esc_color}    (git-svn-id:.+)$/o) {
 867                        (undef, $lc_rev, undef) = ::extract_metadata($1);
 868                }
 869        }
 870        close $log;
 871
 872        Git::SVN::Log::set_local_timezone();
 873
 874        $result .= "Last Changed Author: $lc_author\n";
 875        $result .= "Last Changed Rev: $lc_rev\n";
 876        $result .= "Last Changed Date: " .
 877                   Git::SVN::Log::format_svn_date($lc_date_utc) . "\n";
 878
 879        if ($file_type ne "dir") {
 880                my $text_last_updated_date =
 881                    ($diff_status eq "D" ? $lc_date_utc : (stat $path)[9]);
 882                $result .=
 883                    "Text Last Updated: " .
 884                    Git::SVN::Log::format_svn_date($text_last_updated_date) .
 885                    "\n";
 886                my $checksum;
 887                if ($diff_status eq "D") {
 888                        my ($fh, $ctx) =
 889                            command_output_pipe(qw(cat-file blob), "HEAD:$path");
 890                        if ($file_type eq "link") {
 891                                my $file_name = <$fh>;
 892                                $checksum = md5sum("link $file_name");
 893                        } else {
 894                                $checksum = md5sum($fh);
 895                        }
 896                        command_close_pipe($fh, $ctx);
 897                } elsif ($file_type eq "link") {
 898                        my $file_name =
 899                            command(qw(cat-file blob), "HEAD:$path");
 900                        $checksum =
 901                            md5sum("link " . $file_name);
 902                } else {
 903                        open FILE, "<", $path or die $!;
 904                        $checksum = md5sum(\*FILE);
 905                        close FILE or die $!;
 906                }
 907                $result .= "Checksum: " . $checksum . "\n";
 908        }
 909
 910        print $result, "\n";
 911}
 912
 913########################### utility functions #########################
 914
 915sub rebase_cmd {
 916        my @cmd = qw/rebase/;
 917        push @cmd, '-v' if $_verbose;
 918        push @cmd, qw/--merge/ if $_merge;
 919        push @cmd, "--strategy=$_strategy" if $_strategy;
 920        @cmd;
 921}
 922
 923sub post_fetch_checkout {
 924        return if $_no_checkout;
 925        my $gs = $Git::SVN::_head or return;
 926        return if verify_ref('refs/heads/master^0');
 927
 928        my $valid_head = verify_ref('HEAD^0');
 929        command_noisy(qw(update-ref refs/heads/master), $gs->refname);
 930        return if ($valid_head || !verify_ref('HEAD^0'));
 931
 932        return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
 933        my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
 934        return if -f $index;
 935
 936        return if command_oneline(qw/rev-parse --is-inside-work-tree/) eq 'false';
 937        return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
 938        command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
 939        print STDERR "Checked out HEAD:\n  ",
 940                     $gs->full_url, " r", $gs->last_rev, "\n";
 941}
 942
 943sub complete_svn_url {
 944        my ($url, $path) = @_;
 945        $path =~ s#/+$##;
 946        if ($path !~ m#^[a-z\+]+://#) {
 947                if (!defined $url || $url !~ m#^[a-z\+]+://#) {
 948                        fatal("E: '$path' is not a complete URL ",
 949                              "and a separate URL is not specified");
 950                }
 951                return ($url, $path);
 952        }
 953        return ($path, '');
 954}
 955
 956sub complete_url_ls_init {
 957        my ($ra, $repo_path, $switch, $pfx) = @_;
 958        unless ($repo_path) {
 959                print STDERR "W: $switch not specified\n";
 960                return;
 961        }
 962        $repo_path =~ s#/+$##;
 963        if ($repo_path =~ m#^[a-z\+]+://#) {
 964                $ra = Git::SVN::Ra->new($repo_path);
 965                $repo_path = '';
 966        } else {
 967                $repo_path =~ s#^/+##;
 968                unless ($ra) {
 969                        fatal("E: '$repo_path' is not a complete URL ",
 970                              "and a separate URL is not specified");
 971                }
 972        }
 973        my $url = $ra->{url};
 974        my $gs = Git::SVN->init($url, undef, undef, undef, 1);
 975        my $k = "svn-remote.$gs->{repo_id}.url";
 976        my $orig_url = eval { command_oneline(qw/config --get/, $k) };
 977        if ($orig_url && ($orig_url ne $gs->{url})) {
 978                die "$k already set: $orig_url\n",
 979                    "wanted to set to: $gs->{url}\n";
 980        }
 981        command_oneline('config', $k, $gs->{url}) unless $orig_url;
 982        my $remote_path = "$ra->{svn_path}/$repo_path";
 983        $remote_path =~ s#/+#/#g;
 984        $remote_path =~ s#^/##g;
 985        $remote_path .= "/*" if $remote_path !~ /\*/;
 986        my ($n) = ($switch =~ /^--(\w+)/);
 987        if (length $pfx && $pfx !~ m#/$#) {
 988                die "--prefix='$pfx' must have a trailing slash '/'\n";
 989        }
 990        command_noisy('config',
 991                      "svn-remote.$gs->{repo_id}.$n",
 992                      "$remote_path:refs/remotes/$pfx*" .
 993                        ('/*' x (($remote_path =~ tr/*/*/) - 1)) );
 994}
 995
 996sub verify_ref {
 997        my ($ref) = @_;
 998        eval { command_oneline([ 'rev-parse', '--verify', $ref ],
 999                               { STDERR => 0 }); };
1000}
1001
1002sub get_tree_from_treeish {
1003        my ($treeish) = @_;
1004        # $treeish can be a symbolic ref, too:
1005        my $type = command_oneline(qw/cat-file -t/, $treeish);
1006        my $expected;
1007        while ($type eq 'tag') {
1008                ($treeish, $type) = command(qw/cat-file tag/, $treeish);
1009        }
1010        if ($type eq 'commit') {
1011                $expected = (grep /^tree /, command(qw/cat-file commit/,
1012                                                    $treeish))[0];
1013                ($expected) = ($expected =~ /^tree ($sha1)$/o);
1014                die "Unable to get tree from $treeish\n" unless $expected;
1015        } elsif ($type eq 'tree') {
1016                $expected = $treeish;
1017        } else {
1018                die "$treeish is a $type, expected tree, tag or commit\n";
1019        }
1020        return $expected;
1021}
1022
1023sub get_commit_entry {
1024        my ($treeish) = shift;
1025        my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
1026        my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
1027        my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
1028        open my $log_fh, '>', $commit_editmsg or croak $!;
1029
1030        my $type = command_oneline(qw/cat-file -t/, $treeish);
1031        if ($type eq 'commit' || $type eq 'tag') {
1032                my ($msg_fh, $ctx) = command_output_pipe('cat-file',
1033                                                         $type, $treeish);
1034                my $in_msg = 0;
1035                my $author;
1036                my $saw_from = 0;
1037                my $msgbuf = "";
1038                while (<$msg_fh>) {
1039                        if (!$in_msg) {
1040                                $in_msg = 1 if (/^\s*$/);
1041                                $author = $1 if (/^author (.*>)/);
1042                        } elsif (/^git-svn-id: /) {
1043                                # skip this for now, we regenerate the
1044                                # correct one on re-fetch anyways
1045                                # TODO: set *:merge properties or like...
1046                        } else {
1047                                if (/^From:/ || /^Signed-off-by:/) {
1048                                        $saw_from = 1;
1049                                }
1050                                $msgbuf .= $_;
1051                        }
1052                }
1053                $msgbuf =~ s/\s+$//s;
1054                if ($Git::SVN::_add_author_from && defined($author)
1055                    && !$saw_from) {
1056                        $msgbuf .= "\n\nFrom: $author";
1057                }
1058                print $log_fh $msgbuf or croak $!;
1059                command_close_pipe($msg_fh, $ctx);
1060        }
1061        close $log_fh or croak $!;
1062
1063        if ($_edit || ($type eq 'tree')) {
1064                my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
1065                # TODO: strip out spaces, comments, like git-commit.sh
1066                system($editor, $commit_editmsg);
1067        }
1068        rename $commit_editmsg, $commit_msg or croak $!;
1069        open $log_fh, '<', $commit_msg or croak $!;
1070        { local $/; chomp($log_entry{log} = <$log_fh>); }
1071        close $log_fh or croak $!;
1072        unlink $commit_msg;
1073        \%log_entry;
1074}
1075
1076sub s_to_file {
1077        my ($str, $file, $mode) = @_;
1078        open my $fd,'>',$file or croak $!;
1079        print $fd $str,"\n" or croak $!;
1080        close $fd or croak $!;
1081        chmod ($mode &~ umask, $file) if (defined $mode);
1082}
1083
1084sub file_to_s {
1085        my $file = shift;
1086        open my $fd,'<',$file or croak "$!: file: $file\n";
1087        local $/;
1088        my $ret = <$fd>;
1089        close $fd or croak $!;
1090        $ret =~ s/\s*$//s;
1091        return $ret;
1092}
1093
1094# '<svn username> = real-name <email address>' mapping based on git-svnimport:
1095sub load_authors {
1096        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1097        my $log = $cmd eq 'log';
1098        while (<$authors>) {
1099                chomp;
1100                next unless /^(.+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
1101                my ($user, $name, $email) = ($1, $2, $3);
1102                if ($log) {
1103                        $Git::SVN::Log::rusers{"$name <$email>"} = $user;
1104                } else {
1105                        $users{$user} = [$name, $email];
1106                }
1107        }
1108        close $authors or croak $!;
1109}
1110
1111# convert GetOpt::Long specs for use by git-config
1112sub read_repo_config {
1113        return unless -d $ENV{GIT_DIR};
1114        my $opts = shift;
1115        my @config_only;
1116        foreach my $o (keys %$opts) {
1117                # if we have mixedCase and a long option-only, then
1118                # it's a config-only variable that we don't need for
1119                # the command-line.
1120                push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
1121                my $v = $opts->{$o};
1122                my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
1123                $key =~ s/-//g;
1124                my $arg = 'git-config';
1125                $arg .= ' --int' if ($o =~ /[:=]i$/);
1126                $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
1127                if (ref $v eq 'ARRAY') {
1128                        chomp(my @tmp = `$arg --get-all svn.$key`);
1129                        @$v = @tmp if @tmp;
1130                } else {
1131                        chomp(my $tmp = `$arg --get svn.$key`);
1132                        if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
1133                                $$v = $tmp;
1134                        }
1135                }
1136        }
1137        delete @$opts{@config_only} if @config_only;
1138}
1139
1140sub extract_metadata {
1141        my $id = shift or return (undef, undef, undef);
1142        my ($url, $rev, $uuid) = ($id =~ /^\s*git-svn-id:\s+(.*)\@(\d+)
1143                                                        \s([a-f\d\-]+)$/x);
1144        if (!defined $rev || !$uuid || !$url) {
1145                # some of the original repositories I made had
1146                # identifiers like this:
1147                ($rev, $uuid) = ($id =~/^\s*git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
1148        }
1149        return ($url, $rev, $uuid);
1150}
1151
1152sub cmt_metadata {
1153        return extract_metadata((grep(/^git-svn-id: /,
1154                command(qw/cat-file commit/, shift)))[-1]);
1155}
1156
1157sub working_head_info {
1158        my ($head, $refs) = @_;
1159        my @args = ('log', '--no-color', '--first-parent', '--pretty=medium');
1160        my ($fh, $ctx) = command_output_pipe(@args, $head);
1161        my $hash;
1162        my %max;
1163        while (<$fh>) {
1164                if ( m{^commit ($::sha1)$} ) {
1165                        unshift @$refs, $hash if $hash and $refs;
1166                        $hash = $1;
1167                        next;
1168                }
1169                next unless s{^\s*(git-svn-id:)}{$1};
1170                my ($url, $rev, $uuid) = extract_metadata($_);
1171                if (defined $url && defined $rev) {
1172                        next if $max{$url} and $max{$url} < $rev;
1173                        if (my $gs = Git::SVN->find_by_url($url)) {
1174                                my $c = $gs->rev_map_get($rev, $uuid);
1175                                if ($c && $c eq $hash) {
1176                                        close $fh; # break the pipe
1177                                        return ($url, $rev, $uuid, $gs);
1178                                } else {
1179                                        $max{$url} ||= $gs->rev_map_max;
1180                                }
1181                        }
1182                }
1183        }
1184        command_close_pipe($fh, $ctx);
1185        (undef, undef, undef, undef);
1186}
1187
1188sub read_commit_parents {
1189        my ($parents, $c) = @_;
1190        chomp(my $p = command_oneline(qw/rev-list --parents -1/, $c));
1191        $p =~ s/^($c)\s*// or die "rev-list --parents -1 $c failed!\n";
1192        @{$parents->{$c}} = split(/ /, $p);
1193}
1194
1195sub linearize_history {
1196        my ($gs, $refs) = @_;
1197        my %parents;
1198        foreach my $c (@$refs) {
1199                read_commit_parents(\%parents, $c);
1200        }
1201
1202        my @linear_refs;
1203        my %skip = ();
1204        my $last_svn_commit = $gs->last_commit;
1205        foreach my $c (reverse @$refs) {
1206                next if $c eq $last_svn_commit;
1207                last if $skip{$c};
1208
1209                unshift @linear_refs, $c;
1210                $skip{$c} = 1;
1211
1212                # we only want the first parent to diff against for linear
1213                # history, we save the rest to inject when we finalize the
1214                # svn commit
1215                my $fp_a = verify_ref("$c~1");
1216                my $fp_b = shift @{$parents{$c}} if $parents{$c};
1217                if (!$fp_a || !$fp_b) {
1218                        die "Commit $c\n",
1219                            "has no parent commit, and therefore ",
1220                            "nothing to diff against.\n",
1221                            "You should be working from a repository ",
1222                            "originally created by git-svn\n";
1223                }
1224                if ($fp_a ne $fp_b) {
1225                        die "$c~1 = $fp_a, however parsing commit $c ",
1226                            "revealed that:\n$c~1 = $fp_b\nBUG!\n";
1227                }
1228
1229                foreach my $p (@{$parents{$c}}) {
1230                        $skip{$p} = 1;
1231                }
1232        }
1233        (\@linear_refs, \%parents);
1234}
1235
1236sub find_file_type_and_diff_status {
1237        my ($path) = @_;
1238        return ('dir', '') if $path eq '';
1239
1240        my $diff_output =
1241            command_oneline(qw(diff --cached --name-status --), $path) || "";
1242        my $diff_status = (split(' ', $diff_output))[0] || "";
1243
1244        my $ls_tree = command_oneline(qw(ls-tree HEAD), $path) || "";
1245
1246        return (undef, undef) if !$diff_status && !$ls_tree;
1247
1248        if ($diff_status eq "A") {
1249                return ("link", $diff_status) if -l $path;
1250                return ("dir", $diff_status) if -d $path;
1251                return ("file", $diff_status);
1252        }
1253
1254        my $mode = (split(' ', $ls_tree))[0] || "";
1255
1256        return ("link", $diff_status) if $mode eq "120000";
1257        return ("dir", $diff_status) if $mode eq "040000";
1258        return ("file", $diff_status);
1259}
1260
1261sub md5sum {
1262        my $arg = shift;
1263        my $ref = ref $arg;
1264        my $md5 = Digest::MD5->new();
1265        if ($ref eq 'GLOB' || $ref eq 'IO::File') {
1266                $md5->addfile($arg) or croak $!;
1267        } elsif ($ref eq 'SCALAR') {
1268                $md5->add($$arg) or croak $!;
1269        } elsif (!$ref) {
1270                $md5->add($arg) or croak $!;
1271        } else {
1272                ::fatal "Can't provide MD5 hash for unknown ref type: '", $ref, "'";
1273        }
1274        return $md5->hexdigest();
1275}
1276
1277package Git::SVN;
1278use strict;
1279use warnings;
1280use Fcntl qw/:DEFAULT :seek/;
1281use constant rev_map_fmt => 'NH40';
1282use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
1283            $_repack $_repack_flags $_use_svm_props $_head
1284            $_use_svnsync_props $no_reuse_existing $_minimize_url
1285            $_use_log_author $_add_author_from/;
1286use Carp qw/croak/;
1287use File::Path qw/mkpath/;
1288use File::Copy qw/copy/;
1289use IPC::Open3;
1290
1291my ($_gc_nr, $_gc_period);
1292
1293# properties that we do not log:
1294my %SKIP_PROP;
1295BEGIN {
1296        %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
1297                                        svn:special svn:executable
1298                                        svn:entry:committed-rev
1299                                        svn:entry:last-author
1300                                        svn:entry:uuid
1301                                        svn:entry:committed-date/;
1302
1303        # some options are read globally, but can be overridden locally
1304        # per [svn-remote "..."] section.  Command-line options will *NOT*
1305        # override options set in an [svn-remote "..."] section
1306        no strict 'refs';
1307        for my $option (qw/follow_parent no_metadata use_svm_props
1308                           use_svnsync_props/) {
1309                my $key = $option;
1310                $key =~ tr/_//d;
1311                my $prop = "-$option";
1312                *$option = sub {
1313                        my ($self) = @_;
1314                        return $self->{$prop} if exists $self->{$prop};
1315                        my $k = "svn-remote.$self->{repo_id}.$key";
1316                        eval { command_oneline(qw/config --get/, $k) };
1317                        if ($@) {
1318                                $self->{$prop} = ${"Git::SVN::_$option"};
1319                        } else {
1320                                my $v = command_oneline(qw/config --bool/,$k);
1321                                $self->{$prop} = $v eq 'false' ? 0 : 1;
1322                        }
1323                        return $self->{$prop};
1324                }
1325        }
1326}
1327
1328my (%LOCKFILES, %INDEX_FILES);
1329END {
1330        unlink keys %LOCKFILES if %LOCKFILES;
1331        unlink keys %INDEX_FILES if %INDEX_FILES;
1332}
1333
1334sub resolve_local_globs {
1335        my ($url, $fetch, $glob_spec) = @_;
1336        return unless defined $glob_spec;
1337        my $ref = $glob_spec->{ref};
1338        my $path = $glob_spec->{path};
1339        foreach (command(qw#for-each-ref --format=%(refname) refs/remotes#)) {
1340                next unless m#^refs/remotes/$ref->{regex}$#;
1341                my $p = $1;
1342                my $pathname = desanitize_refname($path->full_path($p));
1343                my $refname = desanitize_refname($ref->full_path($p));
1344                if (my $existing = $fetch->{$pathname}) {
1345                        if ($existing ne $refname) {
1346                                die "Refspec conflict:\n",
1347                                    "existing: refs/remotes/$existing\n",
1348                                    " globbed: refs/remotes/$refname\n";
1349                        }
1350                        my $u = (::cmt_metadata("refs/remotes/$refname"))[0];
1351                        $u =~ s!^\Q$url\E(/|$)!! or die
1352                          "refs/remotes/$refname: '$url' not found in '$u'\n";
1353                        if ($pathname ne $u) {
1354                                warn "W: Refspec glob conflict ",
1355                                     "(ref: refs/remotes/$refname):\n",
1356                                     "expected path: $pathname\n",
1357                                     "    real path: $u\n",
1358                                     "Continuing ahead with $u\n";
1359                                next;
1360                        }
1361                } else {
1362                        $fetch->{$pathname} = $refname;
1363                }
1364        }
1365}
1366
1367sub parse_revision_argument {
1368        my ($base, $head) = @_;
1369        if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
1370                return ($base, $head);
1371        }
1372        return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
1373        return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
1374        return ($head, $head) if ($::_revision eq 'HEAD');
1375        return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
1376        return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
1377        die "revision argument: $::_revision not understood by git-svn\n";
1378}
1379
1380sub fetch_all {
1381        my ($repo_id, $remotes) = @_;
1382        if (ref $repo_id) {
1383                my $gs = $repo_id;
1384                $repo_id = undef;
1385                $repo_id = $gs->{repo_id};
1386        }
1387        $remotes ||= read_all_remotes();
1388        my $remote = $remotes->{$repo_id} or
1389                     die "[svn-remote \"$repo_id\"] unknown\n";
1390        my $fetch = $remote->{fetch};
1391        my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
1392        my (@gs, @globs);
1393        my $ra = Git::SVN::Ra->new($url);
1394        my $uuid = $ra->get_uuid;
1395        my $head = $ra->get_latest_revnum;
1396        my $base = defined $fetch ? $head : 0;
1397
1398        # read the max revs for wildcard expansion (branches/*, tags/*)
1399        foreach my $t (qw/branches tags/) {
1400                defined $remote->{$t} or next;
1401                push @globs, $remote->{$t};
1402                my $max_rev = eval { tmp_config(qw/--int --get/,
1403                                         "svn-remote.$repo_id.${t}-maxRev") };
1404                if (defined $max_rev && ($max_rev < $base)) {
1405                        $base = $max_rev;
1406                } elsif (!defined $max_rev) {
1407                        $base = 0;
1408                }
1409        }
1410
1411        if ($fetch) {
1412                foreach my $p (sort keys %$fetch) {
1413                        my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
1414                        my $lr = $gs->rev_map_max;
1415                        if (defined $lr) {
1416                                $base = $lr if ($lr < $base);
1417                        }
1418                        push @gs, $gs;
1419                }
1420        }
1421
1422        ($base, $head) = parse_revision_argument($base, $head);
1423        $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
1424}
1425
1426sub read_all_remotes {
1427        my $r = {};
1428        my $use_svm_props = eval { command_oneline(qw/config --bool
1429            svn.useSvmProps/) };
1430        $use_svm_props = $use_svm_props eq 'true' if $use_svm_props;
1431        foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
1432                if (m!^(.+)\.fetch=\s*(.*)\s*:\s*(.+)\s*$!) {
1433                        my ($remote, $local_ref, $_remote_ref) = ($1, $2, $3);
1434                        die("svn-remote.$remote: remote ref '$_remote_ref' "
1435                            . "must start with 'refs/remotes/'\n")
1436                                unless $_remote_ref =~ m{^refs/remotes/(.+)};
1437                        my $remote_ref = $1;
1438                        $local_ref =~ s{^/}{};
1439                        $r->{$remote}->{fetch}->{$local_ref} = $remote_ref;
1440                        $r->{$remote}->{svm} = {} if $use_svm_props;
1441                } elsif (m!^(.+)\.usesvmprops=\s*(.*)\s*$!) {
1442                        $r->{$1}->{svm} = {};
1443                } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
1444                        $r->{$1}->{url} = $2;
1445                } elsif (m!^(.+)\.(branches|tags)=
1446                           (.*):refs/remotes/(.+)\s*$/!x) {
1447                        my ($p, $g) = ($3, $4);
1448                        my $rs = $r->{$1}->{$2} = {
1449                                          t => $2,
1450                                          remote => $1,
1451                                          path => Git::SVN::GlobSpec->new($p),
1452                                          ref => Git::SVN::GlobSpec->new($g) };
1453                        if (length($rs->{ref}->{right}) != 0) {
1454                                die "The '*' glob character must be the last ",
1455                                    "character of '$g'\n";
1456                        }
1457                }
1458        }
1459
1460        map {
1461                if (defined $r->{$_}->{svm}) {
1462                        my $svm;
1463                        eval {
1464                                my $section = "svn-remote.$_";
1465                                $svm = {
1466                                        source => tmp_config('--get',
1467                                            "$section.svm-source"),
1468                                        replace => tmp_config('--get',
1469                                            "$section.svm-replace"),
1470                                }
1471                        };
1472                        $r->{$_}->{svm} = $svm;
1473                }
1474        } keys %$r;
1475
1476        $r;
1477}
1478
1479sub init_vars {
1480        $_gc_nr = $_gc_period = 1000;
1481        if (defined $_repack || defined $_repack_flags) {
1482               warn "Repack options are obsolete; they have no effect.\n";
1483        }
1484}
1485
1486sub verify_remotes_sanity {
1487        return unless -d $ENV{GIT_DIR};
1488        my %seen;
1489        foreach (command(qw/config -l/)) {
1490                if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
1491                        if ($seen{$1}) {
1492                                die "Remote ref refs/remote/$1 is tracked by",
1493                                    "\n  \"$_\"\nand\n  \"$seen{$1}\"\n",
1494                                    "Please resolve this ambiguity in ",
1495                                    "your git configuration file before ",
1496                                    "continuing\n";
1497                        }
1498                        $seen{$1} = $_;
1499                }
1500        }
1501}
1502
1503sub find_existing_remote {
1504        my ($url, $remotes) = @_;
1505        return undef if $no_reuse_existing;
1506        my $existing;
1507        foreach my $repo_id (keys %$remotes) {
1508                my $u = $remotes->{$repo_id}->{url} or next;
1509                next if $u ne $url;
1510                $existing = $repo_id;
1511                last;
1512        }
1513        $existing;
1514}
1515
1516sub init_remote_config {
1517        my ($self, $url, $no_write) = @_;
1518        $url =~ s!/+$!!; # strip trailing slash
1519        my $r = read_all_remotes();
1520        my $existing = find_existing_remote($url, $r);
1521        if ($existing) {
1522                unless ($no_write) {
1523                        print STDERR "Using existing ",
1524                                     "[svn-remote \"$existing\"]\n";
1525                }
1526                $self->{repo_id} = $existing;
1527        } elsif ($_minimize_url) {
1528                my $min_url = Git::SVN::Ra->new($url)->minimize_url;
1529                $existing = find_existing_remote($min_url, $r);
1530                if ($existing) {
1531                        unless ($no_write) {
1532                                print STDERR "Using existing ",
1533                                             "[svn-remote \"$existing\"]\n";
1534                        }
1535                        $self->{repo_id} = $existing;
1536                }
1537                if ($min_url ne $url) {
1538                        unless ($no_write) {
1539                                print STDERR "Using higher level of URL: ",
1540                                             "$url => $min_url\n";
1541                        }
1542                        my $old_path = $self->{path};
1543                        $self->{path} = $url;
1544                        $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
1545                        if (length $old_path) {
1546                                $self->{path} .= "/$old_path";
1547                        }
1548                        $url = $min_url;
1549                }
1550        }
1551        my $orig_url;
1552        if (!$existing) {
1553                # verify that we aren't overwriting anything:
1554                $orig_url = eval {
1555                        command_oneline('config', '--get',
1556                                        "svn-remote.$self->{repo_id}.url")
1557                };
1558                if ($orig_url && ($orig_url ne $url)) {
1559                        die "svn-remote.$self->{repo_id}.url already set: ",
1560                            "$orig_url\nwanted to set to: $url\n";
1561                }
1562        }
1563        my ($xrepo_id, $xpath) = find_ref($self->refname);
1564        if (defined $xpath) {
1565                die "svn-remote.$xrepo_id.fetch already set to track ",
1566                    "$xpath:refs/remotes/", $self->refname, "\n";
1567        }
1568        unless ($no_write) {
1569                command_noisy('config',
1570                              "svn-remote.$self->{repo_id}.url", $url);
1571                $self->{path} =~ s{^/}{};
1572                command_noisy('config', '--add',
1573                              "svn-remote.$self->{repo_id}.fetch",
1574                              "$self->{path}:".$self->refname);
1575        }
1576        $self->{url} = $url;
1577}
1578
1579sub find_by_url { # repos_root and, path are optional
1580        my ($class, $full_url, $repos_root, $path) = @_;
1581
1582        return undef unless defined $full_url;
1583        remove_username($full_url);
1584        remove_username($repos_root) if defined $repos_root;
1585        my $remotes = read_all_remotes();
1586        if (defined $full_url && defined $repos_root && !defined $path) {
1587                $path = $full_url;
1588                $path =~ s#^\Q$repos_root\E(?:/|$)##;
1589        }
1590        foreach my $repo_id (keys %$remotes) {
1591                my $u = $remotes->{$repo_id}->{url} or next;
1592                remove_username($u);
1593                next if defined $repos_root && $repos_root ne $u;
1594
1595                my $fetch = $remotes->{$repo_id}->{fetch} || {};
1596                foreach (qw/branches tags/) {
1597                        resolve_local_globs($u, $fetch,
1598                                            $remotes->{$repo_id}->{$_});
1599                }
1600                my $p = $path;
1601                my $rwr = rewrite_root({repo_id => $repo_id});
1602                my $svm = $remotes->{$repo_id}->{svm}
1603                        if defined $remotes->{$repo_id}->{svm};
1604                unless (defined $p) {
1605                        $p = $full_url;
1606                        my $z = $u;
1607                        my $prefix = '';
1608                        if ($rwr) {
1609                                $z = $rwr;
1610                        } elsif (defined $svm) {
1611                                $z = $svm->{source};
1612                                $prefix = $svm->{replace};
1613                                $prefix =~ s#^\Q$u\E(?:/|$)##;
1614                                $prefix =~ s#/$##;
1615                        }
1616                        $p =~ s#^\Q$z\E(?:/|$)#$prefix# or next;
1617                }
1618                foreach my $f (keys %$fetch) {
1619                        next if $f ne $p;
1620                        return Git::SVN->new($fetch->{$f}, $repo_id, $f);
1621                }
1622        }
1623        undef;
1624}
1625
1626sub init {
1627        my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
1628        my $self = _new($class, $repo_id, $ref_id, $path);
1629        if (defined $url) {
1630                $self->init_remote_config($url, $no_write);
1631        }
1632        $self;
1633}
1634
1635sub find_ref {
1636        my ($ref_id) = @_;
1637        foreach (command(qw/config -l/)) {
1638                next unless m!^svn-remote\.(.+)\.fetch=
1639                              \s*(.*)\s*:\s*refs/remotes/(.+)\s*$!x;
1640                my ($repo_id, $path, $ref) = ($1, $2, $3);
1641                if ($ref eq $ref_id) {
1642                        $path = '' if ($path =~ m#^\./?#);
1643                        return ($repo_id, $path);
1644                }
1645        }
1646        (undef, undef, undef);
1647}
1648
1649sub new {
1650        my ($class, $ref_id, $repo_id, $path) = @_;
1651        if (defined $ref_id && !defined $repo_id && !defined $path) {
1652                ($repo_id, $path) = find_ref($ref_id);
1653                if (!defined $repo_id) {
1654                        die "Could not find a \"svn-remote.*.fetch\" key ",
1655                            "in the repository configuration matching: ",
1656                            "refs/remotes/$ref_id\n";
1657                }
1658        }
1659        my $self = _new($class, $repo_id, $ref_id, $path);
1660        if (!defined $self->{path} || !length $self->{path}) {
1661                my $fetch = command_oneline('config', '--get',
1662                                            "svn-remote.$repo_id.fetch",
1663                                            ":refs/remotes/$ref_id\$") or
1664                     die "Failed to read \"svn-remote.$repo_id.fetch\" ",
1665                         "\":refs/remotes/$ref_id\$\" in config\n";
1666                ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
1667        }
1668        $self->{url} = command_oneline('config', '--get',
1669                                       "svn-remote.$repo_id.url") or
1670                  die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
1671        $self->rebuild;
1672        $self;
1673}
1674
1675sub refname {
1676        my ($refname) = "refs/remotes/$_[0]->{ref_id}" ;
1677
1678        # It cannot end with a slash /, we'll throw up on this because
1679        # SVN can't have directories with a slash in their name, either:
1680        if ($refname =~ m{/$}) {
1681                die "ref: '$refname' ends with a trailing slash, this is ",
1682                    "not permitted by git nor Subversion\n";
1683        }
1684
1685        # It cannot have ASCII control character space, tilde ~, caret ^,
1686        # colon :, question-mark ?, asterisk *, space, or open bracket [
1687        # anywhere.
1688        #
1689        # Additionally, % must be escaped because it is used for escaping
1690        # and we want our escaped refname to be reversible
1691        $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
1692
1693        # no slash-separated component can begin with a dot .
1694        # /.* becomes /%2E*
1695        $refname =~ s{/\.}{/%2E}g;
1696
1697        # It cannot have two consecutive dots .. anywhere
1698        # .. becomes %2E%2E
1699        $refname =~ s{\.\.}{%2E%2E}g;
1700
1701        return $refname;
1702}
1703
1704sub desanitize_refname {
1705        my ($refname) = @_;
1706        $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
1707        return $refname;
1708}
1709
1710sub svm_uuid {
1711        my ($self) = @_;
1712        return $self->{svm}->{uuid} if $self->svm;
1713        $self->ra;
1714        unless ($self->{svm}) {
1715                die "SVM UUID not cached, and reading remotely failed\n";
1716        }
1717        $self->{svm}->{uuid};
1718}
1719
1720sub svm {
1721        my ($self) = @_;
1722        return $self->{svm} if $self->{svm};
1723        my $svm;
1724        # see if we have it in our config, first:
1725        eval {
1726                my $section = "svn-remote.$self->{repo_id}";
1727                $svm = {
1728                  source => tmp_config('--get', "$section.svm-source"),
1729                  uuid => tmp_config('--get', "$section.svm-uuid"),
1730                  replace => tmp_config('--get', "$section.svm-replace"),
1731                }
1732        };
1733        if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
1734                $self->{svm} = $svm;
1735        }
1736        $self->{svm};
1737}
1738
1739sub _set_svm_vars {
1740        my ($self, $ra) = @_;
1741        return $ra if $self->svm;
1742
1743        my @err = ( "useSvmProps set, but failed to read SVM properties\n",
1744                    "(svm:source, svm:uuid) ",
1745                    "from the following URLs:\n" );
1746        sub read_svm_props {
1747                my ($self, $ra, $path, $r) = @_;
1748                my $props = ($ra->get_dir($path, $r))[2];
1749                my $src = $props->{'svm:source'};
1750                my $uuid = $props->{'svm:uuid'};
1751                return undef if (!$src || !$uuid);
1752
1753                chomp($src, $uuid);
1754
1755                $uuid =~ m{^[0-9a-f\-]{30,}$}
1756                    or die "doesn't look right - svm:uuid is '$uuid'\n";
1757
1758                # the '!' is used to mark the repos_root!/relative/path
1759                $src =~ s{/?!/?}{/};
1760                $src =~ s{/+$}{}; # no trailing slashes please
1761                # username is of no interest
1762                $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
1763
1764                my $replace = $ra->{url};
1765                $replace .= "/$path" if length $path;
1766
1767                my $section = "svn-remote.$self->{repo_id}";
1768                tmp_config("$section.svm-source", $src);
1769                tmp_config("$section.svm-replace", $replace);
1770                tmp_config("$section.svm-uuid", $uuid);
1771                $self->{svm} = {
1772                        source => $src,
1773                        uuid => $uuid,
1774                        replace => $replace
1775                };
1776        }
1777
1778        my $r = $ra->get_latest_revnum;
1779        my $path = $self->{path};
1780        my %tried;
1781        while (length $path) {
1782                unless ($tried{"$self->{url}/$path"}) {
1783                        return $ra if $self->read_svm_props($ra, $path, $r);
1784                        $tried{"$self->{url}/$path"} = 1;
1785                }
1786                $path =~ s#/?[^/]+$##;
1787        }
1788        die "Path: '$path' should be ''\n" if $path ne '';
1789        return $ra if $self->read_svm_props($ra, $path, $r);
1790        $tried{"$self->{url}/$path"} = 1;
1791
1792        if ($ra->{repos_root} eq $self->{url}) {
1793                die @err, (map { "  $_\n" } keys %tried), "\n";
1794        }
1795
1796        # nope, make sure we're connected to the repository root:
1797        my $ok;
1798        my @tried_b;
1799        $path = $ra->{svn_path};
1800        $ra = Git::SVN::Ra->new($ra->{repos_root});
1801        while (length $path) {
1802                unless ($tried{"$ra->{url}/$path"}) {
1803                        $ok = $self->read_svm_props($ra, $path, $r);
1804                        last if $ok;
1805                        $tried{"$ra->{url}/$path"} = 1;
1806                }
1807                $path =~ s#/?[^/]+$##;
1808        }
1809        die "Path: '$path' should be ''\n" if $path ne '';
1810        $ok ||= $self->read_svm_props($ra, $path, $r);
1811        $tried{"$ra->{url}/$path"} = 1;
1812        if (!$ok) {
1813                die @err, (map { "  $_\n" } keys %tried), "\n";
1814        }
1815        Git::SVN::Ra->new($self->{url});
1816}
1817
1818sub svnsync {
1819        my ($self) = @_;
1820        return $self->{svnsync} if $self->{svnsync};
1821
1822        if ($self->no_metadata) {
1823                die "Can't have both 'noMetadata' and ",
1824                    "'useSvnsyncProps' options set!\n";
1825        }
1826        if ($self->rewrite_root) {
1827                die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
1828                    "options set!\n";
1829        }
1830
1831        my $svnsync;
1832        # see if we have it in our config, first:
1833        eval {
1834                my $section = "svn-remote.$self->{repo_id}";
1835
1836                my $url = tmp_config('--get', "$section.svnsync-url");
1837                ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
1838                   die "doesn't look right - svn:sync-from-url is '$url'\n";
1839
1840                my $uuid = tmp_config('--get', "$section.svnsync-uuid");
1841                ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}) or
1842                   die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
1843
1844                $svnsync = { url => $url, uuid => $uuid }
1845        };
1846        if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
1847                return $self->{svnsync} = $svnsync;
1848        }
1849
1850        my $err = "useSvnsyncProps set, but failed to read " .
1851                  "svnsync property: svn:sync-from-";
1852        my $rp = $self->ra->rev_proplist(0);
1853
1854        my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
1855        ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
1856                   die "doesn't look right - svn:sync-from-url is '$url'\n";
1857
1858        my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
1859        ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}) or
1860                   die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
1861
1862        my $section = "svn-remote.$self->{repo_id}";
1863        tmp_config('--add', "$section.svnsync-uuid", $uuid);
1864        tmp_config('--add', "$section.svnsync-url", $url);
1865        return $self->{svnsync} = { url => $url, uuid => $uuid };
1866}
1867
1868# this allows us to memoize our SVN::Ra UUID locally and avoid a
1869# remote lookup (useful for 'git svn log').
1870sub ra_uuid {
1871        my ($self) = @_;
1872        unless ($self->{ra_uuid}) {
1873                my $key = "svn-remote.$self->{repo_id}.uuid";
1874                my $uuid = eval { tmp_config('--get', $key) };
1875                if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/) {
1876                        $self->{ra_uuid} = $uuid;
1877                } else {
1878                        die "ra_uuid called without URL\n" unless $self->{url};
1879                        $self->{ra_uuid} = $self->ra->get_uuid;
1880                        tmp_config('--add', $key, $self->{ra_uuid});
1881                }
1882        }
1883        $self->{ra_uuid};
1884}
1885
1886sub _set_repos_root {
1887        my ($self, $repos_root) = @_;
1888        my $k = "svn-remote.$self->{repo_id}.reposRoot";
1889        $repos_root ||= $self->ra->{repos_root};
1890        tmp_config($k, $repos_root);
1891        $repos_root;
1892}
1893
1894sub repos_root {
1895        my ($self) = @_;
1896        my $k = "svn-remote.$self->{repo_id}.reposRoot";
1897        eval { tmp_config('--get', $k) } || $self->_set_repos_root;
1898}
1899
1900sub ra {
1901        my ($self) = shift;
1902        my $ra = Git::SVN::Ra->new($self->{url});
1903        $self->_set_repos_root($ra->{repos_root});
1904        if ($self->use_svm_props && !$self->{svm}) {
1905                if ($self->no_metadata) {
1906                        die "Can't have both 'noMetadata' and ",
1907                            "'useSvmProps' options set!\n";
1908                } elsif ($self->use_svnsync_props) {
1909                        die "Can't have both 'useSvnsyncProps' and ",
1910                            "'useSvmProps' options set!\n";
1911                }
1912                $ra = $self->_set_svm_vars($ra);
1913                $self->{-want_revprops} = 1;
1914        }
1915        $ra;
1916}
1917
1918sub rel_path {
1919        my ($self) = @_;
1920        my $repos_root = $self->ra->{repos_root};
1921        return $self->{path} if ($self->{url} eq $repos_root);
1922        my $url = $self->{url} .
1923                  (length $self->{path} ? "/$self->{path}" : $self->{path});
1924        $url =~ s!^\Q$repos_root\E(?:/+|$)!!g;
1925        $url;
1926}
1927
1928# prop_walk(PATH, REV, SUB)
1929# -------------------------
1930# Recursively traverse PATH at revision REV and invoke SUB for each
1931# directory that contains a SVN property.  SUB will be invoked as
1932# follows:  &SUB(gs, path, props);  where `gs' is this instance of
1933# Git::SVN, `path' the path to the directory where the properties
1934# `props' were found.  The `path' will be relative to point of checkout,
1935# that is, if url://repo/trunk is the current Git branch, and that
1936# directory contains a sub-directory `d', SUB will be invoked with `/d/'
1937# as `path' (note the trailing `/').
1938sub prop_walk {
1939        my ($self, $path, $rev, $sub) = @_;
1940
1941        $path =~ s#^/##;
1942        my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev);
1943        $path =~ s#^/*#/#g;
1944        my $p = $path;
1945        # Strip the irrelevant part of the path.
1946        $p =~ s#^/+\Q$self->{path}\E(/|$)#/#;
1947        # Ensure the path is terminated by a `/'.
1948        $p =~ s#/*$#/#;
1949
1950        # The properties contain all the internal SVN stuff nobody
1951        # (usually) cares about.
1952        my $interesting_props = 0;
1953        foreach (keys %{$props}) {
1954                # If it doesn't start with `svn:', it must be a
1955                # user-defined property.
1956                ++$interesting_props and next if $_ !~ /^svn:/;
1957                # FIXME: Fragile, if SVN adds new public properties,
1958                # this needs to be updated.
1959                ++$interesting_props if /^svn:(?:ignore|keywords|executable
1960                                                 |eol-style|mime-type
1961                                                 |externals|needs-lock)$/x;
1962        }
1963        &$sub($self, $p, $props) if $interesting_props;
1964
1965        foreach (sort keys %$dirent) {
1966                next if $dirent->{$_}->{kind} != $SVN::Node::dir;
1967                $self->prop_walk($self->{path} . $p . $_, $rev, $sub);
1968        }
1969}
1970
1971sub last_rev { ($_[0]->last_rev_commit)[0] }
1972sub last_commit { ($_[0]->last_rev_commit)[1] }
1973
1974# returns the newest SVN revision number and newest commit SHA1
1975sub last_rev_commit {
1976        my ($self) = @_;
1977        if (defined $self->{last_rev} && defined $self->{last_commit}) {
1978                return ($self->{last_rev}, $self->{last_commit});
1979        }
1980        my $c = ::verify_ref($self->refname.'^0');
1981        if ($c && !$self->use_svm_props && !$self->no_metadata) {
1982                my $rev = (::cmt_metadata($c))[1];
1983                if (defined $rev) {
1984                        ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
1985                        return ($rev, $c);
1986                }
1987        }
1988        my $map_path = $self->map_path;
1989        unless (-e $map_path) {
1990                ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
1991                return (undef, undef);
1992        }
1993        my ($rev, $commit) = $self->rev_map_max(1);
1994        ($self->{last_rev}, $self->{last_commit}) = ($rev, $commit);
1995        return ($rev, $commit);
1996}
1997
1998sub get_fetch_range {
1999        my ($self, $min, $max) = @_;
2000        $max ||= $self->ra->get_latest_revnum;
2001        $min ||= $self->rev_map_max;
2002        (++$min, $max);
2003}
2004
2005sub tmp_config {
2006        my (@args) = @_;
2007        my $old_def_config = "$ENV{GIT_DIR}/svn/config";
2008        my $config = "$ENV{GIT_DIR}/svn/.metadata";
2009        if (! -f $config && -f $old_def_config) {
2010                rename $old_def_config, $config or
2011                       die "Failed rename $old_def_config => $config: $!\n";
2012        }
2013        my $old_config = $ENV{GIT_CONFIG};
2014        $ENV{GIT_CONFIG} = $config;
2015        $@ = undef;
2016        my @ret = eval {
2017                unless (-f $config) {
2018                        mkfile($config);
2019                        open my $fh, '>', $config or
2020                            die "Can't open $config: $!\n";
2021                        print $fh "; This file is used internally by ",
2022                                  "git-svn\n" or die
2023                                  "Couldn't write to $config: $!\n";
2024                        print $fh "; You should not have to edit it\n" or
2025                              die "Couldn't write to $config: $!\n";
2026                        close $fh or die "Couldn't close $config: $!\n";
2027                }
2028                command('config', @args);
2029        };
2030        my $err = $@;
2031        if (defined $old_config) {
2032                $ENV{GIT_CONFIG} = $old_config;
2033        } else {
2034                delete $ENV{GIT_CONFIG};
2035        }
2036        die $err if $err;
2037        wantarray ? @ret : $ret[0];
2038}
2039
2040sub tmp_index_do {
2041        my ($self, $sub) = @_;
2042        my $old_index = $ENV{GIT_INDEX_FILE};
2043        $ENV{GIT_INDEX_FILE} = $self->{index};
2044        $@ = undef;
2045        my @ret = eval {
2046                my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
2047                mkpath([$dir]) unless -d $dir;
2048                &$sub;
2049        };
2050        my $err = $@;
2051        if (defined $old_index) {
2052                $ENV{GIT_INDEX_FILE} = $old_index;
2053        } else {
2054                delete $ENV{GIT_INDEX_FILE};
2055        }
2056        die $err if $err;
2057        wantarray ? @ret : $ret[0];
2058}
2059
2060sub assert_index_clean {
2061        my ($self, $treeish) = @_;
2062
2063        $self->tmp_index_do(sub {
2064                command_noisy('read-tree', $treeish) unless -e $self->{index};
2065                my $x = command_oneline('write-tree');
2066                my ($y) = (command(qw/cat-file commit/, $treeish) =~
2067                           /^tree ($::sha1)/mo);
2068                return if $y eq $x;
2069
2070                warn "Index mismatch: $y != $x\nrereading $treeish\n";
2071                unlink $self->{index} or die "unlink $self->{index}: $!\n";
2072                command_noisy('read-tree', $treeish);
2073                $x = command_oneline('write-tree');
2074                if ($y ne $x) {
2075                        ::fatal "trees ($treeish) $y != $x\n",
2076                                "Something is seriously wrong...";
2077                }
2078        });
2079}
2080
2081sub get_commit_parents {
2082        my ($self, $log_entry) = @_;
2083        my (%seen, @ret, @tmp);
2084        # legacy support for 'set-tree'; this is only used by set_tree_cb:
2085        if (my $ip = $self->{inject_parents}) {
2086                if (my $commit = delete $ip->{$log_entry->{revision}}) {
2087                        push @tmp, $commit;
2088                }
2089        }
2090        if (my $cur = ::verify_ref($self->refname.'^0')) {
2091                push @tmp, $cur;
2092        }
2093        if (my $ipd = $self->{inject_parents_dcommit}) {
2094                if (my $commit = delete $ipd->{$log_entry->{revision}}) {
2095                        push @tmp, @$commit;
2096                }
2097        }
2098        push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
2099        while (my $p = shift @tmp) {
2100                next if $seen{$p};
2101                $seen{$p} = 1;
2102                push @ret, $p;
2103                # MAXPARENT is defined to 16 in commit-tree.c:
2104                last if @ret >= 16;
2105        }
2106        if (@tmp) {
2107                die "r$log_entry->{revision}: No room for parents:\n\t",
2108                    join("\n\t", @tmp), "\n";
2109        }
2110        @ret;
2111}
2112
2113sub rewrite_root {
2114        my ($self) = @_;
2115        return $self->{-rewrite_root} if exists $self->{-rewrite_root};
2116        my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
2117        my $rwr = eval { command_oneline(qw/config --get/, $k) };
2118        if ($rwr) {
2119                $rwr =~ s#/+$##;
2120                if ($rwr !~ m#^[a-z\+]+://#) {
2121                        die "$rwr is not a valid URL (key: $k)\n";
2122                }
2123        }
2124        $self->{-rewrite_root} = $rwr;
2125}
2126
2127sub metadata_url {
2128        my ($self) = @_;
2129        ($self->rewrite_root || $self->{url}) .
2130           (length $self->{path} ? '/' . $self->{path} : '');
2131}
2132
2133sub full_url {
2134        my ($self) = @_;
2135        $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
2136}
2137
2138
2139sub set_commit_header_env {
2140        my ($log_entry) = @_;
2141        my %env;
2142        foreach my $ned (qw/NAME EMAIL DATE/) {
2143                foreach my $ac (qw/AUTHOR COMMITTER/) {
2144                        $env{"GIT_${ac}_${ned}"} = $ENV{"GIT_${ac}_${ned}"};
2145                }
2146        }
2147
2148        $ENV{GIT_AUTHOR_NAME} = $log_entry->{name};
2149        $ENV{GIT_AUTHOR_EMAIL} = $log_entry->{email};
2150        $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
2151
2152        $ENV{GIT_COMMITTER_NAME} = (defined $log_entry->{commit_name})
2153                                                ? $log_entry->{commit_name}
2154                                                : $log_entry->{name};
2155        $ENV{GIT_COMMITTER_EMAIL} = (defined $log_entry->{commit_email})
2156                                                ? $log_entry->{commit_email}
2157                                                : $log_entry->{email};
2158        \%env;
2159}
2160
2161sub restore_commit_header_env {
2162        my ($env) = @_;
2163        foreach my $ned (qw/NAME EMAIL DATE/) {
2164                foreach my $ac (qw/AUTHOR COMMITTER/) {
2165                        my $k = "GIT_${ac}_${ned}";
2166                        if (defined $env->{$k}) {
2167                                $ENV{$k} = $env->{$k};
2168                        } else {
2169                                delete $ENV{$k};
2170                        }
2171                }
2172        }
2173}
2174
2175sub gc {
2176        command_noisy('gc', '--auto');
2177};
2178
2179sub do_git_commit {
2180        my ($self, $log_entry) = @_;
2181        my $lr = $self->last_rev;
2182        if (defined $lr && $lr >= $log_entry->{revision}) {
2183                die "Last fetched revision of ", $self->refname,
2184                    " was r$lr, but we are about to fetch: ",
2185                    "r$log_entry->{revision}!\n";
2186        }
2187        if (my $c = $self->rev_map_get($log_entry->{revision})) {
2188                croak "$log_entry->{revision} = $c already exists! ",
2189                      "Why are we refetching it?\n";
2190        }
2191        my $old_env = set_commit_header_env($log_entry);
2192        my $tree = $log_entry->{tree};
2193        if (!defined $tree) {
2194                $tree = $self->tmp_index_do(sub {
2195                                            command_oneline('write-tree') });
2196        }
2197        die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
2198
2199        my @exec = ('git-commit-tree', $tree);
2200        foreach ($self->get_commit_parents($log_entry)) {
2201                push @exec, '-p', $_;
2202        }
2203        defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
2204                                                                   or croak $!;
2205        print $msg_fh $log_entry->{log} or croak $!;
2206        restore_commit_header_env($old_env);
2207        unless ($self->no_metadata) {
2208                print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
2209                              or croak $!;
2210        }
2211        $msg_fh->flush == 0 or croak $!;
2212        close $msg_fh or croak $!;
2213        chomp(my $commit = do { local $/; <$out_fh> });
2214        close $out_fh or croak $!;
2215        waitpid $pid, 0;
2216        croak $? if $?;
2217        if ($commit !~ /^$::sha1$/o) {
2218                die "Failed to commit, invalid sha1: $commit\n";
2219        }
2220
2221        $self->rev_map_set($log_entry->{revision}, $commit, 1);
2222
2223        $self->{last_rev} = $log_entry->{revision};
2224        $self->{last_commit} = $commit;
2225        print "r$log_entry->{revision}";
2226        if (defined $log_entry->{svm_revision}) {
2227                 print " (\@$log_entry->{svm_revision})";
2228                 $self->rev_map_set($log_entry->{svm_revision}, $commit,
2229                                   0, $self->svm_uuid);
2230        }
2231        print " = $commit ($self->{ref_id})\n";
2232        if (--$_gc_nr == 0) {
2233                $_gc_nr = $_gc_period;
2234                gc();
2235        }
2236        return $commit;
2237}
2238
2239sub match_paths {
2240        my ($self, $paths, $r) = @_;
2241        return 1 if $self->{path} eq '';
2242        if (my $path = $paths->{"/$self->{path}"}) {
2243                return ($path->{action} eq 'D') ? 0 : 1;
2244        }
2245        $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
2246        if (grep /$self->{path_regex}/, keys %$paths) {
2247                return 1;
2248        }
2249        my $c = '';
2250        foreach (split m#/#, $self->{path}) {
2251                $c .= "/$_";
2252                next unless ($paths->{$c} &&
2253                             ($paths->{$c}->{action} =~ /^[AR]$/));
2254                if ($self->ra->check_path($self->{path}, $r) ==
2255                    $SVN::Node::dir) {
2256                        return 1;
2257                }
2258        }
2259        return 0;
2260}
2261
2262sub find_parent_branch {
2263        my ($self, $paths, $rev) = @_;
2264        return undef unless $self->follow_parent;
2265        unless (defined $paths) {
2266                my $err_handler = $SVN::Error::handler;
2267                $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
2268                $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1, sub {
2269                                   $paths =
2270                                      Git::SVN::Ra::dup_changed_paths($_[0]) });
2271                $SVN::Error::handler = $err_handler;
2272        }
2273        return undef unless defined $paths;
2274
2275        # look for a parent from another branch:
2276        my @b_path_components = split m#/#, $self->rel_path;
2277        my @a_path_components;
2278        my $i;
2279        while (@b_path_components) {
2280                $i = $paths->{'/'.join('/', @b_path_components)};
2281                last if $i && defined $i->{copyfrom_path};
2282                unshift(@a_path_components, pop(@b_path_components));
2283        }
2284        return undef unless defined $i && defined $i->{copyfrom_path};
2285        my $branch_from = $i->{copyfrom_path};
2286        if (@a_path_components) {
2287                print STDERR "branch_from: $branch_from => ";
2288                $branch_from .= '/'.join('/', @a_path_components);
2289                print STDERR $branch_from, "\n";
2290        }
2291        my $r = $i->{copyfrom_rev};
2292        my $repos_root = $self->ra->{repos_root};
2293        my $url = $self->ra->{url};
2294        my $new_url = $repos_root . $branch_from;
2295        print STDERR  "Found possible branch point: ",
2296                      "$new_url => ", $self->full_url, ", $r\n";
2297        $branch_from =~ s#^/##;
2298        my $gs = Git::SVN->find_by_url($new_url, $repos_root, $branch_from);
2299        unless ($gs) {
2300                my $ref_id = $self->{ref_id};
2301                $ref_id =~ s/\@\d+$//;
2302                $ref_id .= "\@$r";
2303                # just grow a tail if we're not unique enough :x
2304                $ref_id .= '-' while find_ref($ref_id);
2305                print STDERR "Initializing parent: $ref_id\n";
2306                my ($u, $p, $repo_id) = ($new_url, '', $ref_id);
2307                if ($u =~ s#^\Q$url\E(/|$)##) {
2308                        $p = $u;
2309                        $u = $url;
2310                        $repo_id = $self->{repo_id};
2311                }
2312                $gs = Git::SVN->init($u, $p, $repo_id, $ref_id, 1);
2313        }
2314        my ($r0, $parent) = $gs->find_rev_before($r, 1);
2315        if (!defined $r0 || !defined $parent) {
2316                my ($base, $head) = parse_revision_argument(0, $r);
2317                if ($base <= $r) {
2318                        $gs->fetch($base, $r);
2319                }
2320                ($r0, $parent) = $gs->last_rev_commit;
2321        }
2322        if (defined $r0 && defined $parent) {
2323                print STDERR "Found branch parent: ($self->{ref_id}) $parent\n";
2324                my $ed;
2325                if ($self->ra->can_do_switch) {
2326                        $self->assert_index_clean($parent);
2327                        print STDERR "Following parent with do_switch\n";
2328                        # do_switch works with svn/trunk >= r22312, but that
2329                        # is not included with SVN 1.4.3 (the latest version
2330                        # at the moment), so we can't rely on it
2331                        $self->{last_commit} = $parent;
2332                        $ed = SVN::Git::Fetcher->new($self);
2333                        $gs->ra->gs_do_switch($r0, $rev, $gs,
2334                                              $self->full_url, $ed)
2335                          or die "SVN connection failed somewhere...\n";
2336                } elsif ($self->ra->trees_match($new_url, $r0,
2337                                                $self->full_url, $rev)) {
2338                        print STDERR "Trees match:\n",
2339                                     "  $new_url\@$r0\n",
2340                                     "  ${\$self->full_url}\@$rev\n",
2341                                     "Following parent with no changes\n";
2342                        $self->tmp_index_do(sub {
2343                            command_noisy('read-tree', $parent);
2344                        });
2345                        $self->{last_commit} = $parent;
2346                } else {
2347                        print STDERR "Following parent with do_update\n";
2348                        $ed = SVN::Git::Fetcher->new($self);
2349                        $self->ra->gs_do_update($rev, $rev, $self, $ed)
2350                          or die "SVN connection failed somewhere...\n";
2351                }
2352                print STDERR "Successfully followed parent\n";
2353                return $self->make_log_entry($rev, [$parent], $ed);
2354        }
2355        return undef;
2356}
2357
2358sub do_fetch {
2359        my ($self, $paths, $rev) = @_;
2360        my $ed;
2361        my ($last_rev, @parents);
2362        if (my $lc = $self->last_commit) {
2363                # we can have a branch that was deleted, then re-added
2364                # under the same name but copied from another path, in
2365                # which case we'll have multiple parents (we don't
2366                # want to break the original ref, nor lose copypath info):
2367                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2368                        push @{$log_entry->{parents}}, $lc;
2369                        return $log_entry;
2370                }
2371                $ed = SVN::Git::Fetcher->new($self);
2372                $last_rev = $self->{last_rev};
2373                $ed->{c} = $lc;
2374                @parents = ($lc);
2375        } else {
2376                $last_rev = $rev;
2377                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2378                        return $log_entry;
2379                }
2380                $ed = SVN::Git::Fetcher->new($self);
2381        }
2382        unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
2383                die "SVN connection failed somewhere...\n";
2384        }
2385        $self->make_log_entry($rev, \@parents, $ed);
2386}
2387
2388sub get_untracked {
2389        my ($self, $ed) = @_;
2390        my @out;
2391        my $h = $ed->{empty};
2392        foreach (sort keys %$h) {
2393                my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
2394                push @out, "  $act: " . uri_encode($_);
2395                warn "W: $act: $_\n";
2396        }
2397        foreach my $t (qw/dir_prop file_prop/) {
2398                $h = $ed->{$t} or next;
2399                foreach my $path (sort keys %$h) {
2400                        my $ppath = $path eq '' ? '.' : $path;
2401                        foreach my $prop (sort keys %{$h->{$path}}) {
2402                                next if $SKIP_PROP{$prop};
2403                                my $v = $h->{$path}->{$prop};
2404                                my $t_ppath_prop = "$t: " .
2405                                                    uri_encode($ppath) . ' ' .
2406                                                    uri_encode($prop);
2407                                if (defined $v) {
2408                                        push @out, "  +$t_ppath_prop " .
2409                                                   uri_encode($v);
2410                                } else {
2411                                        push @out, "  -$t_ppath_prop";
2412                                }
2413                        }
2414                }
2415        }
2416        foreach my $t (qw/absent_file absent_directory/) {
2417                $h = $ed->{$t} or next;
2418                foreach my $parent (sort keys %$h) {
2419                        foreach my $path (sort @{$h->{$parent}}) {
2420                                push @out, "  $t: " .
2421                                           uri_encode("$parent/$path");
2422                                warn "W: $t: $parent/$path ",
2423                                     "Insufficient permissions?\n";
2424                        }
2425                }
2426        }
2427        \@out;
2428}
2429
2430sub parse_svn_date {
2431        my $date = shift || return '+0000 1970-01-01 00:00:00';
2432        my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2433                                            (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x) or
2434                                         croak "Unable to parse date: $date\n";
2435        "+0000 $Y-$m-$d $H:$M:$S";
2436}
2437
2438sub check_author {
2439        my ($author) = @_;
2440        if (!defined $author || length $author == 0) {
2441                $author = '(no author)';
2442        } elsif (defined $::_authors && ! defined $::users{$author}) {
2443                die "Author: $author not defined in $::_authors file\n";
2444        }
2445        $author;
2446}
2447
2448sub make_log_entry {
2449        my ($self, $rev, $parents, $ed) = @_;
2450        my $untracked = $self->get_untracked($ed);
2451
2452        open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
2453        print $un "r$rev\n" or croak $!;
2454        print $un $_, "\n" foreach @$untracked;
2455        my %log_entry = ( parents => $parents || [], revision => $rev,
2456                          log => '');
2457
2458        my $headrev;
2459        my $logged = delete $self->{logged_rev_props};
2460        if (!$logged || $self->{-want_revprops}) {
2461                my $rp = $self->ra->rev_proplist($rev);
2462                foreach (sort keys %$rp) {
2463                        my $v = $rp->{$_};
2464                        if (/^svn:(author|date|log)$/) {
2465                                $log_entry{$1} = $v;
2466                        } elsif ($_ eq 'svm:headrev') {
2467                                $headrev = $v;
2468                        } else {
2469                                print $un "  rev_prop: ", uri_encode($_), ' ',
2470                                          uri_encode($v), "\n";
2471                        }
2472                }
2473        } else {
2474                map { $log_entry{$_} = $logged->{$_} } keys %$logged;
2475        }
2476        close $un or croak $!;
2477
2478        $log_entry{date} = parse_svn_date($log_entry{date});
2479        $log_entry{log} .= "\n";
2480        my $author = $log_entry{author} = check_author($log_entry{author});
2481        my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
2482                                                       : ($author, undef);
2483
2484        my ($commit_name, $commit_email) = ($name, $email);
2485        if ($_use_log_author) {
2486                my $name_field;
2487                if ($log_entry{log} =~ /From:\s+(.*\S)\s*\n/i) {
2488                        $name_field = $1;
2489                } elsif ($log_entry{log} =~ /Signed-off-by:\s+(.*\S)\s*\n/i) {
2490                        $name_field = $1;
2491                }
2492                if (!defined $name_field) {
2493                        if (!defined $email) {
2494                                $email = $name;
2495                        }
2496                } elsif ($name_field =~ /(.*?)\s+<(.*)>/) {
2497                        ($name, $email) = ($1, $2);
2498                } elsif ($name_field =~ /(.*)@/) {
2499                        ($name, $email) = ($1, $name_field);
2500                } else {
2501                        ($name, $email) = ($name_field, $name_field);
2502                }
2503        }
2504        if (defined $headrev && $self->use_svm_props) {
2505                if ($self->rewrite_root) {
2506                        die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
2507                            "options set!\n";
2508                }
2509                my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$};
2510                # we don't want "SVM: initializing mirror for junk" ...
2511                return undef if $r == 0;
2512                my $svm = $self->svm;
2513                if ($uuid ne $svm->{uuid}) {
2514                        die "UUID mismatch on SVM path:\n",
2515                            "expected: $svm->{uuid}\n",
2516                            "     got: $uuid\n";
2517                }
2518                my $full_url = $self->full_url;
2519                $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
2520                             die "Failed to replace '$svm->{replace}' with ",
2521                                 "'$svm->{source}' in $full_url\n";
2522                # throw away username for storing in records
2523                remove_username($full_url);
2524                $log_entry{metadata} = "$full_url\@$r $uuid";
2525                $log_entry{svm_revision} = $r;
2526                $email ||= "$author\@$uuid";
2527                $commit_email ||= "$author\@$uuid";
2528        } elsif ($self->use_svnsync_props) {
2529                my $full_url = $self->svnsync->{url};
2530                $full_url .= "/$self->{path}" if length $self->{path};
2531                remove_username($full_url);
2532                my $uuid = $self->svnsync->{uuid};
2533                $log_entry{metadata} = "$full_url\@$rev $uuid";
2534                $email ||= "$author\@$uuid";
2535                $commit_email ||= "$author\@$uuid";
2536        } else {
2537                my $url = $self->metadata_url;
2538                remove_username($url);
2539                $log_entry{metadata} = "$url\@$rev " .
2540                                       $self->ra->get_uuid;
2541                $email ||= "$author\@" . $self->ra->get_uuid;
2542                $commit_email ||= "$author\@" . $self->ra->get_uuid;
2543        }
2544        $log_entry{name} = $name;
2545        $log_entry{email} = $email;
2546        $log_entry{commit_name} = $commit_name;
2547        $log_entry{commit_email} = $commit_email;
2548        \%log_entry;
2549}
2550
2551sub fetch {
2552        my ($self, $min_rev, $max_rev, @parents) = @_;
2553        my ($last_rev, $last_commit) = $self->last_rev_commit;
2554        my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
2555        $self->ra->gs_fetch_loop_common($base, $head, [$self]);
2556}
2557
2558sub set_tree_cb {
2559        my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
2560        $self->{inject_parents} = { $rev => $tree };
2561        $self->fetch(undef, undef);
2562}
2563
2564sub set_tree {
2565        my ($self, $tree) = (shift, shift);
2566        my $log_entry = ::get_commit_entry($tree);
2567        unless ($self->{last_rev}) {
2568                fatal("Must have an existing revision to commit");
2569        }
2570        my %ed_opts = ( r => $self->{last_rev},
2571                        log => $log_entry->{log},
2572                        ra => $self->ra,
2573                        tree_a => $self->{last_commit},
2574                        tree_b => $tree,
2575                        editor_cb => sub {
2576                               $self->set_tree_cb($log_entry, $tree, @_) },
2577                        svn_path => $self->{path} );
2578        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
2579                print "No changes\nr$self->{last_rev} = $tree\n";
2580        }
2581}
2582
2583sub rebuild_from_rev_db {
2584        my ($self, $path) = @_;
2585        my $r = -1;
2586        open my $fh, '<', $path or croak "open: $!";
2587        binmode $fh or croak "binmode: $!";
2588        while (<$fh>) {
2589                length($_) == 41 or croak "inconsistent size in ($_) != 41";
2590                chomp($_);
2591                ++$r;
2592                next if $_ eq ('0' x 40);
2593                $self->rev_map_set($r, $_);
2594                print "r$r = $_\n";
2595        }
2596        close $fh or croak "close: $!";
2597        unlink $path or croak "unlink: $!";
2598}
2599
2600sub rebuild {
2601        my ($self) = @_;
2602        my $map_path = $self->map_path;
2603        return if (-e $map_path && ! -z $map_path);
2604        return unless ::verify_ref($self->refname.'^0');
2605        if ($self->use_svm_props || $self->no_metadata) {
2606                my $rev_db = $self->rev_db_path;
2607                $self->rebuild_from_rev_db($rev_db);
2608                if ($self->use_svm_props) {
2609                        my $svm_rev_db = $self->rev_db_path($self->svm_uuid);
2610                        $self->rebuild_from_rev_db($svm_rev_db);
2611                }
2612                $self->unlink_rev_db_symlink;
2613                return;
2614        }
2615        print "Rebuilding $map_path ...\n";
2616        my ($log, $ctx) =
2617            command_output_pipe(qw/rev-list --pretty=raw --no-color --reverse/,
2618                                $self->refname, '--');
2619        my $metadata_url = $self->metadata_url;
2620        remove_username($metadata_url);
2621        my $svn_uuid = $self->ra_uuid;
2622        my $c;
2623        while (<$log>) {
2624                if ( m{^commit ($::sha1)$} ) {
2625                        $c = $1;
2626                        next;
2627                }
2628                next unless s{^\s*(git-svn-id:)}{$1};
2629                my ($url, $rev, $uuid) = ::extract_metadata($_);
2630                remove_username($url);
2631
2632                # ignore merges (from set-tree)
2633                next if (!defined $rev || !$uuid);
2634
2635                # if we merged or otherwise started elsewhere, this is
2636                # how we break out of it
2637                if (($uuid ne $svn_uuid) ||
2638                    ($metadata_url && $url && ($url ne $metadata_url))) {
2639                        next;
2640                }
2641
2642                $self->rev_map_set($rev, $c);
2643                print "r$rev = $c\n";
2644        }
2645        command_close_pipe($log, $ctx);
2646        print "Done rebuilding $map_path\n";
2647        my $rev_db_path = $self->rev_db_path;
2648        if (-f $self->rev_db_path) {
2649                unlink $self->rev_db_path or croak "unlink: $!";
2650        }
2651        $self->unlink_rev_db_symlink;
2652}
2653
2654# rev_map:
2655# Tie::File seems to be prone to offset errors if revisions get sparse,
2656# it's not that fast, either.  Tie::File is also not in Perl 5.6.  So
2657# one of my favorite modules is out :<  Next up would be one of the DBM
2658# modules, but I'm not sure which is most portable...
2659#
2660# This is the replacement for the rev_db format, which was too big
2661# and inefficient for large repositories with a lot of sparse history
2662# (mainly tags)
2663#
2664# The format is this:
2665#   - 24 bytes for every record,
2666#     * 4 bytes for the integer representing an SVN revision number
2667#     * 20 bytes representing the sha1 of a git commit
2668#   - No empty padding records like the old format
2669#     (except the last record, which can be overwritten)
2670#   - new records are written append-only since SVN revision numbers
2671#     increase monotonically
2672#   - lookups on SVN revision number are done via a binary search
2673#   - Piping the file to xxd -c24 is a good way of dumping it for
2674#     viewing or editing (piped back through xxd -r), should the need
2675#     ever arise.
2676#   - The last record can be padding revision with an all-zero sha1
2677#     This is used to optimize fetch performance when using multiple
2678#     "fetch" directives in .git/config
2679#
2680# These files are disposable unless noMetadata or useSvmProps is set
2681
2682sub _rev_map_set {
2683        my ($fh, $rev, $commit) = @_;
2684
2685        binmode $fh or croak "binmode: $!";
2686        my $size = (stat($fh))[7];
2687        ($size % 24) == 0 or croak "inconsistent size: $size";
2688
2689        my $wr_offset = 0;
2690        if ($size > 0) {
2691                sysseek($fh, -24, SEEK_END) or croak "seek: $!";
2692                my $read = sysread($fh, my $buf, 24) or croak "read: $!";
2693                $read == 24 or croak "read only $read bytes (!= 24)";
2694                my ($last_rev, $last_commit) = unpack(rev_map_fmt, $buf);
2695                if ($last_commit eq ('0' x40)) {
2696                        if ($size >= 48) {
2697                                sysseek($fh, -48, SEEK_END) or croak "seek: $!";
2698                                $read = sysread($fh, $buf, 24) or
2699                                    croak "read: $!";
2700                                $read == 24 or
2701                                    croak "read only $read bytes (!= 24)";
2702                                ($last_rev, $last_commit) =
2703                                    unpack(rev_map_fmt, $buf);
2704                                if ($last_commit eq ('0' x40)) {
2705                                        croak "inconsistent .rev_map\n";
2706                                }
2707                        }
2708                        if ($last_rev >= $rev) {
2709                                croak "last_rev is higher!: $last_rev >= $rev";
2710                        }
2711                        $wr_offset = -24;
2712                }
2713        }
2714        sysseek($fh, $wr_offset, SEEK_END) or croak "seek: $!";
2715        syswrite($fh, pack(rev_map_fmt, $rev, $commit), 24) == 24 or
2716          croak "write: $!";
2717}
2718
2719sub mkfile {
2720        my ($path) = @_;
2721        unless (-e $path) {
2722                my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
2723                mkpath([$dir]) unless -d $dir;
2724                open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
2725                close $fh or die "Couldn't close (create) $path: $!\n";
2726        }
2727}
2728
2729sub rev_map_set {
2730        my ($self, $rev, $commit, $update_ref, $uuid) = @_;
2731        length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
2732        my $db = $self->map_path($uuid);
2733        my $db_lock = "$db.lock";
2734        my $sig;
2735        if ($update_ref) {
2736                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
2737                            $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
2738        }
2739        mkfile($db);
2740
2741        $LOCKFILES{$db_lock} = 1;
2742        my $sync;
2743        # both of these options make our .rev_db file very, very important
2744        # and we can't afford to lose it because rebuild() won't work
2745        if ($self->use_svm_props || $self->no_metadata) {
2746                $sync = 1;
2747                copy($db, $db_lock) or die "rev_map_set(@_): ",
2748                                           "Failed to copy: ",
2749                                           "$db => $db_lock ($!)\n";
2750        } else {
2751                rename $db, $db_lock or die "rev_map_set(@_): ",
2752                                            "Failed to rename: ",
2753                                            "$db => $db_lock ($!)\n";
2754        }
2755
2756        sysopen(my $fh, $db_lock, O_RDWR | O_CREAT)
2757             or croak "Couldn't open $db_lock: $!\n";
2758        _rev_map_set($fh, $rev, $commit);
2759        if ($sync) {
2760                $fh->flush or die "Couldn't flush $db_lock: $!\n";
2761                $fh->sync or die "Couldn't sync $db_lock: $!\n";
2762        }
2763        close $fh or croak $!;
2764        if ($update_ref) {
2765                $_head = $self;
2766                command_noisy('update-ref', '-m', "r$rev",
2767                              $self->refname, $commit);
2768        }
2769        rename $db_lock, $db or die "rev_map_set(@_): ", "Failed to rename: ",
2770                                    "$db_lock => $db ($!)\n";
2771        delete $LOCKFILES{$db_lock};
2772        if ($update_ref) {
2773                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
2774                            $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
2775                kill $sig, $$ if defined $sig;
2776        }
2777}
2778
2779# If want_commit, this will return an array of (rev, commit) where
2780# commit _must_ be a valid commit in the archive.
2781# Otherwise, it'll return the max revision (whether or not the
2782# commit is valid or just a 0x40 placeholder).
2783sub rev_map_max {
2784        my ($self, $want_commit) = @_;
2785        $self->rebuild;
2786        my $map_path = $self->map_path;
2787        stat $map_path or return $want_commit ? (0, undef) : 0;
2788        sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
2789        binmode $fh or croak "binmode: $!";
2790        my $size = (stat($fh))[7];
2791        ($size % 24) == 0 or croak "inconsistent size: $size";
2792
2793        if ($size == 0) {
2794                close $fh or croak "close: $!";
2795                return $want_commit ? (0, undef) : 0;
2796        }
2797
2798        sysseek($fh, -24, SEEK_END) or croak "seek: $!";
2799        sysread($fh, my $buf, 24) == 24 or croak "read: $!";
2800        my ($r, $c) = unpack(rev_map_fmt, $buf);
2801        if ($want_commit && $c eq ('0' x40)) {
2802                if ($size < 48) {
2803                        return $want_commit ? (0, undef) : 0;
2804                }
2805                sysseek($fh, -48, SEEK_END) or croak "seek: $!";
2806                sysread($fh, $buf, 24) == 24 or croak "read: $!";
2807                ($r, $c) = unpack(rev_map_fmt, $buf);
2808                if ($c eq ('0'x40)) {
2809                        croak "Penultimate record is all-zeroes in $map_path";
2810                }
2811        }
2812        close $fh or croak "close: $!";
2813        $want_commit ? ($r, $c) : $r;
2814}
2815
2816sub rev_map_get {
2817        my ($self, $rev, $uuid) = @_;
2818        my $map_path = $self->map_path($uuid);
2819        return undef unless -e $map_path;
2820
2821        sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
2822        binmode $fh or croak "binmode: $!";
2823        my $size = (stat($fh))[7];
2824        ($size % 24) == 0 or croak "inconsistent size: $size";
2825
2826        if ($size == 0) {
2827                close $fh or croak "close: $fh";
2828                return undef;
2829        }
2830
2831        my ($l, $u) = (0, $size - 24);
2832        my ($r, $c, $buf);
2833
2834        while ($l <= $u) {
2835                my $i = int(($l/24 + $u/24) / 2) * 24;
2836                sysseek($fh, $i, SEEK_SET) or croak "seek: $!";
2837                sysread($fh, my $buf, 24) == 24 or croak "read: $!";
2838                my ($r, $c) = unpack('NH40', $buf);
2839
2840                if ($r < $rev) {
2841                        $l = $i + 24;
2842                } elsif ($r > $rev) {
2843                        $u = $i - 24;
2844                } else { # $r == $rev
2845                        close($fh) or croak "close: $!";
2846                        return $c eq ('0' x 40) ? undef : $c;
2847                }
2848        }
2849        close($fh) or croak "close: $!";
2850        undef;
2851}
2852
2853# Finds the first svn revision that exists on (if $eq_ok is true) or
2854# before $rev for the current branch.  It will not search any lower
2855# than $min_rev.  Returns the git commit hash and svn revision number
2856# if found, else (undef, undef).
2857sub find_rev_before {
2858        my ($self, $rev, $eq_ok, $min_rev) = @_;
2859        --$rev unless $eq_ok;
2860        $min_rev ||= 1;
2861        while ($rev >= $min_rev) {
2862                if (my $c = $self->rev_map_get($rev)) {
2863                        return ($rev, $c);
2864                }
2865                --$rev;
2866        }
2867        return (undef, undef);
2868}
2869
2870# Finds the first svn revision that exists on (if $eq_ok is true) or
2871# after $rev for the current branch.  It will not search any higher
2872# than $max_rev.  Returns the git commit hash and svn revision number
2873# if found, else (undef, undef).
2874sub find_rev_after {
2875        my ($self, $rev, $eq_ok, $max_rev) = @_;
2876        ++$rev unless $eq_ok;
2877        $max_rev ||= $self->rev_map_max;
2878        while ($rev <= $max_rev) {
2879                if (my $c = $self->rev_map_get($rev)) {
2880                        return ($rev, $c);
2881                }
2882                ++$rev;
2883        }
2884        return (undef, undef);
2885}
2886
2887sub _new {
2888        my ($class, $repo_id, $ref_id, $path) = @_;
2889        unless (defined $repo_id && length $repo_id) {
2890                $repo_id = $Git::SVN::default_repo_id;
2891        }
2892        unless (defined $ref_id && length $ref_id) {
2893                $_[2] = $ref_id = $Git::SVN::default_ref_id;
2894        }
2895        $_[1] = $repo_id;
2896        my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
2897        $_[3] = $path = '' unless (defined $path);
2898        mkpath(["$ENV{GIT_DIR}/svn"]);
2899        bless {
2900                ref_id => $ref_id, dir => $dir, index => "$dir/index",
2901                path => $path, config => "$ENV{GIT_DIR}/svn/config",
2902                map_root => "$dir/.rev_map", repo_id => $repo_id }, $class;
2903}
2904
2905# for read-only access of old .rev_db formats
2906sub unlink_rev_db_symlink {
2907        my ($self) = @_;
2908        my $link = $self->rev_db_path;
2909        $link =~ s/\.[\w-]+$// or croak "missing UUID at the end of $link";
2910        if (-l $link) {
2911                unlink $link or croak "unlink: $link failed!";
2912        }
2913}
2914
2915sub rev_db_path {
2916        my ($self, $uuid) = @_;
2917        my $db_path = $self->map_path($uuid);
2918        $db_path =~ s{/\.rev_map\.}{/\.rev_db\.}
2919            or croak "map_path: $db_path does not contain '/.rev_map.' !";
2920        $db_path;
2921}
2922
2923# the new replacement for .rev_db
2924sub map_path {
2925        my ($self, $uuid) = @_;
2926        $uuid ||= $self->ra_uuid;
2927        "$self->{map_root}.$uuid";
2928}
2929
2930sub uri_encode {
2931        my ($f) = @_;
2932        $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
2933        $f
2934}
2935
2936sub remove_username {
2937        $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
2938}
2939
2940package Git::SVN::Prompt;
2941use strict;
2942use warnings;
2943require SVN::Core;
2944use vars qw/$_no_auth_cache $_username/;
2945
2946sub simple {
2947        my ($cred, $realm, $default_username, $may_save, $pool) = @_;
2948        $may_save = undef if $_no_auth_cache;
2949        $default_username = $_username if defined $_username;
2950        if (defined $default_username && length $default_username) {
2951                if (defined $realm && length $realm) {
2952                        print STDERR "Authentication realm: $realm\n";
2953                        STDERR->flush;
2954                }
2955                $cred->username($default_username);
2956        } else {
2957                username($cred, $realm, $may_save, $pool);
2958        }
2959        $cred->password(_read_password("Password for '" .
2960                                       $cred->username . "': ", $realm));
2961        $cred->may_save($may_save);
2962        $SVN::_Core::SVN_NO_ERROR;
2963}
2964
2965sub ssl_server_trust {
2966        my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
2967        $may_save = undef if $_no_auth_cache;
2968        print STDERR "Error validating server certificate for '$realm':\n";
2969        {
2970                no warnings 'once';
2971                # All variables SVN::Auth::SSL::* are used only once,
2972                # so we're shutting up Perl warnings about this.
2973                if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
2974                        print STDERR " - The certificate is not issued ",
2975                            "by a trusted authority. Use the\n",
2976                            "   fingerprint to validate ",
2977                            "the certificate manually!\n";
2978                }
2979                if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
2980                        print STDERR " - The certificate hostname ",
2981                            "does not match.\n";
2982                }
2983                if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
2984                        print STDERR " - The certificate is not yet valid.\n";
2985                }
2986                if ($failures & $SVN::Auth::SSL::EXPIRED) {
2987                        print STDERR " - The certificate has expired.\n";
2988                }
2989                if ($failures & $SVN::Auth::SSL::OTHER) {
2990                        print STDERR " - The certificate has ",
2991                            "an unknown error.\n";
2992                }
2993        } # no warnings 'once'
2994        printf STDERR
2995                "Certificate information:\n".
2996                " - Hostname: %s\n".
2997                " - Valid: from %s until %s\n".
2998                " - Issuer: %s\n".
2999                " - Fingerprint: %s\n",
3000                map $cert_info->$_, qw(hostname valid_from valid_until
3001                                       issuer_dname fingerprint);
3002        my $choice;
3003prompt:
3004        print STDERR $may_save ?
3005              "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
3006              "(R)eject or accept (t)emporarily? ";
3007        STDERR->flush;
3008        $choice = lc(substr(<STDIN> || 'R', 0, 1));
3009        if ($choice =~ /^t$/i) {
3010                $cred->may_save(undef);
3011        } elsif ($choice =~ /^r$/i) {
3012                return -1;
3013        } elsif ($may_save && $choice =~ /^p$/i) {
3014                $cred->may_save($may_save);
3015        } else {
3016                goto prompt;
3017        }
3018        $cred->accepted_failures($failures);
3019        $SVN::_Core::SVN_NO_ERROR;
3020}
3021
3022sub ssl_client_cert {
3023        my ($cred, $realm, $may_save, $pool) = @_;
3024        $may_save = undef if $_no_auth_cache;
3025        print STDERR "Client certificate filename: ";
3026        STDERR->flush;
3027        chomp(my $filename = <STDIN>);
3028        $cred->cert_file($filename);
3029        $cred->may_save($may_save);
3030        $SVN::_Core::SVN_NO_ERROR;
3031}
3032
3033sub ssl_client_cert_pw {
3034        my ($cred, $realm, $may_save, $pool) = @_;
3035        $may_save = undef if $_no_auth_cache;
3036        $cred->password(_read_password("Password: ", $realm));
3037        $cred->may_save($may_save);
3038        $SVN::_Core::SVN_NO_ERROR;
3039}
3040
3041sub username {
3042        my ($cred, $realm, $may_save, $pool) = @_;
3043        $may_save = undef if $_no_auth_cache;
3044        if (defined $realm && length $realm) {
3045                print STDERR "Authentication realm: $realm\n";
3046        }
3047        my $username;
3048        if (defined $_username) {
3049                $username = $_username;
3050        } else {
3051                print STDERR "Username: ";
3052                STDERR->flush;
3053                chomp($username = <STDIN>);
3054        }
3055        $cred->username($username);
3056        $cred->may_save($may_save);
3057        $SVN::_Core::SVN_NO_ERROR;
3058}
3059
3060sub _read_password {
3061        my ($prompt, $realm) = @_;
3062        print STDERR $prompt;
3063        STDERR->flush;
3064        require Term::ReadKey;
3065        Term::ReadKey::ReadMode('noecho');
3066        my $password = '';
3067        while (defined(my $key = Term::ReadKey::ReadKey(0))) {
3068                last if $key =~ /[\012\015]/; # \n\r
3069                $password .= $key;
3070        }
3071        Term::ReadKey::ReadMode('restore');
3072        print STDERR "\n";
3073        STDERR->flush;
3074        $password;
3075}
3076
3077package SVN::Git::Fetcher;
3078use vars qw/@ISA/;
3079use strict;
3080use warnings;
3081use Carp qw/croak/;
3082use File::Temp qw/tempfile/;
3083use IO::File qw//;
3084
3085# file baton members: path, mode_a, mode_b, pool, fh, blob, base
3086sub new {
3087        my ($class, $git_svn) = @_;
3088        my $self = SVN::Delta::Editor->new;
3089        bless $self, $class;
3090        $self->{c} = $git_svn->{last_commit} if exists $git_svn->{last_commit};
3091        $self->{empty} = {};
3092        $self->{dir_prop} = {};
3093        $self->{file_prop} = {};
3094        $self->{absent_dir} = {};
3095        $self->{absent_file} = {};
3096        $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
3097        $self;
3098}
3099
3100sub set_path_strip {
3101        my ($self, $path) = @_;
3102        $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
3103}
3104
3105sub open_root {
3106        { path => '' };
3107}
3108
3109sub open_directory {
3110        my ($self, $path, $pb, $rev) = @_;
3111        { path => $path };
3112}
3113
3114sub git_path {
3115        my ($self, $path) = @_;
3116        if ($self->{path_strip}) {
3117                $path =~ s!$self->{path_strip}!! or
3118                  die "Failed to strip path '$path' ($self->{path_strip})\n";
3119        }
3120        $path;
3121}
3122
3123sub delete_entry {
3124        my ($self, $path, $rev, $pb) = @_;
3125
3126        my $gpath = $self->git_path($path);
3127        return undef if ($gpath eq '');
3128
3129        # remove entire directories.
3130        if (command('ls-tree', $self->{c}, '--', $gpath) =~ /^040000 tree/) {
3131                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3132                                                     -r --name-only -z/,
3133                                                     $self->{c}, '--', $gpath);
3134                local $/ = "\0";
3135                while (<$ls>) {
3136                        chomp;
3137                        $self->{gii}->remove($_);
3138                        print "\tD\t$_\n" unless $::_q;
3139                }
3140                print "\tD\t$gpath/\n" unless $::_q;
3141                command_close_pipe($ls, $ctx);
3142                $self->{empty}->{$path} = 0
3143        } else {
3144                $self->{gii}->remove($gpath);
3145                print "\tD\t$gpath\n" unless $::_q;
3146        }
3147        undef;
3148}
3149
3150sub open_file {
3151        my ($self, $path, $pb, $rev) = @_;
3152        my $gpath = $self->git_path($path);
3153        my ($mode, $blob) = (command('ls-tree', $self->{c}, '--', $gpath)
3154                             =~ /^(\d{6}) blob ([a-f\d]{40})\t/);
3155        unless (defined $mode && defined $blob) {
3156                die "$path was not found in commit $self->{c} (r$rev)\n";
3157        }
3158        { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
3159          pool => SVN::Pool->new, action => 'M' };
3160}
3161
3162sub add_file {
3163        my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
3164        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3165        delete $self->{empty}->{$dir};
3166        { path => $path, mode_a => 100644, mode_b => 100644,
3167          pool => SVN::Pool->new, action => 'A' };
3168}
3169
3170sub add_directory {
3171        my ($self, $path, $cp_path, $cp_rev) = @_;
3172        my $gpath = $self->git_path($path);
3173        if ($gpath eq '') {
3174                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3175                                                     -r --name-only -z/,
3176                                                     $self->{c});
3177                local $/ = "\0";
3178                while (<$ls>) {
3179                        chomp;
3180                        $self->{gii}->remove($_);
3181                        print "\tD\t$_\n" unless $::_q;
3182                }
3183                command_close_pipe($ls, $ctx);
3184                $self->{empty}->{$path} = 0;
3185        }
3186        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3187        delete $self->{empty}->{$dir};
3188        $self->{empty}->{$path} = 1;
3189        { path => $path };
3190}
3191
3192sub change_dir_prop {
3193        my ($self, $db, $prop, $value) = @_;
3194        $self->{dir_prop}->{$db->{path}} ||= {};
3195        $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
3196        undef;
3197}
3198
3199sub absent_directory {
3200        my ($self, $path, $pb) = @_;
3201        $self->{absent_dir}->{$pb->{path}} ||= [];
3202        push @{$self->{absent_dir}->{$pb->{path}}}, $path;
3203        undef;
3204}
3205
3206sub absent_file {
3207        my ($self, $path, $pb) = @_;
3208        $self->{absent_file}->{$pb->{path}} ||= [];
3209        push @{$self->{absent_file}->{$pb->{path}}}, $path;
3210        undef;
3211}
3212
3213sub change_file_prop {
3214        my ($self, $fb, $prop, $value) = @_;
3215        if ($prop eq 'svn:executable') {
3216                if ($fb->{mode_b} != 120000) {
3217                        $fb->{mode_b} = defined $value ? 100755 : 100644;
3218                }
3219        } elsif ($prop eq 'svn:special') {
3220                $fb->{mode_b} = defined $value ? 120000 : 100644;
3221        } else {
3222                $self->{file_prop}->{$fb->{path}} ||= {};
3223                $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
3224        }
3225        undef;
3226}
3227
3228sub apply_textdelta {
3229        my ($self, $fb, $exp) = @_;
3230        my $fh = IO::File->new_tmpfile;
3231        $fh->autoflush(1);
3232        # $fh gets auto-closed() by SVN::TxDelta::apply(),
3233        # (but $base does not,) so dup() it for reading in close_file
3234        open my $dup, '<&', $fh or croak $!;
3235        my $base = IO::File->new_tmpfile;
3236        $base->autoflush(1);
3237        if ($fb->{blob}) {
3238                print $base 'link ' if ($fb->{mode_a} == 120000);
3239                my $size = $::_repository->cat_blob($fb->{blob}, $base);
3240                die "Failed to read object $fb->{blob}" if ($size < 0);
3241
3242                if (defined $exp) {
3243                        seek $base, 0, 0 or croak $!;
3244                        my $got = ::md5sum($base);
3245                        die "Checksum mismatch: $fb->{path} $fb->{blob}\n",
3246                            "expected: $exp\n",
3247                            "     got: $got\n" if ($got ne $exp);
3248                }
3249        }
3250        seek $base, 0, 0 or croak $!;
3251        $fb->{fh} = $dup;
3252        $fb->{base} = $base;
3253        [ SVN::TxDelta::apply($base, $fh, undef, $fb->{path}, $fb->{pool}) ];
3254}
3255
3256sub close_file {
3257        my ($self, $fb, $exp) = @_;
3258        my $hash;
3259        my $path = $self->git_path($fb->{path});
3260        if (my $fh = $fb->{fh}) {
3261                if (defined $exp) {
3262                        seek($fh, 0, 0) or croak $!;
3263                        my $got = ::md5sum($fh);
3264                        if ($got ne $exp) {
3265                                die "Checksum mismatch: $path\n",
3266                                    "expected: $exp\n    got: $got\n";
3267                        }
3268                }
3269                sysseek($fh, 0, 0) or croak $!;
3270                if ($fb->{mode_b} == 120000) {
3271                        eval {
3272                                sysread($fh, my $buf, 5) == 5 or croak $!;
3273                                $buf eq 'link ' or die "$path has mode 120000",
3274                                                       " but is not a link";
3275                        };
3276                        if ($@) {
3277                                warn "$@\n";
3278                                sysseek($fh, 0, 0) or croak $!;
3279                        }
3280                }
3281
3282                my ($tmp_fh, $tmp_filename) = File::Temp::tempfile(UNLINK => 1);
3283                my $result;
3284                while ($result = sysread($fh, my $string, 1024)) {
3285                        my $wrote = syswrite($tmp_fh, $string, $result);
3286                        defined($wrote) && $wrote == $result
3287                                or croak("write $tmp_filename: $!\n");
3288                }
3289                defined $result or croak $!;
3290                close $tmp_fh or croak $!;
3291
3292                close $fh or croak $!;
3293
3294                $hash = $::_repository->hash_and_insert_object($tmp_filename);
3295                unlink($tmp_filename);
3296                $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
3297                close $fb->{base} or croak $!;
3298        } else {
3299                $hash = $fb->{blob} or die "no blob information\n";
3300        }
3301        $fb->{pool}->clear;
3302        $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
3303        print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
3304        undef;
3305}
3306
3307sub abort_edit {
3308        my $self = shift;
3309        $self->{nr} = $self->{gii}->{nr};
3310        delete $self->{gii};
3311        $self->SUPER::abort_edit(@_);
3312}
3313
3314sub close_edit {
3315        my $self = shift;
3316        $self->{git_commit_ok} = 1;
3317        $self->{nr} = $self->{gii}->{nr};
3318        delete $self->{gii};
3319        $self->SUPER::close_edit(@_);
3320}
3321
3322package SVN::Git::Editor;
3323use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
3324use strict;
3325use warnings;
3326use Carp qw/croak/;
3327use IO::File;
3328
3329sub new {
3330        my ($class, $opts) = @_;
3331        foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
3332                die "$_ required!\n" unless (defined $opts->{$_});
3333        }
3334
3335        my $pool = SVN::Pool->new;
3336        my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
3337        my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
3338                                     $opts->{r}, $mods);
3339
3340        # $opts->{ra} functions should not be used after this:
3341        my @ce  = $opts->{ra}->get_commit_editor($opts->{log},
3342                                                $opts->{editor_cb}, $pool);
3343        my $self = SVN::Delta::Editor->new(@ce, $pool);
3344        bless $self, $class;
3345        foreach (qw/svn_path r tree_a tree_b/) {
3346                $self->{$_} = $opts->{$_};
3347        }
3348        $self->{url} = $opts->{ra}->{url};
3349        $self->{mods} = $mods;
3350        $self->{types} = $types;
3351        $self->{pool} = $pool;
3352        $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
3353        $self->{rm} = { };
3354        $self->{path_prefix} = length $self->{svn_path} ?
3355                               "$self->{svn_path}/" : '';
3356        $self->{config} = $opts->{config};
3357        return $self;
3358}
3359
3360sub generate_diff {
3361        my ($tree_a, $tree_b) = @_;
3362        my @diff_tree = qw(diff-tree -z -r);
3363        if ($_cp_similarity) {
3364                push @diff_tree, "-C$_cp_similarity";
3365        } else {
3366                push @diff_tree, '-C';
3367        }
3368        push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
3369        push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
3370        push @diff_tree, $tree_a, $tree_b;
3371        my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
3372        local $/ = "\0";
3373        my $state = 'meta';
3374        my @mods;
3375        while (<$diff_fh>) {
3376                chomp $_; # this gets rid of the trailing "\0"
3377                if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
3378                                        $::sha1\s($::sha1)\s
3379                                        ([MTCRAD])\d*$/xo) {
3380                        push @mods, {   mode_a => $1, mode_b => $2,
3381                                        sha1_b => $3, chg => $4 };
3382                        if ($4 =~ /^(?:C|R)$/) {
3383                                $state = 'file_a';
3384                        } else {
3385                                $state = 'file_b';
3386                        }
3387                } elsif ($state eq 'file_a') {
3388                        my $x = $mods[$#mods] or croak "Empty array\n";
3389                        if ($x->{chg} !~ /^(?:C|R)$/) {
3390                                croak "Error parsing $_, $x->{chg}\n";
3391                        }
3392                        $x->{file_a} = $_;
3393                        $state = 'file_b';
3394                } elsif ($state eq 'file_b') {
3395                        my $x = $mods[$#mods] or croak "Empty array\n";
3396                        if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
3397                                croak "Error parsing $_, $x->{chg}\n";
3398                        }
3399                        if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
3400                                croak "Error parsing $_, $x->{chg}\n";
3401                        }
3402                        $x->{file_b} = $_;
3403                        $state = 'meta';
3404                } else {
3405                        croak "Error parsing $_\n";
3406                }
3407        }
3408        command_close_pipe($diff_fh, $ctx);
3409        \@mods;
3410}
3411
3412sub check_diff_paths {
3413        my ($ra, $pfx, $rev, $mods) = @_;
3414        my %types;
3415        $pfx .= '/' if length $pfx;
3416
3417        sub type_diff_paths {
3418                my ($ra, $types, $path, $rev) = @_;
3419                my @p = split m#/+#, $path;
3420                my $c = shift @p;
3421                unless (defined $types->{$c}) {
3422                        $types->{$c} = $ra->check_path($c, $rev);
3423                }
3424                while (@p) {
3425                        $c .= '/' . shift @p;
3426                        next if defined $types->{$c};
3427                        $types->{$c} = $ra->check_path($c, $rev);
3428                }
3429        }
3430
3431        foreach my $m (@$mods) {
3432                foreach my $f (qw/file_a file_b/) {
3433                        next unless defined $m->{$f};
3434                        my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
3435                        if (length $pfx.$dir && ! defined $types{$dir}) {
3436                                type_diff_paths($ra, \%types, $pfx.$dir, $rev);
3437                        }
3438                }
3439        }
3440        \%types;
3441}
3442
3443sub split_path {
3444        return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
3445}
3446
3447sub repo_path {
3448        my ($self, $path) = @_;
3449        $self->{path_prefix}.(defined $path ? $path : '');
3450}
3451
3452sub url_path {
3453        my ($self, $path) = @_;
3454        if ($self->{url} =~ m#^https?://#) {
3455                $path =~ s/([^a-zA-Z0-9_.-])/uc sprintf("%%%02x",ord($1))/eg;
3456        }
3457        $self->{url} . '/' . $self->repo_path($path);
3458}
3459
3460sub rmdirs {
3461        my ($self) = @_;
3462        my $rm = $self->{rm};
3463        delete $rm->{''}; # we never delete the url we're tracking
3464        return unless %$rm;
3465
3466        foreach (keys %$rm) {
3467                my @d = split m#/#, $_;
3468                my $c = shift @d;
3469                $rm->{$c} = 1;
3470                while (@d) {
3471                        $c .= '/' . shift @d;
3472                        $rm->{$c} = 1;
3473                }
3474        }
3475        delete $rm->{$self->{svn_path}};
3476        delete $rm->{''}; # we never delete the url we're tracking
3477        return unless %$rm;
3478
3479        my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
3480                                             $self->{tree_b});
3481        local $/ = "\0";
3482        while (<$fh>) {
3483                chomp;
3484                my @dn = split m#/#, $_;
3485                while (pop @dn) {
3486                        delete $rm->{join '/', @dn};
3487                }
3488                unless (%$rm) {
3489                        close $fh;
3490                        return;
3491                }
3492        }
3493        command_close_pipe($fh, $ctx);
3494
3495        my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
3496        foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
3497                $self->close_directory($bat->{$d}, $p);
3498                my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
3499                print "\tD+\t$d/\n" unless $::_q;
3500                $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
3501                delete $bat->{$d};
3502        }
3503}
3504
3505sub open_or_add_dir {
3506        my ($self, $full_path, $baton) = @_;
3507        my $t = $self->{types}->{$full_path};
3508        if (!defined $t) {
3509                die "$full_path not known in r$self->{r} or we have a bug!\n";
3510        }
3511        {
3512                no warnings 'once';
3513                # SVN::Node::none and SVN::Node::file are used only once,
3514                # so we're shutting up Perl's warnings about them.
3515                if ($t == $SVN::Node::none) {
3516                        return $self->add_directory($full_path, $baton,
3517                            undef, -1, $self->{pool});
3518                } elsif ($t == $SVN::Node::dir) {
3519                        return $self->open_directory($full_path, $baton,
3520                            $self->{r}, $self->{pool});
3521                } # no warnings 'once'
3522                print STDERR "$full_path already exists in repository at ",
3523                    "r$self->{r} and it is not a directory (",
3524                    ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
3525        } # no warnings 'once'
3526        exit 1;
3527}
3528
3529sub ensure_path {
3530        my ($self, $path) = @_;
3531        my $bat = $self->{bat};
3532        my $repo_path = $self->repo_path($path);
3533        return $bat->{''} unless (length $repo_path);
3534        my @p = split m#/+#, $repo_path;
3535        my $c = shift @p;
3536        $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
3537        while (@p) {
3538                my $c0 = $c;
3539                $c .= '/' . shift @p;
3540                $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
3541        }
3542        return $bat->{$c};
3543}
3544
3545# Subroutine to convert a globbing pattern to a regular expression.
3546# From perl cookbook.
3547sub glob2pat {
3548        my $globstr = shift;
3549        my %patmap = ('*' => '.*', '?' => '.', '[' => '[', ']' => ']');
3550        $globstr =~ s{(.)} { $patmap{$1} || "\Q$1" }ge;
3551        return '^' . $globstr . '$';
3552}
3553
3554sub check_autoprop {
3555        my ($self, $pattern, $properties, $file, $fbat) = @_;
3556        # Convert the globbing pattern to a regular expression.
3557        my $regex = glob2pat($pattern);
3558        # Check if the pattern matches the file name.
3559        if($file =~ m/($regex)/) {
3560                # Parse the list of properties to set.
3561                my @props = split(/;/, $properties);
3562                foreach my $prop (@props) {
3563                        # Parse 'name=value' syntax and set the property.
3564                        if ($prop =~ /([^=]+)=(.*)/) {
3565                                my ($n,$v) = ($1,$2);
3566                                for ($n, $v) {
3567                                        s/^\s+//; s/\s+$//;
3568                                }
3569                                $self->change_file_prop($fbat, $n, $v);
3570                        }
3571                }
3572        }
3573}
3574
3575sub apply_autoprops {
3576        my ($self, $file, $fbat) = @_;
3577        my $conf_t = ${$self->{config}}{'config'};
3578        no warnings 'once';
3579        # Check [miscellany]/enable-auto-props in svn configuration.
3580        if (SVN::_Core::svn_config_get_bool(
3581                $conf_t,
3582                $SVN::_Core::SVN_CONFIG_SECTION_MISCELLANY,
3583                $SVN::_Core::SVN_CONFIG_OPTION_ENABLE_AUTO_PROPS,
3584                0)) {
3585                # Auto-props are enabled.  Enumerate them to look for matches.
3586                my $callback = sub {
3587                        $self->check_autoprop($_[0], $_[1], $file, $fbat);
3588                };
3589                SVN::_Core::svn_config_enumerate(
3590                        $conf_t,
3591                        $SVN::_Core::SVN_CONFIG_SECTION_AUTO_PROPS,
3592                        $callback);
3593        }
3594}
3595
3596sub A {
3597        my ($self, $m) = @_;
3598        my ($dir, $file) = split_path($m->{file_b});
3599        my $pbat = $self->ensure_path($dir);
3600        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3601                                        undef, -1);
3602        print "\tA\t$m->{file_b}\n" unless $::_q;
3603        $self->apply_autoprops($file, $fbat);
3604        $self->chg_file($fbat, $m);
3605        $self->close_file($fbat,undef,$self->{pool});
3606}
3607
3608sub C {
3609        my ($self, $m) = @_;
3610        my ($dir, $file) = split_path($m->{file_b});
3611        my $pbat = $self->ensure_path($dir);
3612        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3613                                $self->url_path($m->{file_a}), $self->{r});
3614        print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
3615        $self->chg_file($fbat, $m);
3616        $self->close_file($fbat,undef,$self->{pool});
3617}
3618
3619sub delete_entry {
3620        my ($self, $path, $pbat) = @_;
3621        my $rpath = $self->repo_path($path);
3622        my ($dir, $file) = split_path($rpath);
3623        $self->{rm}->{$dir} = 1;
3624        $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
3625}
3626
3627sub R {
3628        my ($self, $m) = @_;
3629        my ($dir, $file) = split_path($m->{file_b});
3630        my $pbat = $self->ensure_path($dir);
3631        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3632                                $self->url_path($m->{file_a}), $self->{r});
3633        print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
3634        $self->chg_file($fbat, $m);
3635        $self->close_file($fbat,undef,$self->{pool});
3636
3637        ($dir, $file) = split_path($m->{file_a});
3638        $pbat = $self->ensure_path($dir);
3639        $self->delete_entry($m->{file_a}, $pbat);
3640}
3641
3642sub M {
3643        my ($self, $m) = @_;
3644        my ($dir, $file) = split_path($m->{file_b});
3645        my $pbat = $self->ensure_path($dir);
3646        my $fbat = $self->open_file($self->repo_path($m->{file_b}),
3647                                $pbat,$self->{r},$self->{pool});
3648        print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
3649        $self->chg_file($fbat, $m);
3650        $self->close_file($fbat,undef,$self->{pool});
3651}
3652
3653sub T { shift->M(@_) }
3654
3655sub change_file_prop {
3656        my ($self, $fbat, $pname, $pval) = @_;
3657        $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
3658}
3659
3660sub chg_file {
3661        my ($self, $fbat, $m) = @_;
3662        if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
3663                $self->change_file_prop($fbat,'svn:executable','*');
3664        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
3665                $self->change_file_prop($fbat,'svn:executable',undef);
3666        }
3667        my $fh = IO::File->new_tmpfile or croak $!;
3668        if ($m->{mode_b} =~ /^120/) {
3669                print $fh 'link ' or croak $!;
3670                $self->change_file_prop($fbat,'svn:special','*');
3671        } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
3672                $self->change_file_prop($fbat,'svn:special',undef);
3673        }
3674        my $size = $::_repository->cat_blob($m->{sha1_b}, $fh);
3675        croak "Failed to read object $m->{sha1_b}" if ($size < 0);
3676        $fh->flush == 0 or croak $!;
3677        seek $fh, 0, 0 or croak $!;
3678
3679        my $exp = ::md5sum($fh);
3680        seek $fh, 0, 0 or croak $!;
3681
3682        my $pool = SVN::Pool->new;
3683        my $atd = $self->apply_textdelta($fbat, undef, $pool);
3684        my $got = SVN::TxDelta::send_stream($fh, @$atd, $pool);
3685        die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
3686        $pool->clear;
3687
3688        close $fh or croak $!;
3689}
3690
3691sub D {
3692        my ($self, $m) = @_;
3693        my ($dir, $file) = split_path($m->{file_b});
3694        my $pbat = $self->ensure_path($dir);
3695        print "\tD\t$m->{file_b}\n" unless $::_q;
3696        $self->delete_entry($m->{file_b}, $pbat);
3697}
3698
3699sub close_edit {
3700        my ($self) = @_;
3701        my ($p,$bat) = ($self->{pool}, $self->{bat});
3702        foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
3703                next if $_ eq '';
3704                $self->close_directory($bat->{$_}, $p);
3705        }
3706        $self->close_directory($bat->{''}, $p);
3707        $self->SUPER::close_edit($p);
3708        $p->clear;
3709}
3710
3711sub abort_edit {
3712        my ($self) = @_;
3713        $self->SUPER::abort_edit($self->{pool});
3714}
3715
3716sub DESTROY {
3717        my $self = shift;
3718        $self->SUPER::DESTROY(@_);
3719        $self->{pool}->clear;
3720}
3721
3722# this drives the editor
3723sub apply_diff {
3724        my ($self) = @_;
3725        my $mods = $self->{mods};
3726        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
3727        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
3728                my $f = $m->{chg};
3729                if (defined $o{$f}) {
3730                        $self->$f($m);
3731                } else {
3732                        fatal("Invalid change type: $f");
3733                }
3734        }
3735        $self->rmdirs if $_rmdir;
3736        if (@$mods == 0) {
3737                $self->abort_edit;
3738        } else {
3739                $self->close_edit;
3740        }
3741        return scalar @$mods;
3742}
3743
3744package Git::SVN::Ra;
3745use vars qw/@ISA $config_dir $_log_window_size/;
3746use strict;
3747use warnings;
3748my ($ra_invalid, $can_do_switch, %ignored_err, $RA);
3749
3750BEGIN {
3751        # enforce temporary pool usage for some simple functions
3752        no strict 'refs';
3753        for my $f (qw/rev_proplist get_latest_revnum get_uuid get_repos_root/) {
3754                my $SUPER = "SUPER::$f";
3755                *$f = sub {
3756                        my $self = shift;
3757                        my $pool = SVN::Pool->new;
3758                        my @ret = $self->$SUPER(@_,$pool);
3759                        $pool->clear;
3760                        wantarray ? @ret : $ret[0];
3761                };
3762        }
3763}
3764
3765sub _auth_providers () {
3766        [
3767          SVN::Client::get_simple_provider(),
3768          SVN::Client::get_ssl_server_trust_file_provider(),
3769          SVN::Client::get_simple_prompt_provider(
3770            \&Git::SVN::Prompt::simple, 2),
3771          SVN::Client::get_ssl_client_cert_file_provider(),
3772          SVN::Client::get_ssl_client_cert_prompt_provider(
3773            \&Git::SVN::Prompt::ssl_client_cert, 2),
3774          SVN::Client::get_ssl_client_cert_pw_file_provider(),
3775          SVN::Client::get_ssl_client_cert_pw_prompt_provider(
3776            \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
3777          SVN::Client::get_username_provider(),
3778          SVN::Client::get_ssl_server_trust_prompt_provider(
3779            \&Git::SVN::Prompt::ssl_server_trust),
3780          SVN::Client::get_username_prompt_provider(
3781            \&Git::SVN::Prompt::username, 2)
3782        ]
3783}
3784
3785sub escape_uri_only {
3786        my ($uri) = @_;
3787        my @tmp;
3788        foreach (split m{/}, $uri) {
3789                s/([^\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
3790                push @tmp, $_;
3791        }
3792        join('/', @tmp);
3793}
3794
3795sub escape_url {
3796        my ($url) = @_;
3797        if ($url =~ m#^(https?)://([^/]+)(.*)$#) {
3798                my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
3799                $url = "$scheme://$domain$uri";
3800        }
3801        $url;
3802}
3803
3804sub new {
3805        my ($class, $url) = @_;
3806        $url =~ s!/+$!!;
3807        return $RA if ($RA && $RA->{url} eq $url);
3808
3809        SVN::_Core::svn_config_ensure($config_dir, undef);
3810        my ($baton, $callbacks) = SVN::Core::auth_open_helper(_auth_providers);
3811        my $config = SVN::Core::config_get_config($config_dir);
3812        $RA = undef;
3813        my $dont_store_passwords = 1;
3814        my $conf_t = ${$config}{'config'};
3815        {
3816                no warnings 'once';
3817                # The usage of $SVN::_Core::SVN_CONFIG_* variables
3818                # produces warnings that variables are used only once.
3819                # I had not found the better way to shut them up, so
3820                # the warnings of type 'once' are disabled in this block.
3821                if (SVN::_Core::svn_config_get_bool($conf_t,
3822                    $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
3823                    $SVN::_Core::SVN_CONFIG_OPTION_STORE_PASSWORDS,
3824                    1) == 0) {
3825                        SVN::_Core::svn_auth_set_parameter($baton,
3826                            $SVN::_Core::SVN_AUTH_PARAM_DONT_STORE_PASSWORDS,
3827                            bless (\$dont_store_passwords, "_p_void"));
3828                }
3829                if (SVN::_Core::svn_config_get_bool($conf_t,
3830                    $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
3831                    $SVN::_Core::SVN_CONFIG_OPTION_STORE_AUTH_CREDS,
3832                    1) == 0) {
3833                        $Git::SVN::Prompt::_no_auth_cache = 1;
3834                }
3835        } # no warnings 'once'
3836        my $self = SVN::Ra->new(url => escape_url($url), auth => $baton,
3837                              config => $config,
3838                              pool => SVN::Pool->new,
3839                              auth_provider_callbacks => $callbacks);
3840        $self->{url} = $url;
3841        $self->{svn_path} = $url;
3842        $self->{repos_root} = $self->get_repos_root;
3843        $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
3844        $self->{cache} = { check_path => { r => 0, data => {} },
3845                           get_dir => { r => 0, data => {} } };
3846        $RA = bless $self, $class;
3847}
3848
3849sub check_path {
3850        my ($self, $path, $r) = @_;
3851        my $cache = $self->{cache}->{check_path};
3852        if ($r == $cache->{r} && exists $cache->{data}->{$path}) {
3853                return $cache->{data}->{$path};
3854        }
3855        my $pool = SVN::Pool->new;
3856        my $t = $self->SUPER::check_path($path, $r, $pool);
3857        $pool->clear;
3858        if ($r != $cache->{r}) {
3859                %{$cache->{data}} = ();
3860                $cache->{r} = $r;
3861        }
3862        $cache->{data}->{$path} = $t;
3863}
3864
3865sub get_dir {
3866        my ($self, $dir, $r) = @_;
3867        my $cache = $self->{cache}->{get_dir};
3868        if ($r == $cache->{r}) {
3869                if (my $x = $cache->{data}->{$dir}) {
3870                        return wantarray ? @$x : $x->[0];
3871                }
3872        }
3873        my $pool = SVN::Pool->new;
3874        my ($d, undef, $props) = $self->SUPER::get_dir($dir, $r, $pool);
3875        my %dirents = map { $_ => { kind => $d->{$_}->kind } } keys %$d;
3876        $pool->clear;
3877        if ($r != $cache->{r}) {
3878                %{$cache->{data}} = ();
3879                $cache->{r} = $r;
3880        }
3881        $cache->{data}->{$dir} = [ \%dirents, $r, $props ];
3882        wantarray ? (\%dirents, $r, $props) : \%dirents;
3883}
3884
3885sub DESTROY {
3886        # do not call the real DESTROY since we store ourselves in $RA
3887}
3888
3889sub get_log {
3890        my ($self, @args) = @_;
3891        my $pool = SVN::Pool->new;
3892        splice(@args, 3, 1) if ($SVN::Core::VERSION le '1.2.0');
3893        my $ret = $self->SUPER::get_log(@args, $pool);
3894        $pool->clear;
3895        $ret;
3896}
3897
3898sub trees_match {
3899        my ($self, $url1, $rev1, $url2, $rev2) = @_;
3900        my $ctx = SVN::Client->new(auth => _auth_providers);
3901        my $out = IO::File->new_tmpfile;
3902
3903        # older SVN (1.1.x) doesn't take $pool as the last parameter for
3904        # $ctx->diff(), so we'll create a default one
3905        my $pool = SVN::Pool->new_default_sub;
3906
3907        $ra_invalid = 1; # this will open a new SVN::Ra connection to $url1
3908        $ctx->diff([], $url1, $rev1, $url2, $rev2, 1, 1, 0, $out, $out);
3909        $out->flush;
3910        my $ret = (($out->stat)[7] == 0);
3911        close $out or croak $!;
3912
3913        $ret;
3914}
3915
3916sub get_commit_editor {
3917        my ($self, $log, $cb, $pool) = @_;
3918        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
3919        $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
3920}
3921
3922sub gs_do_update {
3923        my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
3924        my $new = ($rev_a == $rev_b);
3925        my $path = $gs->{path};
3926
3927        if ($new && -e $gs->{index}) {
3928                unlink $gs->{index} or die
3929                  "Couldn't unlink index: $gs->{index}: $!\n";
3930        }
3931        my $pool = SVN::Pool->new;
3932        $editor->set_path_strip($path);
3933        my (@pc) = split m#/#, $path;
3934        my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
3935                                        1, $editor, $pool);
3936        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
3937
3938        # Since we can't rely on svn_ra_reparent being available, we'll
3939        # just have to do some magic with set_path to make it so
3940        # we only want a partial path.
3941        my $sp = '';
3942        my $final = join('/', @pc);
3943        while (@pc) {
3944                $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
3945                $sp .= '/' if length $sp;
3946                $sp .= shift @pc;
3947        }
3948        die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
3949
3950        $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
3951
3952        $reporter->finish_report($pool);
3953        $pool->clear;
3954        $editor->{git_commit_ok};
3955}
3956
3957# this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
3958# svn_ra_reparent didn't work before 1.4)
3959sub gs_do_switch {
3960        my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
3961        my $path = $gs->{path};
3962        my $pool = SVN::Pool->new;
3963
3964        my $full_url = $self->{url};
3965        my $old_url = $full_url;
3966        $full_url .= '/' . escape_uri_only($path) if length $path;
3967        my ($ra, $reparented);
3968        if ($old_url ne $full_url) {
3969                if ($old_url !~ m#^svn(\+ssh)?://#) {
3970                        SVN::_Ra::svn_ra_reparent($self->{session}, $full_url,
3971                                                  $pool);
3972                        $self->{url} = $full_url;
3973                        $reparented = 1;
3974                } else {
3975                        $_[0] = undef;
3976                        $self = undef;
3977                        $RA = undef;
3978                        $ra = Git::SVN::Ra->new($full_url);
3979                        $ra_invalid = 1;
3980                }
3981        }
3982        $ra ||= $self;
3983        my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
3984        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
3985        $reporter->set_path('', $rev_a, 0, @lock, $pool);
3986        $reporter->finish_report($pool);
3987
3988        if ($reparented) {
3989                SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
3990                $self->{url} = $old_url;
3991        }
3992
3993        $pool->clear;
3994        $editor->{git_commit_ok};
3995}
3996
3997sub longest_common_path {
3998        my ($gsv, $globs) = @_;
3999        my %common;
4000        my $common_max = scalar @$gsv;
4001
4002        foreach my $gs (@$gsv) {
4003                my @tmp = split m#/#, $gs->{path};
4004                my $p = '';
4005                foreach (@tmp) {
4006                        $p .= length($p) ? "/$_" : $_;
4007                        $common{$p} ||= 0;
4008                        $common{$p}++;
4009                }
4010        }
4011        $globs ||= [];
4012        $common_max += scalar @$globs;
4013        foreach my $glob (@$globs) {
4014                my @tmp = split m#/#, $glob->{path}->{left};
4015                my $p = '';
4016                foreach (@tmp) {
4017                        $p .= length($p) ? "/$_" : $_;
4018                        $common{$p} ||= 0;
4019                        $common{$p}++;
4020                }
4021        }
4022
4023        my $longest_path = '';
4024        foreach (sort {length $b <=> length $a} keys %common) {
4025                if ($common{$_} == $common_max) {
4026                        $longest_path = $_;
4027                        last;
4028                }
4029        }
4030        $longest_path;
4031}
4032
4033sub gs_fetch_loop_common {
4034        my ($self, $base, $head, $gsv, $globs) = @_;
4035        return if ($base > $head);
4036        my $inc = $_log_window_size;
4037        my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
4038        my $longest_path = longest_common_path($gsv, $globs);
4039        my $ra_url = $self->{url};
4040        while (1) {
4041                my %revs;
4042                my $err;
4043                my $err_handler = $SVN::Error::handler;
4044                $SVN::Error::handler = sub {
4045                        ($err) = @_;
4046                        skip_unknown_revs($err);
4047                };
4048                sub _cb {
4049                        my ($paths, $r, $author, $date, $log) = @_;
4050                        [ dup_changed_paths($paths),
4051                          { author => $author, date => $date, log => $log } ];
4052                }
4053                $self->get_log([$longest_path], $min, $max, 0, 1, 1,
4054                               sub { $revs{$_[1]} = _cb(@_) });
4055                if ($err && $max >= $head) {
4056                        print STDERR "Path '$longest_path' ",
4057                                     "was probably deleted:\n",
4058                                     $err->expanded_message,
4059                                     "\nWill attempt to follow ",
4060                                     "revisions r$min .. r$max ",
4061                                     "committed before the deletion\n";
4062                        my $hi = $max;
4063                        while (--$hi >= $min) {
4064                                my $ok;
4065                                $self->get_log([$longest_path], $min, $hi,
4066                                               0, 1, 1, sub {
4067                                               $ok ||= $_[1];
4068                                               $revs{$_[1]} = _cb(@_) });
4069                                if ($ok) {
4070                                        print STDERR "r$min .. r$ok OK\n";
4071                                        last;
4072                                }
4073                        }
4074                }
4075                $SVN::Error::handler = $err_handler;
4076
4077                my %exists = map { $_->{path} => $_ } @$gsv;
4078                foreach my $r (sort {$a <=> $b} keys %revs) {
4079                        my ($paths, $logged) = @{$revs{$r}};
4080
4081                        foreach my $gs ($self->match_globs(\%exists, $paths,
4082                                                           $globs, $r)) {
4083                                if ($gs->rev_map_max >= $r) {
4084                                        next;
4085                                }
4086                                next unless $gs->match_paths($paths, $r);
4087                                $gs->{logged_rev_props} = $logged;
4088                                if (my $last_commit = $gs->last_commit) {
4089                                        $gs->assert_index_clean($last_commit);
4090                                }
4091                                my $log_entry = $gs->do_fetch($paths, $r);
4092                                if ($log_entry) {
4093                                        $gs->do_git_commit($log_entry);
4094                                }
4095                                $INDEX_FILES{$gs->{index}} = 1;
4096                        }
4097                        foreach my $g (@$globs) {
4098                                my $k = "svn-remote.$g->{remote}." .
4099                                        "$g->{t}-maxRev";
4100                                Git::SVN::tmp_config($k, $r);
4101                        }
4102                        if ($ra_invalid) {
4103                                $_[0] = undef;
4104                                $self = undef;
4105                                $RA = undef;
4106                                $self = Git::SVN::Ra->new($ra_url);
4107                                $ra_invalid = undef;
4108                        }
4109                }
4110                # pre-fill the .rev_db since it'll eventually get filled in
4111                # with '0' x40 if something new gets committed
4112                foreach my $gs (@$gsv) {
4113                        next if $gs->rev_map_max >= $max;
4114                        next if defined $gs->rev_map_get($max);
4115                        $gs->rev_map_set($max, 0 x40);
4116                }
4117                foreach my $g (@$globs) {
4118                        my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
4119                        Git::SVN::tmp_config($k, $max);
4120                }
4121                last if $max >= $head;
4122                $min = $max + 1;
4123                $max += $inc;
4124                $max = $head if ($max > $head);
4125        }
4126        Git::SVN::gc();
4127}
4128
4129sub get_dir_globbed {
4130        my ($self, $left, $depth, $r) = @_;
4131
4132        my @x = eval { $self->get_dir($left, $r) };
4133        return unless scalar @x == 3;
4134        my $dirents = $x[0];
4135        my @finalents;
4136        foreach my $de (keys %$dirents) {
4137                next if $dirents->{$de}->{kind} != $SVN::Node::dir;
4138                if ($depth > 1) {
4139                        my @args = ("$left/$de", $depth - 1, $r);
4140                        foreach my $dir ($self->get_dir_globbed(@args)) {
4141                                push @finalents, "$de/$dir";
4142                        }
4143                } else {
4144                        push @finalents, $de;
4145                }
4146        }
4147        @finalents;
4148}
4149
4150sub match_globs {
4151        my ($self, $exists, $paths, $globs, $r) = @_;
4152
4153        sub get_dir_check {
4154                my ($self, $exists, $g, $r) = @_;
4155
4156                my @dirs = $self->get_dir_globbed($g->{path}->{left},
4157                                                  $g->{path}->{depth},
4158                                                  $r);
4159
4160                foreach my $de (@dirs) {
4161                        my $p = $g->{path}->full_path($de);
4162                        next if $exists->{$p};
4163                        next if (length $g->{path}->{right} &&
4164                                 ($self->check_path($p, $r) !=
4165                                  $SVN::Node::dir));
4166                        $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
4167                                         $g->{ref}->full_path($de), 1);
4168                }
4169        }
4170        foreach my $g (@$globs) {
4171                if (my $path = $paths->{"/$g->{path}->{left}"}) {
4172                        if ($path->{action} =~ /^[AR]$/) {
4173                                get_dir_check($self, $exists, $g, $r);
4174                        }
4175                }
4176                foreach (keys %$paths) {
4177                        if (/$g->{path}->{left_regex}/ &&
4178                            !/$g->{path}->{regex}/) {
4179                                next if $paths->{$_}->{action} !~ /^[AR]$/;
4180                                get_dir_check($self, $exists, $g, $r);
4181                        }
4182                        next unless /$g->{path}->{regex}/;
4183                        my $p = $1;
4184                        my $pathname = $g->{path}->full_path($p);
4185                        next if $exists->{$pathname};
4186                        next if ($self->check_path($pathname, $r) !=
4187                                 $SVN::Node::dir);
4188                        $exists->{$pathname} = Git::SVN->init(
4189                                              $self->{url}, $pathname, undef,
4190                                              $g->{ref}->full_path($p), 1);
4191                }
4192                my $c = '';
4193                foreach (split m#/#, $g->{path}->{left}) {
4194                        $c .= "/$_";
4195                        next unless ($paths->{$c} &&
4196                                     ($paths->{$c}->{action} =~ /^[AR]$/));
4197                        get_dir_check($self, $exists, $g, $r);
4198                }
4199        }
4200        values %$exists;
4201}
4202
4203sub minimize_url {
4204        my ($self) = @_;
4205        return $self->{url} if ($self->{url} eq $self->{repos_root});
4206        my $url = $self->{repos_root};
4207        my @components = split(m!/!, $self->{svn_path});
4208        my $c = '';
4209        do {
4210                $url .= "/$c" if length $c;
4211                eval { (ref $self)->new($url)->get_latest_revnum };
4212        } while ($@ && ($c = shift @components));
4213        $url;
4214}
4215
4216sub can_do_switch {
4217        my $self = shift;
4218        unless (defined $can_do_switch) {
4219                my $pool = SVN::Pool->new;
4220                my $rep = eval {
4221                        $self->do_switch(1, '', 0, $self->{url},
4222                                         SVN::Delta::Editor->new, $pool);
4223                };
4224                if ($@) {
4225                        $can_do_switch = 0;
4226                } else {
4227                        $rep->abort_report($pool);
4228                        $can_do_switch = 1;
4229                }
4230                $pool->clear;
4231        }
4232        $can_do_switch;
4233}
4234
4235sub skip_unknown_revs {
4236        my ($err) = @_;
4237        my $errno = $err->apr_err();
4238        # Maybe the branch we're tracking didn't
4239        # exist when the repo started, so it's
4240        # not an error if it doesn't, just continue
4241        #
4242        # Wonderfully consistent library, eh?
4243        # 160013 - svn:// and file://
4244        # 175002 - http(s)://
4245        # 175007 - http(s):// (this repo required authorization, too...)
4246        #   More codes may be discovered later...
4247        if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
4248                my $err_key = $err->expanded_message;
4249                # revision numbers change every time, filter them out
4250                $err_key =~ s/\d+/\0/g;
4251                $err_key = "$errno\0$err_key";
4252                unless ($ignored_err{$err_key}) {
4253                        warn "W: Ignoring error from SVN, path probably ",
4254                             "does not exist: ($errno): ",
4255                             $err->expanded_message,"\n";
4256                        warn "W: Do not be alarmed at the above message ",
4257                             "git-svn is just searching aggressively for ",
4258                             "old history.\n",
4259                             "This may take a while on large repositories\n";
4260                        $ignored_err{$err_key} = 1;
4261                }
4262                return;
4263        }
4264        die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
4265}
4266
4267# svn_log_changed_path_t objects passed to get_log are likely to be
4268# overwritten even if only the refs are copied to an external variable,
4269# so we should dup the structures in their entirety.  Using an externally
4270# passed pool (instead of our temporary and quickly cleared pool in
4271# Git::SVN::Ra) does not help matters at all...
4272sub dup_changed_paths {
4273        my ($paths) = @_;
4274        return undef unless $paths;
4275        my %ret;
4276        foreach my $p (keys %$paths) {
4277                my $i = $paths->{$p};
4278                my %s = map { $_ => $i->$_ }
4279                              qw/copyfrom_path copyfrom_rev action/;
4280                $ret{$p} = \%s;
4281        }
4282        \%ret;
4283}
4284
4285package Git::SVN::Log;
4286use strict;
4287use warnings;
4288use POSIX qw/strftime/;
4289use constant commit_log_separator => ('-' x 72) . "\n";
4290use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
4291            %rusers $show_commit $incremental/;
4292my $l_fmt;
4293
4294sub cmt_showable {
4295        my ($c) = @_;
4296        return 1 if defined $c->{r};
4297
4298        # big commit message got truncated by the 16k pretty buffer in rev-list
4299        if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
4300                                $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
4301                @{$c->{l}} = ();
4302                my @log = command(qw/cat-file commit/, $c->{c});
4303
4304                # shift off the headers
4305                shift @log while ($log[0] ne '');
4306                shift @log;
4307
4308                # TODO: make $c->{l} not have a trailing newline in the future
4309                @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
4310
4311                (undef, $c->{r}, undef) = ::extract_metadata(
4312                                (grep(/^git-svn-id: /, @log))[-1]);
4313        }
4314        return defined $c->{r};
4315}
4316
4317sub log_use_color {
4318        return $color || Git->repository->get_colorbool('color.diff');
4319}
4320
4321sub git_svn_log_cmd {
4322        my ($r_min, $r_max, @args) = @_;
4323        my $head = 'HEAD';
4324        my (@files, @log_opts);
4325        foreach my $x (@args) {
4326                if ($x eq '--' || @files) {
4327                        push @files, $x;
4328                } else {
4329                        if (::verify_ref("$x^0")) {
4330                                $head = $x;
4331                        } else {
4332                                push @log_opts, $x;
4333                        }
4334                }
4335        }
4336
4337        my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
4338        $gs ||= Git::SVN->_new;
4339        my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
4340                   $gs->refname);
4341        push @cmd, '-r' unless $non_recursive;
4342        push @cmd, qw/--raw --name-status/ if $verbose;
4343        push @cmd, '--color' if log_use_color();
4344        push @cmd, @log_opts;
4345        if (defined $r_max && $r_max == $r_min) {
4346                push @cmd, '--max-count=1';
4347                if (my $c = $gs->rev_map_get($r_max)) {
4348                        push @cmd, $c;
4349                }
4350        } elsif (defined $r_max) {
4351                if ($r_max < $r_min) {
4352                        ($r_min, $r_max) = ($r_max, $r_min);
4353                }
4354                my (undef, $c_max) = $gs->find_rev_before($r_max, 1, $r_min);
4355                my (undef, $c_min) = $gs->find_rev_after($r_min, 1, $r_max);
4356                # If there are no commits in the range, both $c_max and $c_min
4357                # will be undefined.  If there is at least 1 commit in the
4358                # range, both will be defined.
4359                return () if !defined $c_min || !defined $c_max;
4360                if ($c_min eq $c_max) {
4361                        push @cmd, '--max-count=1', $c_min;
4362                } else {
4363                        push @cmd, '--boundary', "$c_min..$c_max";
4364                }
4365        }
4366        return (@cmd, @files);
4367}
4368
4369# adapted from pager.c
4370sub config_pager {
4371        $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
4372        if (!defined $pager) {
4373                $pager = 'less';
4374        } elsif (length $pager == 0 || $pager eq 'cat') {
4375                $pager = undef;
4376        }
4377        $ENV{GIT_PAGER_IN_USE} = defined($pager);
4378}
4379
4380sub run_pager {
4381        return unless -t *STDOUT && defined $pager;
4382        pipe my $rfd, my $wfd or return;
4383        defined(my $pid = fork) or ::fatal "Can't fork: $!";
4384        if (!$pid) {
4385                open STDOUT, '>&', $wfd or
4386                                     ::fatal "Can't redirect to stdout: $!";
4387                return;
4388        }
4389        open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!";
4390        $ENV{LESS} ||= 'FRSX';
4391        exec $pager or ::fatal "Can't run pager: $! ($pager)";
4392}
4393
4394sub format_svn_date {
4395        return strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)", localtime(shift));
4396}
4397
4398sub parse_git_date {
4399        my ($t, $tz) = @_;
4400        # Date::Parse isn't in the standard Perl distro :(
4401        if ($tz =~ s/^\+//) {
4402                $t += tz_to_s_offset($tz);
4403        } elsif ($tz =~ s/^\-//) {
4404                $t -= tz_to_s_offset($tz);
4405        }
4406        return $t;
4407}
4408
4409sub set_local_timezone {
4410        if (defined $TZ) {
4411                $ENV{TZ} = $TZ;
4412        } else {
4413                delete $ENV{TZ};
4414        }
4415}
4416
4417sub tz_to_s_offset {
4418        my ($tz) = @_;
4419        $tz =~ s/(\d\d)$//;
4420        return ($1 * 60) + ($tz * 3600);
4421}
4422
4423sub get_author_info {
4424        my ($dest, $author, $t, $tz) = @_;
4425        $author =~ s/(?:^\s*|\s*$)//g;
4426        $dest->{a_raw} = $author;
4427        my $au;
4428        if ($::_authors) {
4429                $au = $rusers{$author} || undef;
4430        }
4431        if (!$au) {
4432                ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
4433        }
4434        $dest->{t} = $t;
4435        $dest->{tz} = $tz;
4436        $dest->{a} = $au;
4437        $dest->{t_utc} = parse_git_date($t, $tz);
4438}
4439
4440sub process_commit {
4441        my ($c, $r_min, $r_max, $defer) = @_;
4442        if (defined $r_min && defined $r_max) {
4443                if ($r_min == $c->{r} && $r_min == $r_max) {
4444                        show_commit($c);
4445                        return 0;
4446                }
4447                return 1 if $r_min == $r_max;
4448                if ($r_min < $r_max) {
4449                        # we need to reverse the print order
4450                        return 0 if (defined $limit && --$limit < 0);
4451                        push @$defer, $c;
4452                        return 1;
4453                }
4454                if ($r_min != $r_max) {
4455                        return 1 if ($r_min < $c->{r});
4456                        return 1 if ($r_max > $c->{r});
4457                }
4458        }
4459        return 0 if (defined $limit && --$limit < 0);
4460        show_commit($c);
4461        return 1;
4462}
4463
4464sub show_commit {
4465        my $c = shift;
4466        if ($oneline) {
4467                my $x = "\n";
4468                if (my $l = $c->{l}) {
4469                        while ($l->[0] =~ /^\s*$/) { shift @$l }
4470                        $x = $l->[0];
4471                }
4472                $l_fmt ||= 'A' . length($c->{r});
4473                print 'r',pack($l_fmt, $c->{r}),' | ';
4474                print "$c->{c} | " if $show_commit;
4475                print $x;
4476        } else {
4477                show_commit_normal($c);
4478        }
4479}
4480
4481sub show_commit_changed_paths {
4482        my ($c) = @_;
4483        return unless $c->{changed};
4484        print "Changed paths:\n", @{$c->{changed}};
4485}
4486
4487sub show_commit_normal {
4488        my ($c) = @_;
4489        print commit_log_separator, "r$c->{r} | ";
4490        print "$c->{c} | " if $show_commit;
4491        print "$c->{a} | ", format_svn_date($c->{t_utc}), ' | ';
4492        my $nr_line = 0;
4493
4494        if (my $l = $c->{l}) {
4495                while ($l->[$#$l] eq "\n" && $#$l > 0
4496                                          && $l->[($#$l - 1)] eq "\n") {
4497                        pop @$l;
4498                }
4499                $nr_line = scalar @$l;
4500                if (!$nr_line) {
4501                        print "1 line\n\n\n";
4502                } else {
4503                        if ($nr_line == 1) {
4504                                $nr_line = '1 line';
4505                        } else {
4506                                $nr_line .= ' lines';
4507                        }
4508                        print $nr_line, "\n";
4509                        show_commit_changed_paths($c);
4510                        print "\n";
4511                        print $_ foreach @$l;
4512                }
4513        } else {
4514                print "1 line\n";
4515                show_commit_changed_paths($c);
4516                print "\n";
4517
4518        }
4519        foreach my $x (qw/raw stat diff/) {
4520                if ($c->{$x}) {
4521                        print "\n";
4522                        print $_ foreach @{$c->{$x}}
4523                }
4524        }
4525}
4526
4527sub cmd_show_log {
4528        my (@args) = @_;
4529        my ($r_min, $r_max);
4530        my $r_last = -1; # prevent dupes
4531        set_local_timezone();
4532        if (defined $::_revision) {
4533                if ($::_revision =~ /^(\d+):(\d+)$/) {
4534                        ($r_min, $r_max) = ($1, $2);
4535                } elsif ($::_revision =~ /^\d+$/) {
4536                        $r_min = $r_max = $::_revision;
4537                } else {
4538                        ::fatal "-r$::_revision is not supported, use ",
4539                                "standard 'git log' arguments instead";
4540                }
4541        }
4542
4543        config_pager();
4544        @args = git_svn_log_cmd($r_min, $r_max, @args);
4545        if (!@args) {
4546                print commit_log_separator unless $incremental || $oneline;
4547                return;
4548        }
4549        my $log = command_output_pipe(@args);
4550        run_pager();
4551        my (@k, $c, $d, $stat);
4552        my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
4553        while (<$log>) {
4554                if (/^${esc_color}commit -?($::sha1_short)/o) {
4555                        my $cmt = $1;
4556                        if ($c && cmt_showable($c) && $c->{r} != $r_last) {
4557                                $r_last = $c->{r};
4558                                process_commit($c, $r_min, $r_max, \@k) or
4559                                                                goto out;
4560                        }
4561                        $d = undef;
4562                        $c = { c => $cmt };
4563                } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
4564                        get_author_info($c, $1, $2, $3);
4565                } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
4566                        # ignore
4567                } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
4568                        push @{$c->{raw}}, $_;
4569                } elsif (/^${esc_color}[ACRMDT]\t/) {
4570                        # we could add $SVN->{svn_path} here, but that requires
4571                        # remote access at the moment (repo_path_split)...
4572                        s#^(${esc_color})([ACRMDT])\t#$1   $2 #o;
4573                        push @{$c->{changed}}, $_;
4574                } elsif (/^${esc_color}diff /o) {
4575                        $d = 1;
4576                        push @{$c->{diff}}, $_;
4577                } elsif ($d) {
4578                        push @{$c->{diff}}, $_;
4579                } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
4580                          $esc_color*[\+\-]*$esc_color$/x) {
4581                        $stat = 1;
4582                        push @{$c->{stat}}, $_;
4583                } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
4584                        push @{$c->{stat}}, $_;
4585                        $stat = undef;
4586                } elsif (/^${esc_color}    (git-svn-id:.+)$/o) {
4587                        ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
4588                } elsif (s/^${esc_color}    //o) {
4589                        push @{$c->{l}}, $_;
4590                }
4591        }
4592        if ($c && defined $c->{r} && $c->{r} != $r_last) {
4593                $r_last = $c->{r};
4594                process_commit($c, $r_min, $r_max, \@k);
4595        }
4596        if (@k) {
4597                ($r_min, $r_max) = ($r_max, $r_min);
4598                process_commit($_, $r_min, $r_max) foreach reverse @k;
4599        }
4600out:
4601        close $log;
4602        print commit_log_separator unless $incremental || $oneline;
4603}
4604
4605sub cmd_blame {
4606        my $path = pop;
4607
4608        config_pager();
4609        run_pager();
4610
4611        my ($fh, $ctx, $rev);
4612
4613        if ($_git_format) {
4614                ($fh, $ctx) = command_output_pipe('blame', @_, $path);
4615                while (my $line = <$fh>) {
4616                        if ($line =~ /^\^?([[:xdigit:]]+)\s/) {
4617                                # Uncommitted edits show up as a rev ID of
4618                                # all zeros, which we can't look up with
4619                                # cmt_metadata
4620                                if ($1 !~ /^0+$/) {
4621                                        (undef, $rev, undef) =
4622                                                ::cmt_metadata($1);
4623                                        $rev = '0' if (!$rev);
4624                                } else {
4625                                        $rev = '0';
4626                                }
4627                                $rev = sprintf('%-10s', $rev);
4628                                $line =~ s/^\^?[[:xdigit:]]+(\s)/$rev$1/;
4629                        }
4630                        print $line;
4631                }
4632        } else {
4633                ($fh, $ctx) = command_output_pipe('blame', '-p', @_, 'HEAD',
4634                                                  '--', $path);
4635                my ($sha1);
4636                my %authors;
4637                while (my $line = <$fh>) {
4638                        if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
4639                                $sha1 = $1;
4640                                (undef, $rev, undef) = ::cmt_metadata($1);
4641                                $rev = '0' if (!$rev);
4642                        }
4643                        elsif ($line =~ /^author (.*)/) {
4644                                $authors{$rev} = $1;
4645                                $authors{$rev} =~ s/\s/_/g;
4646                        }
4647                        elsif ($line =~ /^\t(.*)$/) {
4648                                printf("%6s %10s %s\n", $rev, $authors{$rev}, $1);
4649                        }
4650                }
4651        }
4652        command_close_pipe($fh, $ctx);
4653}
4654
4655package Git::SVN::Migration;
4656# these version numbers do NOT correspond to actual version numbers
4657# of git nor git-svn.  They are just relative.
4658#
4659# v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
4660#
4661# v1 layout: .git/$id/info/url, refs/remotes/$id
4662#
4663# v2 layout: .git/svn/$id/info/url, refs/remotes/$id
4664#
4665# v3 layout: .git/svn/$id, refs/remotes/$id
4666#            - info/url may remain for backwards compatibility
4667#            - this is what we migrate up to this layout automatically,
4668#            - this will be used by git svn init on single branches
4669# v3.1 layout (auto migrated):
4670#            - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
4671#              for backwards compatibility
4672#
4673# v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
4674#            - this is only created for newly multi-init-ed
4675#              repositories.  Similar in spirit to the
4676#              --use-separate-remotes option in git-clone (now default)
4677#            - we do not automatically migrate to this (following
4678#              the example set by core git)
4679#
4680# v5 layout: .rev_db.$UUID => .rev_map.$UUID
4681#            - newer, more-efficient format that uses 24-bytes per record
4682#              with no filler space.
4683#            - use xxd -c24 < .rev_map.$UUID to view and debug
4684#            - This is a one-way migration, repositories updated to the
4685#              new format will not be able to use old git-svn without
4686#              rebuilding the .rev_db.  Rebuilding the rev_db is not
4687#              possible if noMetadata or useSvmProps are set; but should
4688#              be no problem for users that use the (sensible) defaults.
4689use strict;
4690use warnings;
4691use Carp qw/croak/;
4692use File::Path qw/mkpath/;
4693use File::Basename qw/dirname basename/;
4694use vars qw/$_minimize/;
4695
4696sub migrate_from_v0 {
4697        my $git_dir = $ENV{GIT_DIR};
4698        return undef unless -d $git_dir;
4699        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
4700        my $migrated = 0;
4701        while (<$fh>) {
4702                chomp;
4703                my ($id, $orig_ref) = ($_, $_);
4704                next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
4705                next unless -f "$git_dir/$id/info/url";
4706                my $new_ref = "refs/remotes/$id";
4707                if (::verify_ref("$new_ref^0")) {
4708                        print STDERR "W: $orig_ref is probably an old ",
4709                                     "branch used by an ancient version of ",
4710                                     "git-svn.\n",
4711                                     "However, $new_ref also exists.\n",
4712                                     "We will not be able ",
4713                                     "to use this branch until this ",
4714                                     "ambiguity is resolved.\n";
4715                        next;
4716                }
4717                print STDERR "Migrating from v0 layout...\n" if !$migrated;
4718                print STDERR "Renaming ref: $orig_ref => $new_ref\n";
4719                command_noisy('update-ref', $new_ref, $orig_ref);
4720                command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
4721                $migrated++;
4722        }
4723        command_close_pipe($fh, $ctx);
4724        print STDERR "Done migrating from v0 layout...\n" if $migrated;
4725        $migrated;
4726}
4727
4728sub migrate_from_v1 {
4729        my $git_dir = $ENV{GIT_DIR};
4730        my $migrated = 0;
4731        return $migrated unless -d $git_dir;
4732        my $svn_dir = "$git_dir/svn";
4733
4734        # just in case somebody used 'svn' as their $id at some point...
4735        return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
4736
4737        print STDERR "Migrating from a git-svn v1 layout...\n";
4738        mkpath([$svn_dir]);
4739        print STDERR "Data from a previous version of git-svn exists, but\n\t",
4740                     "$svn_dir\n\t(required for this version ",
4741                     "($::VERSION) of git-svn) does not exist.\n";
4742        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
4743        while (<$fh>) {
4744                my $x = $_;
4745                next unless $x =~ s#^refs/remotes/##;
4746                chomp $x;
4747                next unless -f "$git_dir/$x/info/url";
4748                my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
4749                next unless $u;
4750                my $dn = dirname("$git_dir/svn/$x");
4751                mkpath([$dn]) unless -d $dn;
4752                if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
4753                        mkpath(["$git_dir/svn/svn"]);
4754                        print STDERR " - $git_dir/$x/info => ",
4755                                        "$git_dir/svn/$x/info\n";
4756                        rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
4757                               croak "$!: $x";
4758                        # don't worry too much about these, they probably
4759                        # don't exist with repos this old (save for index,
4760                        # and we can easily regenerate that)
4761                        foreach my $f (qw/unhandled.log index .rev_db/) {
4762                                rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
4763                        }
4764                } else {
4765                        print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
4766                        rename "$git_dir/$x", "$git_dir/svn/$x" or
4767                               croak "$!: $x";
4768                }
4769                $migrated++;
4770        }
4771        command_close_pipe($fh, $ctx);
4772        print STDERR "Done migrating from a git-svn v1 layout\n";
4773        $migrated;
4774}
4775
4776sub read_old_urls {
4777        my ($l_map, $pfx, $path) = @_;
4778        my @dir;
4779        foreach (<$path/*>) {
4780                if (-r "$_/info/url") {
4781                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
4782                        my $ref_id = $pfx . basename $_;
4783                        my $url = ::file_to_s("$_/info/url");
4784                        $l_map->{$ref_id} = $url;
4785                } elsif (-d $_) {
4786                        push @dir, $_;
4787                }
4788        }
4789        foreach (@dir) {
4790                my $x = $_;
4791                $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
4792                read_old_urls($l_map, $x, $_);
4793        }
4794}
4795
4796sub migrate_from_v2 {
4797        my @cfg = command(qw/config -l/);
4798        return if grep /^svn-remote\..+\.url=/, @cfg;
4799        my %l_map;
4800        read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
4801        my $migrated = 0;
4802
4803        foreach my $ref_id (sort keys %l_map) {
4804                eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
4805                if ($@) {
4806                        Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
4807                }
4808                $migrated++;
4809        }
4810        $migrated;
4811}
4812
4813sub minimize_connections {
4814        my $r = Git::SVN::read_all_remotes();
4815        my $new_urls = {};
4816        my $root_repos = {};
4817        foreach my $repo_id (keys %$r) {
4818                my $url = $r->{$repo_id}->{url} or next;
4819                my $fetch = $r->{$repo_id}->{fetch} or next;
4820                my $ra = Git::SVN::Ra->new($url);
4821
4822                # skip existing cases where we already connect to the root
4823                if (($ra->{url} eq $ra->{repos_root}) ||
4824                    ($ra->{repos_root} eq $repo_id)) {
4825                        $root_repos->{$ra->{url}} = $repo_id;
4826                        next;
4827                }
4828
4829                my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
4830                my $root_path = $ra->{url};
4831                $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
4832                foreach my $path (keys %$fetch) {
4833                        my $ref_id = $fetch->{$path};
4834                        my $gs = Git::SVN->new($ref_id, $repo_id, $path);
4835
4836                        # make sure we can read when connecting to
4837                        # a higher level of a repository
4838                        my ($last_rev, undef) = $gs->last_rev_commit;
4839                        if (!defined $last_rev) {
4840                                $last_rev = eval {
4841                                        $root_ra->get_latest_revnum;
4842                                };
4843                                next if $@;
4844                        }
4845                        my $new = $root_path;
4846                        $new .= length $path ? "/$path" : '';
4847                        eval {
4848                                $root_ra->get_log([$new], $last_rev, $last_rev,
4849                                                  0, 0, 1, sub { });
4850                        };
4851                        next if $@;
4852                        $new_urls->{$ra->{repos_root}}->{$new} =
4853                                { ref_id => $ref_id,
4854                                  old_repo_id => $repo_id,
4855                                  old_path => $path };
4856                }
4857        }
4858
4859        my @emptied;
4860        foreach my $url (keys %$new_urls) {
4861                # see if we can re-use an existing [svn-remote "repo_id"]
4862                # instead of creating a(n ugly) new section:
4863                my $repo_id = $root_repos->{$url} || $url;
4864
4865                my $fetch = $new_urls->{$url};
4866                foreach my $path (keys %$fetch) {
4867                        my $x = $fetch->{$path};
4868                        Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
4869                        my $pfx = "svn-remote.$x->{old_repo_id}";
4870
4871                        my $old_fetch = quotemeta("$x->{old_path}:".
4872                                                  "refs/remotes/$x->{ref_id}");
4873                        command_noisy(qw/config --unset/,
4874                                      "$pfx.fetch", '^'. $old_fetch . '$');
4875                        delete $r->{$x->{old_repo_id}}->
4876                               {fetch}->{$x->{old_path}};
4877                        if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
4878                                command_noisy(qw/config --unset/,
4879                                              "$pfx.url");
4880                                push @emptied, $x->{old_repo_id}
4881                        }
4882                }
4883        }
4884        if (@emptied) {
4885                my $file = $ENV{GIT_CONFIG} || $ENV{GIT_CONFIG_LOCAL} ||
4886                           "$ENV{GIT_DIR}/config";
4887                print STDERR <<EOF;
4888The following [svn-remote] sections in your config file ($file) are empty
4889and can be safely removed:
4890EOF
4891                print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
4892        }
4893}
4894
4895sub migration_check {
4896        migrate_from_v0();
4897        migrate_from_v1();
4898        migrate_from_v2();
4899        minimize_connections() if $_minimize;
4900}
4901
4902package Git::IndexInfo;
4903use strict;
4904use warnings;
4905use Git qw/command_input_pipe command_close_pipe/;
4906
4907sub new {
4908        my ($class) = @_;
4909        my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
4910        bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
4911}
4912
4913sub remove {
4914        my ($self, $path) = @_;
4915        if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
4916                return ++$self->{nr};
4917        }
4918        undef;
4919}
4920
4921sub update {
4922        my ($self, $mode, $hash, $path) = @_;
4923        if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
4924                return ++$self->{nr};
4925        }
4926        undef;
4927}
4928
4929sub DESTROY {
4930        my ($self) = @_;
4931        command_close_pipe($self->{gui}, $self->{ctx});
4932}
4933
4934package Git::SVN::GlobSpec;
4935use strict;
4936use warnings;
4937
4938sub new {
4939        my ($class, $glob) = @_;
4940        my $re = $glob;
4941        $re =~ s!/+$!!g; # no need for trailing slashes
4942        $re =~ m!^([^*]*)(\*(?:/\*)*)([^*]*)$!;
4943        my $temp = $re;
4944        my ($left, $right) = ($1, $3);
4945        $re = $2;
4946        my $depth = $re =~ tr/*/*/;
4947        if ($depth != $temp =~ tr/*/*/) {
4948                die "Only one set of wildcard directories " .
4949                        "(e.g. '*' or '*/*/*') is supported: '$glob'\n";
4950        }
4951        if ($depth == 0) {
4952                die "One '*' is needed for glob: '$glob'\n";
4953        }
4954        $re =~ s!\*!\[^/\]*!g;
4955        $re = quotemeta($left) . "($re)" . quotemeta($right);
4956        if (length $left && !($left =~ s!/+$!!g)) {
4957                die "Missing trailing '/' on left side of: '$glob' ($left)\n";
4958        }
4959        if (length $right && !($right =~ s!^/+!!g)) {
4960                die "Missing leading '/' on right side of: '$glob' ($right)\n";
4961        }
4962        my $left_re = qr/^\/\Q$left\E(\/|$)/;
4963        bless { left => $left, right => $right, left_regex => $left_re,
4964                regex => qr/$re/, glob => $glob, depth => $depth }, $class;
4965}
4966
4967sub full_path {
4968        my ($self, $path) = @_;
4969        return (length $self->{left} ? "$self->{left}/" : '') .
4970               $path . (length $self->{right} ? "/$self->{right}" : '');
4971}
4972
4973__END__
4974
4975Data structures:
4976
4977
4978$remotes = { # returned by read_all_remotes()
4979        'svn' => {
4980                # svn-remote.svn.url=https://svn.musicpd.org
4981                url => 'https://svn.musicpd.org',
4982                # svn-remote.svn.fetch=mpd/trunk:trunk
4983                fetch => {
4984                        'mpd/trunk' => 'trunk',
4985                },
4986                # svn-remote.svn.tags=mpd/tags/*:tags/*
4987                tags => {
4988                        path => {
4989                                left => 'mpd/tags',
4990                                right => '',
4991                                regex => qr!mpd/tags/([^/]+)$!,
4992                                glob => 'tags/*',
4993                        },
4994                        ref => {
4995                                left => 'tags',
4996                                right => '',
4997                                regex => qr!tags/([^/]+)$!,
4998                                glob => 'tags/*',
4999                        },
5000                }
5001        }
5002};
5003
5004$log_entry hashref as returned by libsvn_log_entry()
5005{
5006        log => 'whitespace-formatted log entry
5007',                                              # trailing newline is preserved
5008        revision => '8',                        # integer
5009        date => '2004-02-24T17:01:44.108345Z',  # commit date
5010        author => 'committer name'
5011};
5012
5013
5014# this is generated by generate_diff();
5015@mods = array of diff-index line hashes, each element represents one line
5016        of diff-index output
5017
5018diff-index line ($m hash)
5019{
5020        mode_a => first column of diff-index output, no leading ':',
5021        mode_b => second column of diff-index output,
5022        sha1_b => sha1sum of the final blob,
5023        chg => change type [MCRADT],
5024        file_a => original file name of a file (iff chg is 'C' or 'R')
5025        file_b => new/current file name of a file (any chg)
5026}
5027;
5028
5029# retval of read_url_paths{,_all}();
5030$l_map = {
5031        # repository root url
5032        'https://svn.musicpd.org' => {
5033                # repository path               # GIT_SVN_ID
5034                'mpd/trunk'             =>      'trunk',
5035                'mpd/tags/0.11.5'       =>      'tags/0.11.5',
5036        },
5037}
5038
5039Notes:
5040        I don't trust the each() function on unless I created %hash myself
5041        because the internal iterator may not have started at base.