git-svn.perlon commit autodetect number of CPUs by default when using threads (43cc2b4)
   1#!/usr/bin/env perl
   2# Copyright (C) 2006, Eric Wong <normalperson@yhbt.net>
   3# License: GPL v2 or later
   4use warnings;
   5use strict;
   6use vars qw/    $AUTHOR $VERSION
   7                $sha1 $sha1_short $_revision $_repository
   8                $_q $_authors %users/;
   9$AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
  10$VERSION = '@@GIT_VERSION@@';
  11
  12# From which subdir have we been invoked?
  13my $cmd_dir_prefix = eval {
  14        command_oneline([qw/rev-parse --show-prefix/], STDERR => 0)
  15} || '';
  16
  17my $git_dir_user_set = 1 if defined $ENV{GIT_DIR};
  18$ENV{GIT_DIR} ||= '.git';
  19$Git::SVN::default_repo_id = 'svn';
  20$Git::SVN::default_ref_id = $ENV{GIT_SVN_ID} || 'git-svn';
  21$Git::SVN::Ra::_log_window_size = 100;
  22
  23$Git::SVN::Log::TZ = $ENV{TZ};
  24$ENV{TZ} = 'UTC';
  25$| = 1; # unbuffer STDOUT
  26
  27sub fatal (@) { print STDERR "@_\n"; exit 1 }
  28require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
  29require SVN::Ra;
  30require SVN::Delta;
  31if ($SVN::Core::VERSION lt '1.1.0') {
  32        fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)";
  33}
  34push @Git::SVN::Ra::ISA, 'SVN::Ra';
  35push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
  36push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
  37use Carp qw/croak/;
  38use Digest::MD5;
  39use IO::File qw//;
  40use File::Basename qw/dirname basename/;
  41use File::Path qw/mkpath/;
  42use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev/;
  43use IPC::Open3;
  44use Git;
  45
  46BEGIN {
  47        # import functions from Git into our packages, en masse
  48        no strict 'refs';
  49        foreach (qw/command command_oneline command_noisy command_output_pipe
  50                    command_input_pipe command_close_pipe/) {
  51                for my $package ( qw(SVN::Git::Editor SVN::Git::Fetcher
  52                        Git::SVN::Migration Git::SVN::Log Git::SVN),
  53                        __PACKAGE__) {
  54                        *{"${package}::$_"} = \&{"Git::$_"};
  55                }
  56        }
  57}
  58
  59my ($SVN);
  60
  61$sha1 = qr/[a-f\d]{40}/;
  62$sha1_short = qr/[a-f\d]{4,40}/;
  63my ($_stdin, $_help, $_edit,
  64        $_message, $_file,
  65        $_template, $_shared,
  66        $_version, $_fetch_all, $_no_rebase,
  67        $_merge, $_strategy, $_dry_run, $_local,
  68        $_prefix, $_no_checkout, $_url, $_verbose,
  69        $_git_format, $_commit_url, $_tag);
  70$Git::SVN::_follow_parent = 1;
  71my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
  72                    'config-dir=s' => \$Git::SVN::Ra::config_dir,
  73                    'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache );
  74my %fc_opts = ( 'follow-parent|follow!' => \$Git::SVN::_follow_parent,
  75                'authors-file|A=s' => \$_authors,
  76                'repack:i' => \$Git::SVN::_repack,
  77                'noMetadata' => \$Git::SVN::_no_metadata,
  78                'useSvmProps' => \$Git::SVN::_use_svm_props,
  79                'useSvnsyncProps' => \$Git::SVN::_use_svnsync_props,
  80                'log-window-size=i' => \$Git::SVN::Ra::_log_window_size,
  81                'no-checkout' => \$_no_checkout,
  82                'quiet|q' => \$_q,
  83                'repack-flags|repack-args|repack-opts=s' =>
  84                   \$Git::SVN::_repack_flags,
  85                'use-log-author' => \$Git::SVN::_use_log_author,
  86                'add-author-from' => \$Git::SVN::_add_author_from,
  87                %remote_opts );
  88
  89my ($_trunk, $_tags, $_branches, $_stdlayout);
  90my %icv;
  91my %init_opts = ( 'template=s' => \$_template, 'shared:s' => \$_shared,
  92                  'trunk|T=s' => \$_trunk, 'tags|t=s' => \$_tags,
  93                  'branches|b=s' => \$_branches, 'prefix=s' => \$_prefix,
  94                  'stdlayout|s' => \$_stdlayout,
  95                  'minimize-url|m' => \$Git::SVN::_minimize_url,
  96                  'no-metadata' => sub { $icv{noMetadata} = 1 },
  97                  'use-svm-props' => sub { $icv{useSvmProps} = 1 },
  98                  'use-svnsync-props' => sub { $icv{useSvnsyncProps} = 1 },
  99                  'rewrite-root=s' => sub { $icv{rewriteRoot} = $_[1] },
 100                  %remote_opts );
 101my %cmt_opts = ( 'edit|e' => \$_edit,
 102                'rmdir' => \$SVN::Git::Editor::_rmdir,
 103                'find-copies-harder' => \$SVN::Git::Editor::_find_copies_harder,
 104                'l=i' => \$SVN::Git::Editor::_rename_limit,
 105                'copy-similarity|C=i'=> \$SVN::Git::Editor::_cp_similarity
 106);
 107
 108my %cmd = (
 109        fetch => [ \&cmd_fetch, "Download new revisions from SVN",
 110                        { 'revision|r=s' => \$_revision,
 111                          'fetch-all|all' => \$_fetch_all,
 112                           %fc_opts } ],
 113        clone => [ \&cmd_clone, "Initialize and fetch revisions",
 114                        { 'revision|r=s' => \$_revision,
 115                           %fc_opts, %init_opts } ],
 116        init => [ \&cmd_init, "Initialize a repo for tracking" .
 117                          " (requires URL argument)",
 118                          \%init_opts ],
 119        'multi-init' => [ \&cmd_multi_init,
 120                          "Deprecated alias for ".
 121                          "'$0 init -T<trunk> -b<branches> -t<tags>'",
 122                          \%init_opts ],
 123        dcommit => [ \&cmd_dcommit,
 124                     'Commit several diffs to merge with upstream',
 125                        { 'merge|m|M' => \$_merge,
 126                          'strategy|s=s' => \$_strategy,
 127                          'verbose|v' => \$_verbose,
 128                          'dry-run|n' => \$_dry_run,
 129                          'fetch-all|all' => \$_fetch_all,
 130                          'commit-url=s' => \$_commit_url,
 131                          'revision|r=i' => \$_revision,
 132                          'no-rebase' => \$_no_rebase,
 133                        %cmt_opts, %fc_opts } ],
 134        branch => [ \&cmd_branch,
 135                    'Create a branch in the SVN repository',
 136                    { 'message|m=s' => \$_message,
 137                      'dry-run|n' => \$_dry_run,
 138                      'tag|t' => \$_tag } ],
 139        tag => [ sub { $_tag = 1; cmd_branch(@_) },
 140                 'Create a tag in the SVN repository',
 141                 { 'message|m=s' => \$_message,
 142                   'dry-run|n' => \$_dry_run } ],
 143        'set-tree' => [ \&cmd_set_tree,
 144                        "Set an SVN repository to a git tree-ish",
 145                        { 'stdin|' => \$_stdin, %cmt_opts, %fc_opts, } ],
 146        'create-ignore' => [ \&cmd_create_ignore,
 147                             'Create a .gitignore per svn:ignore',
 148                             { 'revision|r=i' => \$_revision
 149                             } ],
 150        'propget' => [ \&cmd_propget,
 151                       'Print the value of a property on a file or directory',
 152                       { 'revision|r=i' => \$_revision } ],
 153        'proplist' => [ \&cmd_proplist,
 154                       'List all properties of a file or directory',
 155                       { 'revision|r=i' => \$_revision } ],
 156        'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
 157                        { 'revision|r=i' => \$_revision
 158                        } ],
 159        'show-externals' => [ \&cmd_show_externals, "Show svn:externals listings",
 160                        { 'revision|r=i' => \$_revision
 161                        } ],
 162        'multi-fetch' => [ \&cmd_multi_fetch,
 163                           "Deprecated alias for $0 fetch --all",
 164                           { 'revision|r=s' => \$_revision, %fc_opts } ],
 165        'migrate' => [ sub { },
 166                       # no-op, we automatically run this anyways,
 167                       'Migrate configuration/metadata/layout from
 168                        previous versions of git-svn',
 169                       { 'minimize' => \$Git::SVN::Migration::_minimize,
 170                         %remote_opts } ],
 171        'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
 172                        { 'limit=i' => \$Git::SVN::Log::limit,
 173                          'revision|r=s' => \$_revision,
 174                          'verbose|v' => \$Git::SVN::Log::verbose,
 175                          'incremental' => \$Git::SVN::Log::incremental,
 176                          'oneline' => \$Git::SVN::Log::oneline,
 177                          'show-commit' => \$Git::SVN::Log::show_commit,
 178                          'non-recursive' => \$Git::SVN::Log::non_recursive,
 179                          'authors-file|A=s' => \$_authors,
 180                          'color' => \$Git::SVN::Log::color,
 181                          'pager=s' => \$Git::SVN::Log::pager
 182                        } ],
 183        'find-rev' => [ \&cmd_find_rev,
 184                        "Translate between SVN revision numbers and tree-ish",
 185                        {} ],
 186        'rebase' => [ \&cmd_rebase, "Fetch and rebase your working directory",
 187                        { 'merge|m|M' => \$_merge,
 188                          'verbose|v' => \$_verbose,
 189                          'strategy|s=s' => \$_strategy,
 190                          'local|l' => \$_local,
 191                          'fetch-all|all' => \$_fetch_all,
 192                          'dry-run|n' => \$_dry_run,
 193                          %fc_opts } ],
 194        'commit-diff' => [ \&cmd_commit_diff,
 195                           'Commit a diff between two trees',
 196                        { 'message|m=s' => \$_message,
 197                          'file|F=s' => \$_file,
 198                          'revision|r=s' => \$_revision,
 199                        %cmt_opts } ],
 200        'info' => [ \&cmd_info,
 201                    "Show info about the latest SVN revision
 202                     on the current branch",
 203                    { 'url' => \$_url, } ],
 204        'blame' => [ \&Git::SVN::Log::cmd_blame,
 205                    "Show what revision and author last modified each line of a file",
 206                    { 'git-format' => \$_git_format } ],
 207);
 208
 209my $cmd;
 210for (my $i = 0; $i < @ARGV; $i++) {
 211        if (defined $cmd{$ARGV[$i]}) {
 212                $cmd = $ARGV[$i];
 213                splice @ARGV, $i, 1;
 214                last;
 215        }
 216};
 217
 218# make sure we're always running at the top-level working directory
 219unless ($cmd && $cmd =~ /(?:clone|init|multi-init)$/) {
 220        unless (-d $ENV{GIT_DIR}) {
 221                if ($git_dir_user_set) {
 222                        die "GIT_DIR=$ENV{GIT_DIR} explicitly set, ",
 223                            "but it is not a directory\n";
 224                }
 225                my $git_dir = delete $ENV{GIT_DIR};
 226                my $cdup = undef;
 227                git_cmd_try {
 228                        $cdup = command_oneline(qw/rev-parse --show-cdup/);
 229                        $git_dir = '.' unless ($cdup);
 230                        chomp $cdup if ($cdup);
 231                        $cdup = "." unless ($cdup && length $cdup);
 232                } "Already at toplevel, but $git_dir not found\n";
 233                chdir $cdup or die "Unable to chdir up to '$cdup'\n";
 234                unless (-d $git_dir) {
 235                        die "$git_dir still not found after going to ",
 236                            "'$cdup'\n";
 237                }
 238                $ENV{GIT_DIR} = $git_dir;
 239        }
 240        $_repository = Git->repository(Repository => $ENV{GIT_DIR});
 241}
 242
 243my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
 244
 245read_repo_config(\%opts);
 246if ($cmd && ($cmd eq 'log' || $cmd eq 'blame')) {
 247        Getopt::Long::Configure('pass_through');
 248}
 249my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
 250                    'minimize-connections' => \$Git::SVN::Migration::_minimize,
 251                    'id|i=s' => \$Git::SVN::default_ref_id,
 252                    'svn-remote|remote|R=s' => sub {
 253                       $Git::SVN::no_reuse_existing = 1;
 254                       $Git::SVN::default_repo_id = $_[1] });
 255exit 1 if (!$rv && $cmd && $cmd ne 'log');
 256
 257usage(0) if $_help;
 258version() if $_version;
 259usage(1) unless defined $cmd;
 260load_authors() if $_authors;
 261
 262unless ($cmd =~ /^(?:clone|init|multi-init|commit-diff)$/) {
 263        Git::SVN::Migration::migration_check();
 264}
 265Git::SVN::init_vars();
 266eval {
 267        Git::SVN::verify_remotes_sanity();
 268        $cmd{$cmd}->[0]->(@ARGV);
 269};
 270fatal $@ if $@;
 271post_fetch_checkout();
 272exit 0;
 273
 274####################### primary functions ######################
 275sub usage {
 276        my $exit = shift || 0;
 277        my $fd = $exit ? \*STDERR : \*STDOUT;
 278        print $fd <<"";
 279git-svn - bidirectional operations between a single Subversion tree and git
 280Usage: git svn <command> [options] [arguments]\n
 281
 282        print $fd "Available commands:\n" unless $cmd;
 283
 284        foreach (sort keys %cmd) {
 285                next if $cmd && $cmd ne $_;
 286                next if /^multi-/; # don't show deprecated commands
 287                print $fd '  ',pack('A17',$_),$cmd{$_}->[1],"\n";
 288                foreach (sort keys %{$cmd{$_}->[2]}) {
 289                        # mixed-case options are for .git/config only
 290                        next if /[A-Z]/ && /^[a-z]+$/i;
 291                        # prints out arguments as they should be passed:
 292                        my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
 293                        print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
 294                                                        "--$_" : "-$_" }
 295                                                split /\|/,$_)," $x\n";
 296                }
 297        }
 298        print $fd <<"";
 299\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
 300arbitrary identifier if you're tracking multiple SVN branches/repositories in
 301one git repository and want to keep them separate.  See git-svn(1) for more
 302information.
 303
 304        exit $exit;
 305}
 306
 307sub version {
 308        print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
 309        exit 0;
 310}
 311
 312sub do_git_init_db {
 313        unless (-d $ENV{GIT_DIR}) {
 314                my @init_db = ('init');
 315                push @init_db, "--template=$_template" if defined $_template;
 316                if (defined $_shared) {
 317                        if ($_shared =~ /[a-z]/) {
 318                                push @init_db, "--shared=$_shared";
 319                        } else {
 320                                push @init_db, "--shared";
 321                        }
 322                }
 323                command_noisy(@init_db);
 324                $_repository = Git->repository(Repository => ".git");
 325        }
 326        my $set;
 327        my $pfx = "svn-remote.$Git::SVN::default_repo_id";
 328        foreach my $i (keys %icv) {
 329                die "'$set' and '$i' cannot both be set\n" if $set;
 330                next unless defined $icv{$i};
 331                command_noisy('config', "$pfx.$i", $icv{$i});
 332                $set = $i;
 333        }
 334}
 335
 336sub init_subdir {
 337        my $repo_path = shift or return;
 338        mkpath([$repo_path]) unless -d $repo_path;
 339        chdir $repo_path or die "Couldn't chdir to $repo_path: $!\n";
 340        $ENV{GIT_DIR} = '.git';
 341        $_repository = Git->repository(Repository => $ENV{GIT_DIR});
 342}
 343
 344sub cmd_clone {
 345        my ($url, $path) = @_;
 346        if (!defined $path &&
 347            (defined $_trunk || defined $_branches || defined $_tags ||
 348             defined $_stdlayout) &&
 349            $url !~ m#^[a-z\+]+://#) {
 350                $path = $url;
 351        }
 352        $path = basename($url) if !defined $path || !length $path;
 353        cmd_init($url, $path);
 354        Git::SVN::fetch_all($Git::SVN::default_repo_id);
 355}
 356
 357sub cmd_init {
 358        if (defined $_stdlayout) {
 359                $_trunk = 'trunk' if (!defined $_trunk);
 360                $_tags = 'tags' if (!defined $_tags);
 361                $_branches = 'branches' if (!defined $_branches);
 362        }
 363        if (defined $_trunk || defined $_branches || defined $_tags) {
 364                return cmd_multi_init(@_);
 365        }
 366        my $url = shift or die "SVN repository location required ",
 367                               "as a command-line argument\n";
 368        init_subdir(@_);
 369        do_git_init_db();
 370
 371        Git::SVN->init($url);
 372}
 373
 374sub cmd_fetch {
 375        if (grep /^\d+=./, @_) {
 376                die "'<rev>=<commit>' fetch arguments are ",
 377                    "no longer supported.\n";
 378        }
 379        my ($remote) = @_;
 380        if (@_ > 1) {
 381                die "Usage: $0 fetch [--all] [svn-remote]\n";
 382        }
 383        $remote ||= $Git::SVN::default_repo_id;
 384        if ($_fetch_all) {
 385                cmd_multi_fetch();
 386        } else {
 387                Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
 388        }
 389}
 390
 391sub cmd_set_tree {
 392        my (@commits) = @_;
 393        if ($_stdin || !@commits) {
 394                print "Reading from stdin...\n";
 395                @commits = ();
 396                while (<STDIN>) {
 397                        if (/\b($sha1_short)\b/o) {
 398                                unshift @commits, $1;
 399                        }
 400                }
 401        }
 402        my @revs;
 403        foreach my $c (@commits) {
 404                my @tmp = command('rev-parse',$c);
 405                if (scalar @tmp == 1) {
 406                        push @revs, $tmp[0];
 407                } elsif (scalar @tmp > 1) {
 408                        push @revs, reverse(command('rev-list',@tmp));
 409                } else {
 410                        fatal "Failed to rev-parse $c";
 411                }
 412        }
 413        my $gs = Git::SVN->new;
 414        my ($r_last, $cmt_last) = $gs->last_rev_commit;
 415        $gs->fetch;
 416        if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
 417                fatal "There are new revisions that were fetched ",
 418                      "and need to be merged (or acknowledged) ",
 419                      "before committing.\nlast rev: $r_last\n",
 420                      " current: $gs->{last_rev}";
 421        }
 422        $gs->set_tree($_) foreach @revs;
 423        print "Done committing ",scalar @revs," revisions to SVN\n";
 424        unlink $gs->{index};
 425}
 426
 427sub cmd_dcommit {
 428        my $head = shift;
 429        git_cmd_try { command_oneline(qw/diff-index --quiet HEAD/) }
 430                'Cannot dcommit with a dirty index.  Commit your changes first, '
 431                . "or stash them with `git stash'.\n";
 432        $head ||= 'HEAD';
 433        my @refs;
 434        my ($url, $rev, $uuid, $gs) = working_head_info($head, \@refs);
 435        unless ($gs) {
 436                die "Unable to determine upstream SVN information from ",
 437                    "$head history.\nPerhaps the repository is empty.";
 438        }
 439        $url = defined $_commit_url ? $_commit_url : $gs->full_url;
 440        my $last_rev = $_revision if defined $_revision;
 441        if ($url) {
 442                print "Committing to $url ...\n";
 443        }
 444        my ($linear_refs, $parents) = linearize_history($gs, \@refs);
 445        if ($_no_rebase && scalar(@$linear_refs) > 1) {
 446                warn "Attempting to commit more than one change while ",
 447                     "--no-rebase is enabled.\n",
 448                     "If these changes depend on each other, re-running ",
 449                     "without --no-rebase may be required."
 450        }
 451        my $expect_url = $url;
 452        Git::SVN::remove_username($expect_url);
 453        while (1) {
 454                my $d = shift @$linear_refs or last;
 455                unless (defined $last_rev) {
 456                        (undef, $last_rev, undef) = cmt_metadata("$d~1");
 457                        unless (defined $last_rev) {
 458                                fatal "Unable to extract revision information ",
 459                                      "from commit $d~1";
 460                        }
 461                }
 462                if ($_dry_run) {
 463                        print "diff-tree $d~1 $d\n";
 464                } else {
 465                        my $cmt_rev;
 466                        my %ed_opts = ( r => $last_rev,
 467                                        log => get_commit_entry($d)->{log},
 468                                        ra => Git::SVN::Ra->new($url),
 469                                        config => SVN::Core::config_get_config(
 470                                                $Git::SVN::Ra::config_dir
 471                                        ),
 472                                        tree_a => "$d~1",
 473                                        tree_b => $d,
 474                                        editor_cb => sub {
 475                                               print "Committed r$_[0]\n";
 476                                               $cmt_rev = $_[0];
 477                                        },
 478                                        svn_path => '');
 479                        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
 480                                print "No changes\n$d~1 == $d\n";
 481                        } elsif ($parents->{$d} && @{$parents->{$d}}) {
 482                                $gs->{inject_parents_dcommit}->{$cmt_rev} =
 483                                                               $parents->{$d};
 484                        }
 485                        $_fetch_all ? $gs->fetch_all : $gs->fetch;
 486                        $last_rev = $cmt_rev;
 487                        next if $_no_rebase;
 488
 489                        # we always want to rebase against the current HEAD,
 490                        # not any head that was passed to us
 491                        my @diff = command('diff-tree', $d,
 492                                           $gs->refname, '--');
 493                        my @finish;
 494                        if (@diff) {
 495                                @finish = rebase_cmd();
 496                                print STDERR "W: $d and ", $gs->refname,
 497                                             " differ, using @finish:\n",
 498                                             join("\n", @diff), "\n";
 499                        } else {
 500                                print "No changes between current HEAD and ",
 501                                      $gs->refname,
 502                                      "\nResetting to the latest ",
 503                                      $gs->refname, "\n";
 504                                @finish = qw/reset --mixed/;
 505                        }
 506                        command_noisy(@finish, $gs->refname);
 507                        if (@diff) {
 508                                @refs = ();
 509                                my ($url_, $rev_, $uuid_, $gs_) =
 510                                              working_head_info($head, \@refs);
 511                                my ($linear_refs_, $parents_) =
 512                                              linearize_history($gs_, \@refs);
 513                                if (scalar(@$linear_refs) !=
 514                                    scalar(@$linear_refs_)) {
 515                                        fatal "# of revisions changed ",
 516                                          "\nbefore:\n",
 517                                          join("\n", @$linear_refs),
 518                                          "\n\nafter:\n",
 519                                          join("\n", @$linear_refs_), "\n",
 520                                          'If you are attempting to commit ',
 521                                          "merges, try running:\n\t",
 522                                          'git rebase --interactive',
 523                                          '--preserve-merges ',
 524                                          $gs->refname,
 525                                          "\nBefore dcommitting";
 526                                }
 527                                if ($url_ ne $expect_url) {
 528                                        fatal "URL mismatch after rebase: ",
 529                                              "$url_ != $expect_url";
 530                                }
 531                                if ($uuid_ ne $uuid) {
 532                                        fatal "uuid mismatch after rebase: ",
 533                                              "$uuid_ != $uuid";
 534                                }
 535                                # remap parents
 536                                my (%p, @l, $i);
 537                                for ($i = 0; $i < scalar @$linear_refs; $i++) {
 538                                        my $new = $linear_refs_->[$i] or next;
 539                                        $p{$new} =
 540                                                $parents->{$linear_refs->[$i]};
 541                                        push @l, $new;
 542                                }
 543                                $parents = \%p;
 544                                $linear_refs = \@l;
 545                        }
 546                }
 547        }
 548        unlink $gs->{index};
 549}
 550
 551sub cmd_branch {
 552        my ($branch_name, $head) = @_;
 553
 554        unless (defined $branch_name && length $branch_name) {
 555                die(($_tag ? "tag" : "branch") . " name required\n");
 556        }
 557        $head ||= 'HEAD';
 558
 559        my ($src, $rev, undef, $gs) = working_head_info($head);
 560
 561        my $remote = Git::SVN::read_all_remotes()->{$gs->{repo_id}};
 562        my $glob = $remote->{ $_tag ? 'tags' : 'branches' };
 563        my ($lft, $rgt) = @{ $glob->{path} }{qw/left right/};
 564        my $dst = join '/', $remote->{url}, $lft, $branch_name, ($rgt || ());
 565
 566        my $ctx = SVN::Client->new(
 567                auth    => Git::SVN::Ra::_auth_providers(),
 568                log_msg => sub {
 569                        ${ $_[0] } = defined $_message
 570                                ? $_message
 571                                : 'Create ' . ($_tag ? 'tag ' : 'branch ' )
 572                                . $branch_name;
 573                },
 574        );
 575
 576        eval {
 577                $ctx->ls($dst, 'HEAD', 0);
 578        } and die "branch ${branch_name} already exists\n";
 579
 580        print "Copying ${src} at r${rev} to ${dst}...\n";
 581        $ctx->copy($src, $rev, $dst)
 582                unless $_dry_run;
 583
 584        $gs->fetch_all;
 585}
 586
 587sub cmd_find_rev {
 588        my $revision_or_hash = shift or die "SVN or git revision required ",
 589                                            "as a command-line argument\n";
 590        my $result;
 591        if ($revision_or_hash =~ /^r\d+$/) {
 592                my $head = shift;
 593                $head ||= 'HEAD';
 594                my @refs;
 595                my (undef, undef, $uuid, $gs) = working_head_info($head, \@refs);
 596                unless ($gs) {
 597                        die "Unable to determine upstream SVN information from ",
 598                            "$head history\n";
 599                }
 600                my $desired_revision = substr($revision_or_hash, 1);
 601                $result = $gs->rev_map_get($desired_revision, $uuid);
 602        } else {
 603                my (undef, $rev, undef) = cmt_metadata($revision_or_hash);
 604                $result = $rev;
 605        }
 606        print "$result\n" if $result;
 607}
 608
 609sub cmd_rebase {
 610        command_noisy(qw/update-index --refresh/);
 611        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 612        unless ($gs) {
 613                die "Unable to determine upstream SVN information from ",
 614                    "working tree history\n";
 615        }
 616        if ($_dry_run) {
 617                print "Remote Branch: " . $gs->refname . "\n";
 618                print "SVN URL: " . $url . "\n";
 619                return;
 620        }
 621        if (command(qw/diff-index HEAD --/)) {
 622                print STDERR "Cannot rebase with uncommited changes:\n";
 623                command_noisy('status');
 624                exit 1;
 625        }
 626        unless ($_local) {
 627                # rebase will checkout for us, so no need to do it explicitly
 628                $_no_checkout = 'true';
 629                $_fetch_all ? $gs->fetch_all : $gs->fetch;
 630        }
 631        command_noisy(rebase_cmd(), $gs->refname);
 632}
 633
 634sub cmd_show_ignore {
 635        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 636        $gs ||= Git::SVN->new;
 637        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 638        $gs->prop_walk($gs->{path}, $r, sub {
 639                my ($gs, $path, $props) = @_;
 640                print STDOUT "\n# $path\n";
 641                my $s = $props->{'svn:ignore'} or return;
 642                $s =~ s/[\r\n]+/\n/g;
 643                chomp $s;
 644                $s =~ s#^#$path#gm;
 645                print STDOUT "$s\n";
 646        });
 647}
 648
 649sub cmd_show_externals {
 650        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 651        $gs ||= Git::SVN->new;
 652        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 653        $gs->prop_walk($gs->{path}, $r, sub {
 654                my ($gs, $path, $props) = @_;
 655                print STDOUT "\n# $path\n";
 656                my $s = $props->{'svn:externals'} or return;
 657                $s =~ s/[\r\n]+/\n/g;
 658                chomp $s;
 659                $s =~ s#^#$path#gm;
 660                print STDOUT "$s\n";
 661        });
 662}
 663
 664sub cmd_create_ignore {
 665        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 666        $gs ||= Git::SVN->new;
 667        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 668        $gs->prop_walk($gs->{path}, $r, sub {
 669                my ($gs, $path, $props) = @_;
 670                # $path is of the form /path/to/dir/
 671                my $ignore = '.' . $path . '.gitignore';
 672                my $s = $props->{'svn:ignore'} or return;
 673                open(GITIGNORE, '>', $ignore)
 674                  or fatal("Failed to open `$ignore' for writing: $!");
 675                $s =~ s/[\r\n]+/\n/g;
 676                chomp $s;
 677                # Prefix all patterns so that the ignore doesn't apply
 678                # to sub-directories.
 679                $s =~ s#^#/#gm;
 680                print GITIGNORE "$s\n";
 681                close(GITIGNORE)
 682                  or fatal("Failed to close `$ignore': $!");
 683                command_noisy('add', '-f', $ignore);
 684        });
 685}
 686
 687sub canonicalize_path {
 688        my ($path) = @_;
 689        my $dot_slash_added = 0;
 690        if (substr($path, 0, 1) ne "/") {
 691                $path = "./" . $path;
 692                $dot_slash_added = 1;
 693        }
 694        # File::Spec->canonpath doesn't collapse x/../y into y (for a
 695        # good reason), so let's do this manually.
 696        $path =~ s#/+#/#g;
 697        $path =~ s#/\.(?:/|$)#/#g;
 698        $path =~ s#/[^/]+/\.\.##g;
 699        $path =~ s#/$##g;
 700        $path =~ s#^\./## if $dot_slash_added;
 701        $path =~ s#^/##;
 702        $path =~ s#^\.$##;
 703        return $path;
 704}
 705
 706# get_svnprops(PATH)
 707# ------------------
 708# Helper for cmd_propget and cmd_proplist below.
 709sub get_svnprops {
 710        my $path = shift;
 711        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 712        $gs ||= Git::SVN->new;
 713
 714        # prefix THE PATH by the sub-directory from which the user
 715        # invoked us.
 716        $path = $cmd_dir_prefix . $path;
 717        fatal("No such file or directory: $path") unless -e $path;
 718        my $is_dir = -d $path ? 1 : 0;
 719        $path = $gs->{path} . '/' . $path;
 720
 721        # canonicalize the path (otherwise libsvn will abort or fail to
 722        # find the file)
 723        $path = canonicalize_path($path);
 724
 725        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 726        my $props;
 727        if ($is_dir) {
 728                (undef, undef, $props) = $gs->ra->get_dir($path, $r);
 729        }
 730        else {
 731                (undef, $props) = $gs->ra->get_file($path, $r, undef);
 732        }
 733        return $props;
 734}
 735
 736# cmd_propget (PROP, PATH)
 737# ------------------------
 738# Print the SVN property PROP for PATH.
 739sub cmd_propget {
 740        my ($prop, $path) = @_;
 741        $path = '.' if not defined $path;
 742        usage(1) if not defined $prop;
 743        my $props = get_svnprops($path);
 744        if (not defined $props->{$prop}) {
 745                fatal("`$path' does not have a `$prop' SVN property.");
 746        }
 747        print $props->{$prop} . "\n";
 748}
 749
 750# cmd_proplist (PATH)
 751# -------------------
 752# Print the list of SVN properties for PATH.
 753sub cmd_proplist {
 754        my $path = shift;
 755        $path = '.' if not defined $path;
 756        my $props = get_svnprops($path);
 757        print "Properties on '$path':\n";
 758        foreach (sort keys %{$props}) {
 759                print "  $_\n";
 760        }
 761}
 762
 763sub cmd_multi_init {
 764        my $url = shift;
 765        unless (defined $_trunk || defined $_branches || defined $_tags) {
 766                usage(1);
 767        }
 768
 769        # there are currently some bugs that prevent multi-init/multi-fetch
 770        # setups from working well without this.
 771        $Git::SVN::_minimize_url = 1;
 772
 773        $_prefix = '' unless defined $_prefix;
 774        if (defined $url) {
 775                $url =~ s#/+$##;
 776                init_subdir(@_);
 777        }
 778        do_git_init_db();
 779        if (defined $_trunk) {
 780                my $trunk_ref = $_prefix . 'trunk';
 781                # try both old-style and new-style lookups:
 782                my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
 783                unless ($gs_trunk) {
 784                        my ($trunk_url, $trunk_path) =
 785                                              complete_svn_url($url, $_trunk);
 786                        $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
 787                                                   undef, $trunk_ref);
 788                }
 789        }
 790        return unless defined $_branches || defined $_tags;
 791        my $ra = $url ? Git::SVN::Ra->new($url) : undef;
 792        complete_url_ls_init($ra, $_branches, '--branches/-b', $_prefix);
 793        complete_url_ls_init($ra, $_tags, '--tags/-t', $_prefix . 'tags/');
 794}
 795
 796sub cmd_multi_fetch {
 797        my $remotes = Git::SVN::read_all_remotes();
 798        foreach my $repo_id (sort keys %$remotes) {
 799                if ($remotes->{$repo_id}->{url}) {
 800                        Git::SVN::fetch_all($repo_id, $remotes);
 801                }
 802        }
 803}
 804
 805# this command is special because it requires no metadata
 806sub cmd_commit_diff {
 807        my ($ta, $tb, $url) = @_;
 808        my $usage = "Usage: $0 commit-diff -r<revision> ".
 809                    "<tree-ish> <tree-ish> [<URL>]";
 810        fatal($usage) if (!defined $ta || !defined $tb);
 811        my $svn_path = '';
 812        if (!defined $url) {
 813                my $gs = eval { Git::SVN->new };
 814                if (!$gs) {
 815                        fatal("Needed URL or usable git-svn --id in ",
 816                              "the command-line\n", $usage);
 817                }
 818                $url = $gs->{url};
 819                $svn_path = $gs->{path};
 820        }
 821        unless (defined $_revision) {
 822                fatal("-r|--revision is a required argument\n", $usage);
 823        }
 824        if (defined $_message && defined $_file) {
 825                fatal("Both --message/-m and --file/-F specified ",
 826                      "for the commit message.\n",
 827                      "I have no idea what you mean");
 828        }
 829        if (defined $_file) {
 830                $_message = file_to_s($_file);
 831        } else {
 832                $_message ||= get_commit_entry($tb)->{log};
 833        }
 834        my $ra ||= Git::SVN::Ra->new($url);
 835        my $r = $_revision;
 836        if ($r eq 'HEAD') {
 837                $r = $ra->get_latest_revnum;
 838        } elsif ($r !~ /^\d+$/) {
 839                die "revision argument: $r not understood by git-svn\n";
 840        }
 841        my %ed_opts = ( r => $r,
 842                        log => $_message,
 843                        ra => $ra,
 844                        tree_a => $ta,
 845                        tree_b => $tb,
 846                        editor_cb => sub { print "Committed r$_[0]\n" },
 847                        svn_path => $svn_path );
 848        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
 849                print "No changes\n$ta == $tb\n";
 850        }
 851}
 852
 853sub escape_uri_only {
 854        my ($uri) = @_;
 855        my @tmp;
 856        foreach (split m{/}, $uri) {
 857                s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
 858                push @tmp, $_;
 859        }
 860        join('/', @tmp);
 861}
 862
 863sub escape_url {
 864        my ($url) = @_;
 865        if ($url =~ m#^([^:]+)://([^/]*)(.*)$#) {
 866                my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
 867                $url = "$scheme://$domain$uri";
 868        }
 869        $url;
 870}
 871
 872sub cmd_info {
 873        my $path = canonicalize_path(defined($_[0]) ? $_[0] : ".");
 874        my $fullpath = canonicalize_path($cmd_dir_prefix . $path);
 875        if (exists $_[1]) {
 876                die "Too many arguments specified\n";
 877        }
 878
 879        my ($file_type, $diff_status) = find_file_type_and_diff_status($path);
 880
 881        if (!$file_type && !$diff_status) {
 882                print STDERR "svn: '$path' is not under version control\n";
 883                exit 1;
 884        }
 885
 886        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 887        unless ($gs) {
 888                die "Unable to determine upstream SVN information from ",
 889                    "working tree history\n";
 890        }
 891
 892        # canonicalize_path() will return "" to make libsvn 1.5.x happy,
 893        $path = "." if $path eq "";
 894
 895        my $full_url = $url . ($fullpath eq "" ? "" : "/$fullpath");
 896
 897        if ($_url) {
 898                print escape_url($full_url), "\n";
 899                return;
 900        }
 901
 902        my $result = "Path: $path\n";
 903        $result .= "Name: " . basename($path) . "\n" if $file_type ne "dir";
 904        $result .= "URL: " . escape_url($full_url) . "\n";
 905
 906        eval {
 907                my $repos_root = $gs->repos_root;
 908                Git::SVN::remove_username($repos_root);
 909                $result .= "Repository Root: " . escape_url($repos_root) . "\n";
 910        };
 911        if ($@) {
 912                $result .= "Repository Root: (offline)\n";
 913        }
 914        $result .= "Repository UUID: $uuid\n" unless $diff_status eq "A";
 915        $result .= "Revision: " . ($diff_status eq "A" ? 0 : $rev) . "\n";
 916
 917        $result .= "Node Kind: " .
 918                   ($file_type eq "dir" ? "directory" : "file") . "\n";
 919
 920        my $schedule = $diff_status eq "A"
 921                       ? "add"
 922                       : ($diff_status eq "D" ? "delete" : "normal");
 923        $result .= "Schedule: $schedule\n";
 924
 925        if ($diff_status eq "A") {
 926                print $result, "\n";
 927                return;
 928        }
 929
 930        my ($lc_author, $lc_rev, $lc_date_utc);
 931        my @args = Git::SVN::Log::git_svn_log_cmd($rev, $rev, "--", $fullpath);
 932        my $log = command_output_pipe(@args);
 933        my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
 934        while (<$log>) {
 935                if (/^${esc_color}author (.+) <[^>]+> (\d+) ([\-\+]?\d+)$/o) {
 936                        $lc_author = $1;
 937                        $lc_date_utc = Git::SVN::Log::parse_git_date($2, $3);
 938                } elsif (/^${esc_color}    (git-svn-id:.+)$/o) {
 939                        (undef, $lc_rev, undef) = ::extract_metadata($1);
 940                }
 941        }
 942        close $log;
 943
 944        Git::SVN::Log::set_local_timezone();
 945
 946        $result .= "Last Changed Author: $lc_author\n";
 947        $result .= "Last Changed Rev: $lc_rev\n";
 948        $result .= "Last Changed Date: " .
 949                   Git::SVN::Log::format_svn_date($lc_date_utc) . "\n";
 950
 951        if ($file_type ne "dir") {
 952                my $text_last_updated_date =
 953                    ($diff_status eq "D" ? $lc_date_utc : (stat $path)[9]);
 954                $result .=
 955                    "Text Last Updated: " .
 956                    Git::SVN::Log::format_svn_date($text_last_updated_date) .
 957                    "\n";
 958                my $checksum;
 959                if ($diff_status eq "D") {
 960                        my ($fh, $ctx) =
 961                            command_output_pipe(qw(cat-file blob), "HEAD:$path");
 962                        if ($file_type eq "link") {
 963                                my $file_name = <$fh>;
 964                                $checksum = md5sum("link $file_name");
 965                        } else {
 966                                $checksum = md5sum($fh);
 967                        }
 968                        command_close_pipe($fh, $ctx);
 969                } elsif ($file_type eq "link") {
 970                        my $file_name =
 971                            command(qw(cat-file blob), "HEAD:$path");
 972                        $checksum =
 973                            md5sum("link " . $file_name);
 974                } else {
 975                        open FILE, "<", $path or die $!;
 976                        $checksum = md5sum(\*FILE);
 977                        close FILE or die $!;
 978                }
 979                $result .= "Checksum: " . $checksum . "\n";
 980        }
 981
 982        print $result, "\n";
 983}
 984
 985########################### utility functions #########################
 986
 987sub rebase_cmd {
 988        my @cmd = qw/rebase/;
 989        push @cmd, '-v' if $_verbose;
 990        push @cmd, qw/--merge/ if $_merge;
 991        push @cmd, "--strategy=$_strategy" if $_strategy;
 992        @cmd;
 993}
 994
 995sub post_fetch_checkout {
 996        return if $_no_checkout;
 997        my $gs = $Git::SVN::_head or return;
 998        return if verify_ref('refs/heads/master^0');
 999
1000        my $valid_head = verify_ref('HEAD^0');
1001        command_noisy(qw(update-ref refs/heads/master), $gs->refname);
1002        return if ($valid_head || !verify_ref('HEAD^0'));
1003
1004        return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
1005        my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
1006        return if -f $index;
1007
1008        return if command_oneline(qw/rev-parse --is-inside-work-tree/) eq 'false';
1009        return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
1010        command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
1011        print STDERR "Checked out HEAD:\n  ",
1012                     $gs->full_url, " r", $gs->last_rev, "\n";
1013}
1014
1015sub complete_svn_url {
1016        my ($url, $path) = @_;
1017        $path =~ s#/+$##;
1018        if ($path !~ m#^[a-z\+]+://#) {
1019                if (!defined $url || $url !~ m#^[a-z\+]+://#) {
1020                        fatal("E: '$path' is not a complete URL ",
1021                              "and a separate URL is not specified");
1022                }
1023                return ($url, $path);
1024        }
1025        return ($path, '');
1026}
1027
1028sub complete_url_ls_init {
1029        my ($ra, $repo_path, $switch, $pfx) = @_;
1030        unless ($repo_path) {
1031                print STDERR "W: $switch not specified\n";
1032                return;
1033        }
1034        $repo_path =~ s#/+$##;
1035        if ($repo_path =~ m#^[a-z\+]+://#) {
1036                $ra = Git::SVN::Ra->new($repo_path);
1037                $repo_path = '';
1038        } else {
1039                $repo_path =~ s#^/+##;
1040                unless ($ra) {
1041                        fatal("E: '$repo_path' is not a complete URL ",
1042                              "and a separate URL is not specified");
1043                }
1044        }
1045        my $url = $ra->{url};
1046        my $gs = Git::SVN->init($url, undef, undef, undef, 1);
1047        my $k = "svn-remote.$gs->{repo_id}.url";
1048        my $orig_url = eval { command_oneline(qw/config --get/, $k) };
1049        if ($orig_url && ($orig_url ne $gs->{url})) {
1050                die "$k already set: $orig_url\n",
1051                    "wanted to set to: $gs->{url}\n";
1052        }
1053        command_oneline('config', $k, $gs->{url}) unless $orig_url;
1054        my $remote_path = "$ra->{svn_path}/$repo_path";
1055        $remote_path =~ s#/+#/#g;
1056        $remote_path =~ s#^/##g;
1057        $remote_path .= "/*" if $remote_path !~ /\*/;
1058        my ($n) = ($switch =~ /^--(\w+)/);
1059        if (length $pfx && $pfx !~ m#/$#) {
1060                die "--prefix='$pfx' must have a trailing slash '/'\n";
1061        }
1062        command_noisy('config',
1063                      "svn-remote.$gs->{repo_id}.$n",
1064                      "$remote_path:refs/remotes/$pfx*" .
1065                        ('/*' x (($remote_path =~ tr/*/*/) - 1)) );
1066}
1067
1068sub verify_ref {
1069        my ($ref) = @_;
1070        eval { command_oneline([ 'rev-parse', '--verify', $ref ],
1071                               { STDERR => 0 }); };
1072}
1073
1074sub get_tree_from_treeish {
1075        my ($treeish) = @_;
1076        # $treeish can be a symbolic ref, too:
1077        my $type = command_oneline(qw/cat-file -t/, $treeish);
1078        my $expected;
1079        while ($type eq 'tag') {
1080                ($treeish, $type) = command(qw/cat-file tag/, $treeish);
1081        }
1082        if ($type eq 'commit') {
1083                $expected = (grep /^tree /, command(qw/cat-file commit/,
1084                                                    $treeish))[0];
1085                ($expected) = ($expected =~ /^tree ($sha1)$/o);
1086                die "Unable to get tree from $treeish\n" unless $expected;
1087        } elsif ($type eq 'tree') {
1088                $expected = $treeish;
1089        } else {
1090                die "$treeish is a $type, expected tree, tag or commit\n";
1091        }
1092        return $expected;
1093}
1094
1095sub get_commit_entry {
1096        my ($treeish) = shift;
1097        my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
1098        my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
1099        my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
1100        open my $log_fh, '>', $commit_editmsg or croak $!;
1101
1102        my $type = command_oneline(qw/cat-file -t/, $treeish);
1103        if ($type eq 'commit' || $type eq 'tag') {
1104                my ($msg_fh, $ctx) = command_output_pipe('cat-file',
1105                                                         $type, $treeish);
1106                my $in_msg = 0;
1107                my $author;
1108                my $saw_from = 0;
1109                my $msgbuf = "";
1110                while (<$msg_fh>) {
1111                        if (!$in_msg) {
1112                                $in_msg = 1 if (/^\s*$/);
1113                                $author = $1 if (/^author (.*>)/);
1114                        } elsif (/^git-svn-id: /) {
1115                                # skip this for now, we regenerate the
1116                                # correct one on re-fetch anyways
1117                                # TODO: set *:merge properties or like...
1118                        } else {
1119                                if (/^From:/ || /^Signed-off-by:/) {
1120                                        $saw_from = 1;
1121                                }
1122                                $msgbuf .= $_;
1123                        }
1124                }
1125                $msgbuf =~ s/\s+$//s;
1126                if ($Git::SVN::_add_author_from && defined($author)
1127                    && !$saw_from) {
1128                        $msgbuf .= "\n\nFrom: $author";
1129                }
1130                print $log_fh $msgbuf or croak $!;
1131                command_close_pipe($msg_fh, $ctx);
1132        }
1133        close $log_fh or croak $!;
1134
1135        if ($_edit || ($type eq 'tree')) {
1136                my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
1137                # TODO: strip out spaces, comments, like git-commit.sh
1138                system($editor, $commit_editmsg);
1139        }
1140        rename $commit_editmsg, $commit_msg or croak $!;
1141        {
1142                # SVN requires messages to be UTF-8 when entering the repo
1143                local $/;
1144                open $log_fh, '<', $commit_msg or croak $!;
1145                binmode $log_fh;
1146                chomp($log_entry{log} = <$log_fh>);
1147
1148                if (my $enc = Git::config('i18n.commitencoding')) {
1149                        require Encode;
1150                        Encode::from_to($log_entry{log}, $enc, 'UTF-8');
1151                }
1152                close $log_fh or croak $!;
1153        }
1154        unlink $commit_msg;
1155        \%log_entry;
1156}
1157
1158sub s_to_file {
1159        my ($str, $file, $mode) = @_;
1160        open my $fd,'>',$file or croak $!;
1161        print $fd $str,"\n" or croak $!;
1162        close $fd or croak $!;
1163        chmod ($mode &~ umask, $file) if (defined $mode);
1164}
1165
1166sub file_to_s {
1167        my $file = shift;
1168        open my $fd,'<',$file or croak "$!: file: $file\n";
1169        local $/;
1170        my $ret = <$fd>;
1171        close $fd or croak $!;
1172        $ret =~ s/\s*$//s;
1173        return $ret;
1174}
1175
1176# '<svn username> = real-name <email address>' mapping based on git-svnimport:
1177sub load_authors {
1178        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1179        my $log = $cmd eq 'log';
1180        while (<$authors>) {
1181                chomp;
1182                next unless /^(.+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
1183                my ($user, $name, $email) = ($1, $2, $3);
1184                if ($log) {
1185                        $Git::SVN::Log::rusers{"$name <$email>"} = $user;
1186                } else {
1187                        $users{$user} = [$name, $email];
1188                }
1189        }
1190        close $authors or croak $!;
1191}
1192
1193# convert GetOpt::Long specs for use by git-config
1194sub read_repo_config {
1195        return unless -d $ENV{GIT_DIR};
1196        my $opts = shift;
1197        my @config_only;
1198        foreach my $o (keys %$opts) {
1199                # if we have mixedCase and a long option-only, then
1200                # it's a config-only variable that we don't need for
1201                # the command-line.
1202                push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
1203                my $v = $opts->{$o};
1204                my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
1205                $key =~ s/-//g;
1206                my $arg = 'git config';
1207                $arg .= ' --int' if ($o =~ /[:=]i$/);
1208                $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
1209                if (ref $v eq 'ARRAY') {
1210                        chomp(my @tmp = `$arg --get-all svn.$key`);
1211                        @$v = @tmp if @tmp;
1212                } else {
1213                        chomp(my $tmp = `$arg --get svn.$key`);
1214                        if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
1215                                $$v = $tmp;
1216                        }
1217                }
1218        }
1219        delete @$opts{@config_only} if @config_only;
1220}
1221
1222sub extract_metadata {
1223        my $id = shift or return (undef, undef, undef);
1224        my ($url, $rev, $uuid) = ($id =~ /^\s*git-svn-id:\s+(.*)\@(\d+)
1225                                                        \s([a-f\d\-]+)$/x);
1226        if (!defined $rev || !$uuid || !$url) {
1227                # some of the original repositories I made had
1228                # identifiers like this:
1229                ($rev, $uuid) = ($id =~/^\s*git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
1230        }
1231        return ($url, $rev, $uuid);
1232}
1233
1234sub cmt_metadata {
1235        return extract_metadata((grep(/^git-svn-id: /,
1236                command(qw/cat-file commit/, shift)))[-1]);
1237}
1238
1239sub working_head_info {
1240        my ($head, $refs) = @_;
1241        my @args = ('log', '--no-color', '--first-parent', '--pretty=medium');
1242        my ($fh, $ctx) = command_output_pipe(@args, $head);
1243        my $hash;
1244        my %max;
1245        while (<$fh>) {
1246                if ( m{^commit ($::sha1)$} ) {
1247                        unshift @$refs, $hash if $hash and $refs;
1248                        $hash = $1;
1249                        next;
1250                }
1251                next unless s{^\s*(git-svn-id:)}{$1};
1252                my ($url, $rev, $uuid) = extract_metadata($_);
1253                if (defined $url && defined $rev) {
1254                        next if $max{$url} and $max{$url} < $rev;
1255                        if (my $gs = Git::SVN->find_by_url($url)) {
1256                                my $c = $gs->rev_map_get($rev, $uuid);
1257                                if ($c && $c eq $hash) {
1258                                        close $fh; # break the pipe
1259                                        return ($url, $rev, $uuid, $gs);
1260                                } else {
1261                                        $max{$url} ||= $gs->rev_map_max;
1262                                }
1263                        }
1264                }
1265        }
1266        command_close_pipe($fh, $ctx);
1267        (undef, undef, undef, undef);
1268}
1269
1270sub read_commit_parents {
1271        my ($parents, $c) = @_;
1272        chomp(my $p = command_oneline(qw/rev-list --parents -1/, $c));
1273        $p =~ s/^($c)\s*// or die "rev-list --parents -1 $c failed!\n";
1274        @{$parents->{$c}} = split(/ /, $p);
1275}
1276
1277sub linearize_history {
1278        my ($gs, $refs) = @_;
1279        my %parents;
1280        foreach my $c (@$refs) {
1281                read_commit_parents(\%parents, $c);
1282        }
1283
1284        my @linear_refs;
1285        my %skip = ();
1286        my $last_svn_commit = $gs->last_commit;
1287        foreach my $c (reverse @$refs) {
1288                next if $c eq $last_svn_commit;
1289                last if $skip{$c};
1290
1291                unshift @linear_refs, $c;
1292                $skip{$c} = 1;
1293
1294                # we only want the first parent to diff against for linear
1295                # history, we save the rest to inject when we finalize the
1296                # svn commit
1297                my $fp_a = verify_ref("$c~1");
1298                my $fp_b = shift @{$parents{$c}} if $parents{$c};
1299                if (!$fp_a || !$fp_b) {
1300                        die "Commit $c\n",
1301                            "has no parent commit, and therefore ",
1302                            "nothing to diff against.\n",
1303                            "You should be working from a repository ",
1304                            "originally created by git-svn\n";
1305                }
1306                if ($fp_a ne $fp_b) {
1307                        die "$c~1 = $fp_a, however parsing commit $c ",
1308                            "revealed that:\n$c~1 = $fp_b\nBUG!\n";
1309                }
1310
1311                foreach my $p (@{$parents{$c}}) {
1312                        $skip{$p} = 1;
1313                }
1314        }
1315        (\@linear_refs, \%parents);
1316}
1317
1318sub find_file_type_and_diff_status {
1319        my ($path) = @_;
1320        return ('dir', '') if $path eq '';
1321
1322        my $diff_output =
1323            command_oneline(qw(diff --cached --name-status --), $path) || "";
1324        my $diff_status = (split(' ', $diff_output))[0] || "";
1325
1326        my $ls_tree = command_oneline(qw(ls-tree HEAD), $path) || "";
1327
1328        return (undef, undef) if !$diff_status && !$ls_tree;
1329
1330        if ($diff_status eq "A") {
1331                return ("link", $diff_status) if -l $path;
1332                return ("dir", $diff_status) if -d $path;
1333                return ("file", $diff_status);
1334        }
1335
1336        my $mode = (split(' ', $ls_tree))[0] || "";
1337
1338        return ("link", $diff_status) if $mode eq "120000";
1339        return ("dir", $diff_status) if $mode eq "040000";
1340        return ("file", $diff_status);
1341}
1342
1343sub md5sum {
1344        my $arg = shift;
1345        my $ref = ref $arg;
1346        my $md5 = Digest::MD5->new();
1347        if ($ref eq 'GLOB' || $ref eq 'IO::File' || $ref eq 'File::Temp') {
1348                $md5->addfile($arg) or croak $!;
1349        } elsif ($ref eq 'SCALAR') {
1350                $md5->add($$arg) or croak $!;
1351        } elsif (!$ref) {
1352                $md5->add($arg) or croak $!;
1353        } else {
1354                ::fatal "Can't provide MD5 hash for unknown ref type: '", $ref, "'";
1355        }
1356        return $md5->hexdigest();
1357}
1358
1359package Git::SVN;
1360use strict;
1361use warnings;
1362use Fcntl qw/:DEFAULT :seek/;
1363use constant rev_map_fmt => 'NH40';
1364use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
1365            $_repack $_repack_flags $_use_svm_props $_head
1366            $_use_svnsync_props $no_reuse_existing $_minimize_url
1367            $_use_log_author $_add_author_from/;
1368use Carp qw/croak/;
1369use File::Path qw/mkpath/;
1370use File::Copy qw/copy/;
1371use IPC::Open3;
1372
1373my ($_gc_nr, $_gc_period);
1374
1375# properties that we do not log:
1376my %SKIP_PROP;
1377BEGIN {
1378        %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
1379                                        svn:special svn:executable
1380                                        svn:entry:committed-rev
1381                                        svn:entry:last-author
1382                                        svn:entry:uuid
1383                                        svn:entry:committed-date/;
1384
1385        # some options are read globally, but can be overridden locally
1386        # per [svn-remote "..."] section.  Command-line options will *NOT*
1387        # override options set in an [svn-remote "..."] section
1388        no strict 'refs';
1389        for my $option (qw/follow_parent no_metadata use_svm_props
1390                           use_svnsync_props/) {
1391                my $key = $option;
1392                $key =~ tr/_//d;
1393                my $prop = "-$option";
1394                *$option = sub {
1395                        my ($self) = @_;
1396                        return $self->{$prop} if exists $self->{$prop};
1397                        my $k = "svn-remote.$self->{repo_id}.$key";
1398                        eval { command_oneline(qw/config --get/, $k) };
1399                        if ($@) {
1400                                $self->{$prop} = ${"Git::SVN::_$option"};
1401                        } else {
1402                                my $v = command_oneline(qw/config --bool/,$k);
1403                                $self->{$prop} = $v eq 'false' ? 0 : 1;
1404                        }
1405                        return $self->{$prop};
1406                }
1407        }
1408}
1409
1410
1411my (%LOCKFILES, %INDEX_FILES);
1412END {
1413        unlink keys %LOCKFILES if %LOCKFILES;
1414        unlink keys %INDEX_FILES if %INDEX_FILES;
1415}
1416
1417sub resolve_local_globs {
1418        my ($url, $fetch, $glob_spec) = @_;
1419        return unless defined $glob_spec;
1420        my $ref = $glob_spec->{ref};
1421        my $path = $glob_spec->{path};
1422        foreach (command(qw#for-each-ref --format=%(refname) refs/remotes#)) {
1423                next unless m#^refs/remotes/$ref->{regex}$#;
1424                my $p = $1;
1425                my $pathname = desanitize_refname($path->full_path($p));
1426                my $refname = desanitize_refname($ref->full_path($p));
1427                if (my $existing = $fetch->{$pathname}) {
1428                        if ($existing ne $refname) {
1429                                die "Refspec conflict:\n",
1430                                    "existing: refs/remotes/$existing\n",
1431                                    " globbed: refs/remotes/$refname\n";
1432                        }
1433                        my $u = (::cmt_metadata("refs/remotes/$refname"))[0];
1434                        $u =~ s!^\Q$url\E(/|$)!! or die
1435                          "refs/remotes/$refname: '$url' not found in '$u'\n";
1436                        if ($pathname ne $u) {
1437                                warn "W: Refspec glob conflict ",
1438                                     "(ref: refs/remotes/$refname):\n",
1439                                     "expected path: $pathname\n",
1440                                     "    real path: $u\n",
1441                                     "Continuing ahead with $u\n";
1442                                next;
1443                        }
1444                } else {
1445                        $fetch->{$pathname} = $refname;
1446                }
1447        }
1448}
1449
1450sub parse_revision_argument {
1451        my ($base, $head) = @_;
1452        if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
1453                return ($base, $head);
1454        }
1455        return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
1456        return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
1457        return ($head, $head) if ($::_revision eq 'HEAD');
1458        return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
1459        return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
1460        die "revision argument: $::_revision not understood by git-svn\n";
1461}
1462
1463sub fetch_all {
1464        my ($repo_id, $remotes) = @_;
1465        if (ref $repo_id) {
1466                my $gs = $repo_id;
1467                $repo_id = undef;
1468                $repo_id = $gs->{repo_id};
1469        }
1470        $remotes ||= read_all_remotes();
1471        my $remote = $remotes->{$repo_id} or
1472                     die "[svn-remote \"$repo_id\"] unknown\n";
1473        my $fetch = $remote->{fetch};
1474        my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
1475        my (@gs, @globs);
1476        my $ra = Git::SVN::Ra->new($url);
1477        my $uuid = $ra->get_uuid;
1478        my $head = $ra->get_latest_revnum;
1479        my $base = defined $fetch ? $head : 0;
1480
1481        # read the max revs for wildcard expansion (branches/*, tags/*)
1482        foreach my $t (qw/branches tags/) {
1483                defined $remote->{$t} or next;
1484                push @globs, $remote->{$t};
1485                my $max_rev = eval { tmp_config(qw/--int --get/,
1486                                         "svn-remote.$repo_id.${t}-maxRev") };
1487                if (defined $max_rev && ($max_rev < $base)) {
1488                        $base = $max_rev;
1489                } elsif (!defined $max_rev) {
1490                        $base = 0;
1491                }
1492        }
1493
1494        if ($fetch) {
1495                foreach my $p (sort keys %$fetch) {
1496                        my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
1497                        my $lr = $gs->rev_map_max;
1498                        if (defined $lr) {
1499                                $base = $lr if ($lr < $base);
1500                        }
1501                        push @gs, $gs;
1502                }
1503        }
1504
1505        ($base, $head) = parse_revision_argument($base, $head);
1506        $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
1507}
1508
1509sub read_all_remotes {
1510        my $r = {};
1511        my $use_svm_props = eval { command_oneline(qw/config --bool
1512            svn.useSvmProps/) };
1513        $use_svm_props = $use_svm_props eq 'true' if $use_svm_props;
1514        foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
1515                if (m!^(.+)\.fetch=\s*(.*)\s*:\s*(.+)\s*$!) {
1516                        my ($remote, $local_ref, $_remote_ref) = ($1, $2, $3);
1517                        die("svn-remote.$remote: remote ref '$_remote_ref' "
1518                            . "must start with 'refs/remotes/'\n")
1519                                unless $_remote_ref =~ m{^refs/remotes/(.+)};
1520                        my $remote_ref = $1;
1521                        $local_ref =~ s{^/}{};
1522                        $r->{$remote}->{fetch}->{$local_ref} = $remote_ref;
1523                        $r->{$remote}->{svm} = {} if $use_svm_props;
1524                } elsif (m!^(.+)\.usesvmprops=\s*(.*)\s*$!) {
1525                        $r->{$1}->{svm} = {};
1526                } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
1527                        $r->{$1}->{url} = $2;
1528                } elsif (m!^(.+)\.(branches|tags)=
1529                           (.*):refs/remotes/(.+)\s*$/!x) {
1530                        my ($p, $g) = ($3, $4);
1531                        my $rs = $r->{$1}->{$2} = {
1532                                          t => $2,
1533                                          remote => $1,
1534                                          path => Git::SVN::GlobSpec->new($p),
1535                                          ref => Git::SVN::GlobSpec->new($g) };
1536                        if (length($rs->{ref}->{right}) != 0) {
1537                                die "The '*' glob character must be the last ",
1538                                    "character of '$g'\n";
1539                        }
1540                }
1541        }
1542
1543        map {
1544                if (defined $r->{$_}->{svm}) {
1545                        my $svm;
1546                        eval {
1547                                my $section = "svn-remote.$_";
1548                                $svm = {
1549                                        source => tmp_config('--get',
1550                                            "$section.svm-source"),
1551                                        replace => tmp_config('--get',
1552                                            "$section.svm-replace"),
1553                                }
1554                        };
1555                        $r->{$_}->{svm} = $svm;
1556                }
1557        } keys %$r;
1558
1559        $r;
1560}
1561
1562sub init_vars {
1563        $_gc_nr = $_gc_period = 1000;
1564        if (defined $_repack || defined $_repack_flags) {
1565               warn "Repack options are obsolete; they have no effect.\n";
1566        }
1567}
1568
1569sub verify_remotes_sanity {
1570        return unless -d $ENV{GIT_DIR};
1571        my %seen;
1572        foreach (command(qw/config -l/)) {
1573                if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
1574                        if ($seen{$1}) {
1575                                die "Remote ref refs/remote/$1 is tracked by",
1576                                    "\n  \"$_\"\nand\n  \"$seen{$1}\"\n",
1577                                    "Please resolve this ambiguity in ",
1578                                    "your git configuration file before ",
1579                                    "continuing\n";
1580                        }
1581                        $seen{$1} = $_;
1582                }
1583        }
1584}
1585
1586sub find_existing_remote {
1587        my ($url, $remotes) = @_;
1588        return undef if $no_reuse_existing;
1589        my $existing;
1590        foreach my $repo_id (keys %$remotes) {
1591                my $u = $remotes->{$repo_id}->{url} or next;
1592                next if $u ne $url;
1593                $existing = $repo_id;
1594                last;
1595        }
1596        $existing;
1597}
1598
1599sub init_remote_config {
1600        my ($self, $url, $no_write) = @_;
1601        $url =~ s!/+$!!; # strip trailing slash
1602        my $r = read_all_remotes();
1603        my $existing = find_existing_remote($url, $r);
1604        if ($existing) {
1605                unless ($no_write) {
1606                        print STDERR "Using existing ",
1607                                     "[svn-remote \"$existing\"]\n";
1608                }
1609                $self->{repo_id} = $existing;
1610        } elsif ($_minimize_url) {
1611                my $min_url = Git::SVN::Ra->new($url)->minimize_url;
1612                $existing = find_existing_remote($min_url, $r);
1613                if ($existing) {
1614                        unless ($no_write) {
1615                                print STDERR "Using existing ",
1616                                             "[svn-remote \"$existing\"]\n";
1617                        }
1618                        $self->{repo_id} = $existing;
1619                }
1620                if ($min_url ne $url) {
1621                        unless ($no_write) {
1622                                print STDERR "Using higher level of URL: ",
1623                                             "$url => $min_url\n";
1624                        }
1625                        my $old_path = $self->{path};
1626                        $self->{path} = $url;
1627                        $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
1628                        if (length $old_path) {
1629                                $self->{path} .= "/$old_path";
1630                        }
1631                        $url = $min_url;
1632                }
1633        }
1634        my $orig_url;
1635        if (!$existing) {
1636                # verify that we aren't overwriting anything:
1637                $orig_url = eval {
1638                        command_oneline('config', '--get',
1639                                        "svn-remote.$self->{repo_id}.url")
1640                };
1641                if ($orig_url && ($orig_url ne $url)) {
1642                        die "svn-remote.$self->{repo_id}.url already set: ",
1643                            "$orig_url\nwanted to set to: $url\n";
1644                }
1645        }
1646        my ($xrepo_id, $xpath) = find_ref($self->refname);
1647        if (defined $xpath) {
1648                die "svn-remote.$xrepo_id.fetch already set to track ",
1649                    "$xpath:refs/remotes/", $self->refname, "\n";
1650        }
1651        unless ($no_write) {
1652                command_noisy('config',
1653                              "svn-remote.$self->{repo_id}.url", $url);
1654                $self->{path} =~ s{^/}{};
1655                command_noisy('config', '--add',
1656                              "svn-remote.$self->{repo_id}.fetch",
1657                              "$self->{path}:".$self->refname);
1658        }
1659        $self->{url} = $url;
1660}
1661
1662sub find_by_url { # repos_root and, path are optional
1663        my ($class, $full_url, $repos_root, $path) = @_;
1664
1665        return undef unless defined $full_url;
1666        remove_username($full_url);
1667        remove_username($repos_root) if defined $repos_root;
1668        my $remotes = read_all_remotes();
1669        if (defined $full_url && defined $repos_root && !defined $path) {
1670                $path = $full_url;
1671                $path =~ s#^\Q$repos_root\E(?:/|$)##;
1672        }
1673        foreach my $repo_id (keys %$remotes) {
1674                my $u = $remotes->{$repo_id}->{url} or next;
1675                remove_username($u);
1676                next if defined $repos_root && $repos_root ne $u;
1677
1678                my $fetch = $remotes->{$repo_id}->{fetch} || {};
1679                foreach (qw/branches tags/) {
1680                        resolve_local_globs($u, $fetch,
1681                                            $remotes->{$repo_id}->{$_});
1682                }
1683                my $p = $path;
1684                my $rwr = rewrite_root({repo_id => $repo_id});
1685                my $svm = $remotes->{$repo_id}->{svm}
1686                        if defined $remotes->{$repo_id}->{svm};
1687                unless (defined $p) {
1688                        $p = $full_url;
1689                        my $z = $u;
1690                        my $prefix = '';
1691                        if ($rwr) {
1692                                $z = $rwr;
1693                        } elsif (defined $svm) {
1694                                $z = $svm->{source};
1695                                $prefix = $svm->{replace};
1696                                $prefix =~ s#^\Q$u\E(?:/|$)##;
1697                                $prefix =~ s#/$##;
1698                        }
1699                        $p =~ s#^\Q$z\E(?:/|$)#$prefix# or next;
1700                }
1701                foreach my $f (keys %$fetch) {
1702                        next if $f ne $p;
1703                        return Git::SVN->new($fetch->{$f}, $repo_id, $f);
1704                }
1705        }
1706        undef;
1707}
1708
1709sub init {
1710        my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
1711        my $self = _new($class, $repo_id, $ref_id, $path);
1712        if (defined $url) {
1713                $self->init_remote_config($url, $no_write);
1714        }
1715        $self;
1716}
1717
1718sub find_ref {
1719        my ($ref_id) = @_;
1720        foreach (command(qw/config -l/)) {
1721                next unless m!^svn-remote\.(.+)\.fetch=
1722                              \s*(.*)\s*:\s*refs/remotes/(.+)\s*$!x;
1723                my ($repo_id, $path, $ref) = ($1, $2, $3);
1724                if ($ref eq $ref_id) {
1725                        $path = '' if ($path =~ m#^\./?#);
1726                        return ($repo_id, $path);
1727                }
1728        }
1729        (undef, undef, undef);
1730}
1731
1732sub new {
1733        my ($class, $ref_id, $repo_id, $path) = @_;
1734        if (defined $ref_id && !defined $repo_id && !defined $path) {
1735                ($repo_id, $path) = find_ref($ref_id);
1736                if (!defined $repo_id) {
1737                        die "Could not find a \"svn-remote.*.fetch\" key ",
1738                            "in the repository configuration matching: ",
1739                            "refs/remotes/$ref_id\n";
1740                }
1741        }
1742        my $self = _new($class, $repo_id, $ref_id, $path);
1743        if (!defined $self->{path} || !length $self->{path}) {
1744                my $fetch = command_oneline('config', '--get',
1745                                            "svn-remote.$repo_id.fetch",
1746                                            ":refs/remotes/$ref_id\$") or
1747                     die "Failed to read \"svn-remote.$repo_id.fetch\" ",
1748                         "\":refs/remotes/$ref_id\$\" in config\n";
1749                ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
1750        }
1751        $self->{url} = command_oneline('config', '--get',
1752                                       "svn-remote.$repo_id.url") or
1753                  die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
1754        $self->rebuild;
1755        $self;
1756}
1757
1758sub refname {
1759        my ($refname) = "refs/remotes/$_[0]->{ref_id}" ;
1760
1761        # It cannot end with a slash /, we'll throw up on this because
1762        # SVN can't have directories with a slash in their name, either:
1763        if ($refname =~ m{/$}) {
1764                die "ref: '$refname' ends with a trailing slash, this is ",
1765                    "not permitted by git nor Subversion\n";
1766        }
1767
1768        # It cannot have ASCII control character space, tilde ~, caret ^,
1769        # colon :, question-mark ?, asterisk *, space, or open bracket [
1770        # anywhere.
1771        #
1772        # Additionally, % must be escaped because it is used for escaping
1773        # and we want our escaped refname to be reversible
1774        $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
1775
1776        # no slash-separated component can begin with a dot .
1777        # /.* becomes /%2E*
1778        $refname =~ s{/\.}{/%2E}g;
1779
1780        # It cannot have two consecutive dots .. anywhere
1781        # .. becomes %2E%2E
1782        $refname =~ s{\.\.}{%2E%2E}g;
1783
1784        return $refname;
1785}
1786
1787sub desanitize_refname {
1788        my ($refname) = @_;
1789        $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
1790        return $refname;
1791}
1792
1793sub svm_uuid {
1794        my ($self) = @_;
1795        return $self->{svm}->{uuid} if $self->svm;
1796        $self->ra;
1797        unless ($self->{svm}) {
1798                die "SVM UUID not cached, and reading remotely failed\n";
1799        }
1800        $self->{svm}->{uuid};
1801}
1802
1803sub svm {
1804        my ($self) = @_;
1805        return $self->{svm} if $self->{svm};
1806        my $svm;
1807        # see if we have it in our config, first:
1808        eval {
1809                my $section = "svn-remote.$self->{repo_id}";
1810                $svm = {
1811                  source => tmp_config('--get', "$section.svm-source"),
1812                  uuid => tmp_config('--get', "$section.svm-uuid"),
1813                  replace => tmp_config('--get', "$section.svm-replace"),
1814                }
1815        };
1816        if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
1817                $self->{svm} = $svm;
1818        }
1819        $self->{svm};
1820}
1821
1822sub _set_svm_vars {
1823        my ($self, $ra) = @_;
1824        return $ra if $self->svm;
1825
1826        my @err = ( "useSvmProps set, but failed to read SVM properties\n",
1827                    "(svm:source, svm:uuid) ",
1828                    "from the following URLs:\n" );
1829        sub read_svm_props {
1830                my ($self, $ra, $path, $r) = @_;
1831                my $props = ($ra->get_dir($path, $r))[2];
1832                my $src = $props->{'svm:source'};
1833                my $uuid = $props->{'svm:uuid'};
1834                return undef if (!$src || !$uuid);
1835
1836                chomp($src, $uuid);
1837
1838                $uuid =~ m{^[0-9a-f\-]{30,}$}
1839                    or die "doesn't look right - svm:uuid is '$uuid'\n";
1840
1841                # the '!' is used to mark the repos_root!/relative/path
1842                $src =~ s{/?!/?}{/};
1843                $src =~ s{/+$}{}; # no trailing slashes please
1844                # username is of no interest
1845                $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
1846
1847                my $replace = $ra->{url};
1848                $replace .= "/$path" if length $path;
1849
1850                my $section = "svn-remote.$self->{repo_id}";
1851                tmp_config("$section.svm-source", $src);
1852                tmp_config("$section.svm-replace", $replace);
1853                tmp_config("$section.svm-uuid", $uuid);
1854                $self->{svm} = {
1855                        source => $src,
1856                        uuid => $uuid,
1857                        replace => $replace
1858                };
1859        }
1860
1861        my $r = $ra->get_latest_revnum;
1862        my $path = $self->{path};
1863        my %tried;
1864        while (length $path) {
1865                unless ($tried{"$self->{url}/$path"}) {
1866                        return $ra if $self->read_svm_props($ra, $path, $r);
1867                        $tried{"$self->{url}/$path"} = 1;
1868                }
1869                $path =~ s#/?[^/]+$##;
1870        }
1871        die "Path: '$path' should be ''\n" if $path ne '';
1872        return $ra if $self->read_svm_props($ra, $path, $r);
1873        $tried{"$self->{url}/$path"} = 1;
1874
1875        if ($ra->{repos_root} eq $self->{url}) {
1876                die @err, (map { "  $_\n" } keys %tried), "\n";
1877        }
1878
1879        # nope, make sure we're connected to the repository root:
1880        my $ok;
1881        my @tried_b;
1882        $path = $ra->{svn_path};
1883        $ra = Git::SVN::Ra->new($ra->{repos_root});
1884        while (length $path) {
1885                unless ($tried{"$ra->{url}/$path"}) {
1886                        $ok = $self->read_svm_props($ra, $path, $r);
1887                        last if $ok;
1888                        $tried{"$ra->{url}/$path"} = 1;
1889                }
1890                $path =~ s#/?[^/]+$##;
1891        }
1892        die "Path: '$path' should be ''\n" if $path ne '';
1893        $ok ||= $self->read_svm_props($ra, $path, $r);
1894        $tried{"$ra->{url}/$path"} = 1;
1895        if (!$ok) {
1896                die @err, (map { "  $_\n" } keys %tried), "\n";
1897        }
1898        Git::SVN::Ra->new($self->{url});
1899}
1900
1901sub svnsync {
1902        my ($self) = @_;
1903        return $self->{svnsync} if $self->{svnsync};
1904
1905        if ($self->no_metadata) {
1906                die "Can't have both 'noMetadata' and ",
1907                    "'useSvnsyncProps' options set!\n";
1908        }
1909        if ($self->rewrite_root) {
1910                die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
1911                    "options set!\n";
1912        }
1913
1914        my $svnsync;
1915        # see if we have it in our config, first:
1916        eval {
1917                my $section = "svn-remote.$self->{repo_id}";
1918
1919                my $url = tmp_config('--get', "$section.svnsync-url");
1920                ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
1921                   die "doesn't look right - svn:sync-from-url is '$url'\n";
1922
1923                my $uuid = tmp_config('--get', "$section.svnsync-uuid");
1924                ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}) or
1925                   die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
1926
1927                $svnsync = { url => $url, uuid => $uuid }
1928        };
1929        if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
1930                return $self->{svnsync} = $svnsync;
1931        }
1932
1933        my $err = "useSvnsyncProps set, but failed to read " .
1934                  "svnsync property: svn:sync-from-";
1935        my $rp = $self->ra->rev_proplist(0);
1936
1937        my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
1938        ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
1939                   die "doesn't look right - svn:sync-from-url is '$url'\n";
1940
1941        my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
1942        ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}) or
1943                   die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
1944
1945        my $section = "svn-remote.$self->{repo_id}";
1946        tmp_config('--add', "$section.svnsync-uuid", $uuid);
1947        tmp_config('--add', "$section.svnsync-url", $url);
1948        return $self->{svnsync} = { url => $url, uuid => $uuid };
1949}
1950
1951# this allows us to memoize our SVN::Ra UUID locally and avoid a
1952# remote lookup (useful for 'git svn log').
1953sub ra_uuid {
1954        my ($self) = @_;
1955        unless ($self->{ra_uuid}) {
1956                my $key = "svn-remote.$self->{repo_id}.uuid";
1957                my $uuid = eval { tmp_config('--get', $key) };
1958                if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/) {
1959                        $self->{ra_uuid} = $uuid;
1960                } else {
1961                        die "ra_uuid called without URL\n" unless $self->{url};
1962                        $self->{ra_uuid} = $self->ra->get_uuid;
1963                        tmp_config('--add', $key, $self->{ra_uuid});
1964                }
1965        }
1966        $self->{ra_uuid};
1967}
1968
1969sub _set_repos_root {
1970        my ($self, $repos_root) = @_;
1971        my $k = "svn-remote.$self->{repo_id}.reposRoot";
1972        $repos_root ||= $self->ra->{repos_root};
1973        tmp_config($k, $repos_root);
1974        $repos_root;
1975}
1976
1977sub repos_root {
1978        my ($self) = @_;
1979        my $k = "svn-remote.$self->{repo_id}.reposRoot";
1980        eval { tmp_config('--get', $k) } || $self->_set_repos_root;
1981}
1982
1983sub ra {
1984        my ($self) = shift;
1985        my $ra = Git::SVN::Ra->new($self->{url});
1986        $self->_set_repos_root($ra->{repos_root});
1987        if ($self->use_svm_props && !$self->{svm}) {
1988                if ($self->no_metadata) {
1989                        die "Can't have both 'noMetadata' and ",
1990                            "'useSvmProps' options set!\n";
1991                } elsif ($self->use_svnsync_props) {
1992                        die "Can't have both 'useSvnsyncProps' and ",
1993                            "'useSvmProps' options set!\n";
1994                }
1995                $ra = $self->_set_svm_vars($ra);
1996                $self->{-want_revprops} = 1;
1997        }
1998        $ra;
1999}
2000
2001sub rel_path {
2002        my ($self) = @_;
2003        my $repos_root = $self->ra->{repos_root};
2004        return $self->{path} if ($self->{url} eq $repos_root);
2005        my $url = $self->{url} .
2006                  (length $self->{path} ? "/$self->{path}" : $self->{path});
2007        $url =~ s!^\Q$repos_root\E(?:/+|$)!!g;
2008        $url;
2009}
2010
2011# prop_walk(PATH, REV, SUB)
2012# -------------------------
2013# Recursively traverse PATH at revision REV and invoke SUB for each
2014# directory that contains a SVN property.  SUB will be invoked as
2015# follows:  &SUB(gs, path, props);  where `gs' is this instance of
2016# Git::SVN, `path' the path to the directory where the properties
2017# `props' were found.  The `path' will be relative to point of checkout,
2018# that is, if url://repo/trunk is the current Git branch, and that
2019# directory contains a sub-directory `d', SUB will be invoked with `/d/'
2020# as `path' (note the trailing `/').
2021sub prop_walk {
2022        my ($self, $path, $rev, $sub) = @_;
2023
2024        $path =~ s#^/##;
2025        my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev);
2026        $path =~ s#^/*#/#g;
2027        my $p = $path;
2028        # Strip the irrelevant part of the path.
2029        $p =~ s#^/+\Q$self->{path}\E(/|$)#/#;
2030        # Ensure the path is terminated by a `/'.
2031        $p =~ s#/*$#/#;
2032
2033        # The properties contain all the internal SVN stuff nobody
2034        # (usually) cares about.
2035        my $interesting_props = 0;
2036        foreach (keys %{$props}) {
2037                # If it doesn't start with `svn:', it must be a
2038                # user-defined property.
2039                ++$interesting_props and next if $_ !~ /^svn:/;
2040                # FIXME: Fragile, if SVN adds new public properties,
2041                # this needs to be updated.
2042                ++$interesting_props if /^svn:(?:ignore|keywords|executable
2043                                                 |eol-style|mime-type
2044                                                 |externals|needs-lock)$/x;
2045        }
2046        &$sub($self, $p, $props) if $interesting_props;
2047
2048        foreach (sort keys %$dirent) {
2049                next if $dirent->{$_}->{kind} != $SVN::Node::dir;
2050                $self->prop_walk($self->{path} . $p . $_, $rev, $sub);
2051        }
2052}
2053
2054sub last_rev { ($_[0]->last_rev_commit)[0] }
2055sub last_commit { ($_[0]->last_rev_commit)[1] }
2056
2057# returns the newest SVN revision number and newest commit SHA1
2058sub last_rev_commit {
2059        my ($self) = @_;
2060        if (defined $self->{last_rev} && defined $self->{last_commit}) {
2061                return ($self->{last_rev}, $self->{last_commit});
2062        }
2063        my $c = ::verify_ref($self->refname.'^0');
2064        if ($c && !$self->use_svm_props && !$self->no_metadata) {
2065                my $rev = (::cmt_metadata($c))[1];
2066                if (defined $rev) {
2067                        ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
2068                        return ($rev, $c);
2069                }
2070        }
2071        my $map_path = $self->map_path;
2072        unless (-e $map_path) {
2073                ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
2074                return (undef, undef);
2075        }
2076        my ($rev, $commit) = $self->rev_map_max(1);
2077        ($self->{last_rev}, $self->{last_commit}) = ($rev, $commit);
2078        return ($rev, $commit);
2079}
2080
2081sub get_fetch_range {
2082        my ($self, $min, $max) = @_;
2083        $max ||= $self->ra->get_latest_revnum;
2084        $min ||= $self->rev_map_max;
2085        (++$min, $max);
2086}
2087
2088sub tmp_config {
2089        my (@args) = @_;
2090        my $old_def_config = "$ENV{GIT_DIR}/svn/config";
2091        my $config = "$ENV{GIT_DIR}/svn/.metadata";
2092        if (! -f $config && -f $old_def_config) {
2093                rename $old_def_config, $config or
2094                       die "Failed rename $old_def_config => $config: $!\n";
2095        }
2096        my $old_config = $ENV{GIT_CONFIG};
2097        $ENV{GIT_CONFIG} = $config;
2098        $@ = undef;
2099        my @ret = eval {
2100                unless (-f $config) {
2101                        mkfile($config);
2102                        open my $fh, '>', $config or
2103                            die "Can't open $config: $!\n";
2104                        print $fh "; This file is used internally by ",
2105                                  "git-svn\n" or die
2106                                  "Couldn't write to $config: $!\n";
2107                        print $fh "; You should not have to edit it\n" or
2108                              die "Couldn't write to $config: $!\n";
2109                        close $fh or die "Couldn't close $config: $!\n";
2110                }
2111                command('config', @args);
2112        };
2113        my $err = $@;
2114        if (defined $old_config) {
2115                $ENV{GIT_CONFIG} = $old_config;
2116        } else {
2117                delete $ENV{GIT_CONFIG};
2118        }
2119        die $err if $err;
2120        wantarray ? @ret : $ret[0];
2121}
2122
2123sub tmp_index_do {
2124        my ($self, $sub) = @_;
2125        my $old_index = $ENV{GIT_INDEX_FILE};
2126        $ENV{GIT_INDEX_FILE} = $self->{index};
2127        $@ = undef;
2128        my @ret = eval {
2129                my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
2130                mkpath([$dir]) unless -d $dir;
2131                &$sub;
2132        };
2133        my $err = $@;
2134        if (defined $old_index) {
2135                $ENV{GIT_INDEX_FILE} = $old_index;
2136        } else {
2137                delete $ENV{GIT_INDEX_FILE};
2138        }
2139        die $err if $err;
2140        wantarray ? @ret : $ret[0];
2141}
2142
2143sub assert_index_clean {
2144        my ($self, $treeish) = @_;
2145
2146        $self->tmp_index_do(sub {
2147                command_noisy('read-tree', $treeish) unless -e $self->{index};
2148                my $x = command_oneline('write-tree');
2149                my ($y) = (command(qw/cat-file commit/, $treeish) =~
2150                           /^tree ($::sha1)/mo);
2151                return if $y eq $x;
2152
2153                warn "Index mismatch: $y != $x\nrereading $treeish\n";
2154                unlink $self->{index} or die "unlink $self->{index}: $!\n";
2155                command_noisy('read-tree', $treeish);
2156                $x = command_oneline('write-tree');
2157                if ($y ne $x) {
2158                        ::fatal "trees ($treeish) $y != $x\n",
2159                                "Something is seriously wrong...";
2160                }
2161        });
2162}
2163
2164sub get_commit_parents {
2165        my ($self, $log_entry) = @_;
2166        my (%seen, @ret, @tmp);
2167        # legacy support for 'set-tree'; this is only used by set_tree_cb:
2168        if (my $ip = $self->{inject_parents}) {
2169                if (my $commit = delete $ip->{$log_entry->{revision}}) {
2170                        push @tmp, $commit;
2171                }
2172        }
2173        if (my $cur = ::verify_ref($self->refname.'^0')) {
2174                push @tmp, $cur;
2175        }
2176        if (my $ipd = $self->{inject_parents_dcommit}) {
2177                if (my $commit = delete $ipd->{$log_entry->{revision}}) {
2178                        push @tmp, @$commit;
2179                }
2180        }
2181        push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
2182        while (my $p = shift @tmp) {
2183                next if $seen{$p};
2184                $seen{$p} = 1;
2185                push @ret, $p;
2186                # MAXPARENT is defined to 16 in commit-tree.c:
2187                last if @ret >= 16;
2188        }
2189        if (@tmp) {
2190                die "r$log_entry->{revision}: No room for parents:\n\t",
2191                    join("\n\t", @tmp), "\n";
2192        }
2193        @ret;
2194}
2195
2196sub rewrite_root {
2197        my ($self) = @_;
2198        return $self->{-rewrite_root} if exists $self->{-rewrite_root};
2199        my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
2200        my $rwr = eval { command_oneline(qw/config --get/, $k) };
2201        if ($rwr) {
2202                $rwr =~ s#/+$##;
2203                if ($rwr !~ m#^[a-z\+]+://#) {
2204                        die "$rwr is not a valid URL (key: $k)\n";
2205                }
2206        }
2207        $self->{-rewrite_root} = $rwr;
2208}
2209
2210sub metadata_url {
2211        my ($self) = @_;
2212        ($self->rewrite_root || $self->{url}) .
2213           (length $self->{path} ? '/' . $self->{path} : '');
2214}
2215
2216sub full_url {
2217        my ($self) = @_;
2218        $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
2219}
2220
2221
2222sub set_commit_header_env {
2223        my ($log_entry) = @_;
2224        my %env;
2225        foreach my $ned (qw/NAME EMAIL DATE/) {
2226                foreach my $ac (qw/AUTHOR COMMITTER/) {
2227                        $env{"GIT_${ac}_${ned}"} = $ENV{"GIT_${ac}_${ned}"};
2228                }
2229        }
2230
2231        $ENV{GIT_AUTHOR_NAME} = $log_entry->{name};
2232        $ENV{GIT_AUTHOR_EMAIL} = $log_entry->{email};
2233        $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
2234
2235        $ENV{GIT_COMMITTER_NAME} = (defined $log_entry->{commit_name})
2236                                                ? $log_entry->{commit_name}
2237                                                : $log_entry->{name};
2238        $ENV{GIT_COMMITTER_EMAIL} = (defined $log_entry->{commit_email})
2239                                                ? $log_entry->{commit_email}
2240                                                : $log_entry->{email};
2241        \%env;
2242}
2243
2244sub restore_commit_header_env {
2245        my ($env) = @_;
2246        foreach my $ned (qw/NAME EMAIL DATE/) {
2247                foreach my $ac (qw/AUTHOR COMMITTER/) {
2248                        my $k = "GIT_${ac}_${ned}";
2249                        if (defined $env->{$k}) {
2250                                $ENV{$k} = $env->{$k};
2251                        } else {
2252                                delete $ENV{$k};
2253                        }
2254                }
2255        }
2256}
2257
2258sub gc {
2259        command_noisy('gc', '--auto');
2260};
2261
2262sub do_git_commit {
2263        my ($self, $log_entry) = @_;
2264        my $lr = $self->last_rev;
2265        if (defined $lr && $lr >= $log_entry->{revision}) {
2266                die "Last fetched revision of ", $self->refname,
2267                    " was r$lr, but we are about to fetch: ",
2268                    "r$log_entry->{revision}!\n";
2269        }
2270        if (my $c = $self->rev_map_get($log_entry->{revision})) {
2271                croak "$log_entry->{revision} = $c already exists! ",
2272                      "Why are we refetching it?\n";
2273        }
2274        my $old_env = set_commit_header_env($log_entry);
2275        my $tree = $log_entry->{tree};
2276        if (!defined $tree) {
2277                $tree = $self->tmp_index_do(sub {
2278                                            command_oneline('write-tree') });
2279        }
2280        die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
2281
2282        my @exec = ('git', 'commit-tree', $tree);
2283        foreach ($self->get_commit_parents($log_entry)) {
2284                push @exec, '-p', $_;
2285        }
2286        defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
2287                                                                   or croak $!;
2288        binmode $msg_fh;
2289
2290        # we always get UTF-8 from SVN, but we may want our commits in
2291        # a different encoding.
2292        if (my $enc = Git::config('i18n.commitencoding')) {
2293                require Encode;
2294                Encode::from_to($log_entry->{log}, 'UTF-8', $enc);
2295        }
2296        print $msg_fh $log_entry->{log} or croak $!;
2297        restore_commit_header_env($old_env);
2298        unless ($self->no_metadata) {
2299                print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
2300                              or croak $!;
2301        }
2302        $msg_fh->flush == 0 or croak $!;
2303        close $msg_fh or croak $!;
2304        chomp(my $commit = do { local $/; <$out_fh> });
2305        close $out_fh or croak $!;
2306        waitpid $pid, 0;
2307        croak $? if $?;
2308        if ($commit !~ /^$::sha1$/o) {
2309                die "Failed to commit, invalid sha1: $commit\n";
2310        }
2311
2312        $self->rev_map_set($log_entry->{revision}, $commit, 1);
2313
2314        $self->{last_rev} = $log_entry->{revision};
2315        $self->{last_commit} = $commit;
2316        print "r$log_entry->{revision}";
2317        if (defined $log_entry->{svm_revision}) {
2318                 print " (\@$log_entry->{svm_revision})";
2319                 $self->rev_map_set($log_entry->{svm_revision}, $commit,
2320                                   0, $self->svm_uuid);
2321        }
2322        print " = $commit ($self->{ref_id})\n";
2323        if (--$_gc_nr == 0) {
2324                $_gc_nr = $_gc_period;
2325                gc();
2326        }
2327        return $commit;
2328}
2329
2330sub match_paths {
2331        my ($self, $paths, $r) = @_;
2332        return 1 if $self->{path} eq '';
2333        if (my $path = $paths->{"/$self->{path}"}) {
2334                return ($path->{action} eq 'D') ? 0 : 1;
2335        }
2336        $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
2337        if (grep /$self->{path_regex}/, keys %$paths) {
2338                return 1;
2339        }
2340        my $c = '';
2341        foreach (split m#/#, $self->{path}) {
2342                $c .= "/$_";
2343                next unless ($paths->{$c} &&
2344                             ($paths->{$c}->{action} =~ /^[AR]$/));
2345                if ($self->ra->check_path($self->{path}, $r) ==
2346                    $SVN::Node::dir) {
2347                        return 1;
2348                }
2349        }
2350        return 0;
2351}
2352
2353sub find_parent_branch {
2354        my ($self, $paths, $rev) = @_;
2355        return undef unless $self->follow_parent;
2356        unless (defined $paths) {
2357                my $err_handler = $SVN::Error::handler;
2358                $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
2359                $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1, sub {
2360                                   $paths =
2361                                      Git::SVN::Ra::dup_changed_paths($_[0]) });
2362                $SVN::Error::handler = $err_handler;
2363        }
2364        return undef unless defined $paths;
2365
2366        # look for a parent from another branch:
2367        my @b_path_components = split m#/#, $self->rel_path;
2368        my @a_path_components;
2369        my $i;
2370        while (@b_path_components) {
2371                $i = $paths->{'/'.join('/', @b_path_components)};
2372                last if $i && defined $i->{copyfrom_path};
2373                unshift(@a_path_components, pop(@b_path_components));
2374        }
2375        return undef unless defined $i && defined $i->{copyfrom_path};
2376        my $branch_from = $i->{copyfrom_path};
2377        if (@a_path_components) {
2378                print STDERR "branch_from: $branch_from => ";
2379                $branch_from .= '/'.join('/', @a_path_components);
2380                print STDERR $branch_from, "\n";
2381        }
2382        my $r = $i->{copyfrom_rev};
2383        my $repos_root = $self->ra->{repos_root};
2384        my $url = $self->ra->{url};
2385        my $new_url = $repos_root . $branch_from;
2386        print STDERR  "Found possible branch point: ",
2387                      "$new_url => ", $self->full_url, ", $r\n";
2388        $branch_from =~ s#^/##;
2389        my $gs = Git::SVN->find_by_url($new_url, $repos_root, $branch_from);
2390        unless ($gs) {
2391                my $ref_id = $self->{ref_id};
2392                $ref_id =~ s/\@\d+$//;
2393                $ref_id .= "\@$r";
2394                # just grow a tail if we're not unique enough :x
2395                $ref_id .= '-' while find_ref($ref_id);
2396                print STDERR "Initializing parent: $ref_id\n";
2397                my ($u, $p, $repo_id) = ($new_url, '', $ref_id);
2398                if ($u =~ s#^\Q$url\E(/|$)##) {
2399                        $p = $u;
2400                        $u = $url;
2401                        $repo_id = $self->{repo_id};
2402                }
2403                $gs = Git::SVN->init($u, $p, $repo_id, $ref_id, 1);
2404        }
2405        my ($r0, $parent) = $gs->find_rev_before($r, 1);
2406        {
2407                my ($base, $head);
2408                if (!defined $r0 || !defined $parent) {
2409                        ($base, $head) = parse_revision_argument(0, $r);
2410                } else {
2411                        if ($r0 < $r) {
2412                                $gs->ra->get_log([$gs->{path}], $r0 + 1, $r, 1,
2413                                        0, 1, sub { $base = $_[1] - 1 });
2414                        }
2415                }
2416                if (defined $base && $base <= $r) {
2417                        $gs->fetch($base, $r);
2418                }
2419                ($r0, $parent) = $gs->find_rev_before($r, 1);
2420        }
2421        if (defined $r0 && defined $parent) {
2422                print STDERR "Found branch parent: ($self->{ref_id}) $parent\n";
2423                my $ed;
2424                if ($self->ra->can_do_switch) {
2425                        $self->assert_index_clean($parent);
2426                        print STDERR "Following parent with do_switch\n";
2427                        # do_switch works with svn/trunk >= r22312, but that
2428                        # is not included with SVN 1.4.3 (the latest version
2429                        # at the moment), so we can't rely on it
2430                        $self->{last_commit} = $parent;
2431                        $ed = SVN::Git::Fetcher->new($self);
2432                        $gs->ra->gs_do_switch($r0, $rev, $gs,
2433                                              $self->full_url, $ed)
2434                          or die "SVN connection failed somewhere...\n";
2435                } elsif ($self->ra->trees_match($new_url, $r0,
2436                                                $self->full_url, $rev)) {
2437                        print STDERR "Trees match:\n",
2438                                     "  $new_url\@$r0\n",
2439                                     "  ${\$self->full_url}\@$rev\n",
2440                                     "Following parent with no changes\n";
2441                        $self->tmp_index_do(sub {
2442                            command_noisy('read-tree', $parent);
2443                        });
2444                        $self->{last_commit} = $parent;
2445                } else {
2446                        print STDERR "Following parent with do_update\n";
2447                        $ed = SVN::Git::Fetcher->new($self);
2448                        $self->ra->gs_do_update($rev, $rev, $self, $ed)
2449                          or die "SVN connection failed somewhere...\n";
2450                }
2451                print STDERR "Successfully followed parent\n";
2452                return $self->make_log_entry($rev, [$parent], $ed);
2453        }
2454        return undef;
2455}
2456
2457sub do_fetch {
2458        my ($self, $paths, $rev) = @_;
2459        my $ed;
2460        my ($last_rev, @parents);
2461        if (my $lc = $self->last_commit) {
2462                # we can have a branch that was deleted, then re-added
2463                # under the same name but copied from another path, in
2464                # which case we'll have multiple parents (we don't
2465                # want to break the original ref, nor lose copypath info):
2466                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2467                        push @{$log_entry->{parents}}, $lc;
2468                        return $log_entry;
2469                }
2470                $ed = SVN::Git::Fetcher->new($self);
2471                $last_rev = $self->{last_rev};
2472                $ed->{c} = $lc;
2473                @parents = ($lc);
2474        } else {
2475                $last_rev = $rev;
2476                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2477                        return $log_entry;
2478                }
2479                $ed = SVN::Git::Fetcher->new($self);
2480        }
2481        unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
2482                die "SVN connection failed somewhere...\n";
2483        }
2484        $self->make_log_entry($rev, \@parents, $ed);
2485}
2486
2487sub get_untracked {
2488        my ($self, $ed) = @_;
2489        my @out;
2490        my $h = $ed->{empty};
2491        foreach (sort keys %$h) {
2492                my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
2493                push @out, "  $act: " . uri_encode($_);
2494                warn "W: $act: $_\n";
2495        }
2496        foreach my $t (qw/dir_prop file_prop/) {
2497                $h = $ed->{$t} or next;
2498                foreach my $path (sort keys %$h) {
2499                        my $ppath = $path eq '' ? '.' : $path;
2500                        foreach my $prop (sort keys %{$h->{$path}}) {
2501                                next if $SKIP_PROP{$prop};
2502                                my $v = $h->{$path}->{$prop};
2503                                my $t_ppath_prop = "$t: " .
2504                                                    uri_encode($ppath) . ' ' .
2505                                                    uri_encode($prop);
2506                                if (defined $v) {
2507                                        push @out, "  +$t_ppath_prop " .
2508                                                   uri_encode($v);
2509                                } else {
2510                                        push @out, "  -$t_ppath_prop";
2511                                }
2512                        }
2513                }
2514        }
2515        foreach my $t (qw/absent_file absent_directory/) {
2516                $h = $ed->{$t} or next;
2517                foreach my $parent (sort keys %$h) {
2518                        foreach my $path (sort @{$h->{$parent}}) {
2519                                push @out, "  $t: " .
2520                                           uri_encode("$parent/$path");
2521                                warn "W: $t: $parent/$path ",
2522                                     "Insufficient permissions?\n";
2523                        }
2524                }
2525        }
2526        \@out;
2527}
2528
2529sub parse_svn_date {
2530        my $date = shift || return '+0000 1970-01-01 00:00:00';
2531        my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2532                                            (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x) or
2533                                         croak "Unable to parse date: $date\n";
2534        "+0000 $Y-$m-$d $H:$M:$S";
2535}
2536
2537sub check_author {
2538        my ($author) = @_;
2539        if (!defined $author || length $author == 0) {
2540                $author = '(no author)';
2541        } elsif (defined $::_authors && ! defined $::users{$author}) {
2542                die "Author: $author not defined in $::_authors file\n";
2543        }
2544        $author;
2545}
2546
2547sub make_log_entry {
2548        my ($self, $rev, $parents, $ed) = @_;
2549        my $untracked = $self->get_untracked($ed);
2550
2551        open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
2552        print $un "r$rev\n" or croak $!;
2553        print $un $_, "\n" foreach @$untracked;
2554        my %log_entry = ( parents => $parents || [], revision => $rev,
2555                          log => '');
2556
2557        my $headrev;
2558        my $logged = delete $self->{logged_rev_props};
2559        if (!$logged || $self->{-want_revprops}) {
2560                my $rp = $self->ra->rev_proplist($rev);
2561                foreach (sort keys %$rp) {
2562                        my $v = $rp->{$_};
2563                        if (/^svn:(author|date|log)$/) {
2564                                $log_entry{$1} = $v;
2565                        } elsif ($_ eq 'svm:headrev') {
2566                                $headrev = $v;
2567                        } else {
2568                                print $un "  rev_prop: ", uri_encode($_), ' ',
2569                                          uri_encode($v), "\n";
2570                        }
2571                }
2572        } else {
2573                map { $log_entry{$_} = $logged->{$_} } keys %$logged;
2574        }
2575        close $un or croak $!;
2576
2577        $log_entry{date} = parse_svn_date($log_entry{date});
2578        $log_entry{log} .= "\n";
2579        my $author = $log_entry{author} = check_author($log_entry{author});
2580        my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
2581                                                       : ($author, undef);
2582
2583        my ($commit_name, $commit_email) = ($name, $email);
2584        if ($_use_log_author) {
2585                my $name_field;
2586                if ($log_entry{log} =~ /From:\s+(.*\S)\s*\n/i) {
2587                        $name_field = $1;
2588                } elsif ($log_entry{log} =~ /Signed-off-by:\s+(.*\S)\s*\n/i) {
2589                        $name_field = $1;
2590                }
2591                if (!defined $name_field) {
2592                        if (!defined $email) {
2593                                $email = $name;
2594                        }
2595                } elsif ($name_field =~ /(.*?)\s+<(.*)>/) {
2596                        ($name, $email) = ($1, $2);
2597                } elsif ($name_field =~ /(.*)@/) {
2598                        ($name, $email) = ($1, $name_field);
2599                } else {
2600                        ($name, $email) = ($name_field, $name_field);
2601                }
2602        }
2603        if (defined $headrev && $self->use_svm_props) {
2604                if ($self->rewrite_root) {
2605                        die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
2606                            "options set!\n";
2607                }
2608                my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$};
2609                # we don't want "SVM: initializing mirror for junk" ...
2610                return undef if $r == 0;
2611                my $svm = $self->svm;
2612                if ($uuid ne $svm->{uuid}) {
2613                        die "UUID mismatch on SVM path:\n",
2614                            "expected: $svm->{uuid}\n",
2615                            "     got: $uuid\n";
2616                }
2617                my $full_url = $self->full_url;
2618                $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
2619                             die "Failed to replace '$svm->{replace}' with ",
2620                                 "'$svm->{source}' in $full_url\n";
2621                # throw away username for storing in records
2622                remove_username($full_url);
2623                $log_entry{metadata} = "$full_url\@$r $uuid";
2624                $log_entry{svm_revision} = $r;
2625                $email ||= "$author\@$uuid";
2626                $commit_email ||= "$author\@$uuid";
2627        } elsif ($self->use_svnsync_props) {
2628                my $full_url = $self->svnsync->{url};
2629                $full_url .= "/$self->{path}" if length $self->{path};
2630                remove_username($full_url);
2631                my $uuid = $self->svnsync->{uuid};
2632                $log_entry{metadata} = "$full_url\@$rev $uuid";
2633                $email ||= "$author\@$uuid";
2634                $commit_email ||= "$author\@$uuid";
2635        } else {
2636                my $url = $self->metadata_url;
2637                remove_username($url);
2638                $log_entry{metadata} = "$url\@$rev " .
2639                                       $self->ra->get_uuid;
2640                $email ||= "$author\@" . $self->ra->get_uuid;
2641                $commit_email ||= "$author\@" . $self->ra->get_uuid;
2642        }
2643        $log_entry{name} = $name;
2644        $log_entry{email} = $email;
2645        $log_entry{commit_name} = $commit_name;
2646        $log_entry{commit_email} = $commit_email;
2647        \%log_entry;
2648}
2649
2650sub fetch {
2651        my ($self, $min_rev, $max_rev, @parents) = @_;
2652        my ($last_rev, $last_commit) = $self->last_rev_commit;
2653        my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
2654        $self->ra->gs_fetch_loop_common($base, $head, [$self]);
2655}
2656
2657sub set_tree_cb {
2658        my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
2659        $self->{inject_parents} = { $rev => $tree };
2660        $self->fetch(undef, undef);
2661}
2662
2663sub set_tree {
2664        my ($self, $tree) = (shift, shift);
2665        my $log_entry = ::get_commit_entry($tree);
2666        unless ($self->{last_rev}) {
2667                ::fatal("Must have an existing revision to commit");
2668        }
2669        my %ed_opts = ( r => $self->{last_rev},
2670                        log => $log_entry->{log},
2671                        ra => $self->ra,
2672                        tree_a => $self->{last_commit},
2673                        tree_b => $tree,
2674                        editor_cb => sub {
2675                               $self->set_tree_cb($log_entry, $tree, @_) },
2676                        svn_path => $self->{path} );
2677        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
2678                print "No changes\nr$self->{last_rev} = $tree\n";
2679        }
2680}
2681
2682sub rebuild_from_rev_db {
2683        my ($self, $path) = @_;
2684        my $r = -1;
2685        open my $fh, '<', $path or croak "open: $!";
2686        binmode $fh or croak "binmode: $!";
2687        while (<$fh>) {
2688                length($_) == 41 or croak "inconsistent size in ($_) != 41";
2689                chomp($_);
2690                ++$r;
2691                next if $_ eq ('0' x 40);
2692                $self->rev_map_set($r, $_);
2693                print "r$r = $_\n";
2694        }
2695        close $fh or croak "close: $!";
2696        unlink $path or croak "unlink: $!";
2697}
2698
2699sub rebuild {
2700        my ($self) = @_;
2701        my $map_path = $self->map_path;
2702        my $partial = (-e $map_path && ! -z $map_path);
2703        return unless ::verify_ref($self->refname.'^0');
2704        if (!$partial && ($self->use_svm_props || $self->no_metadata)) {
2705                my $rev_db = $self->rev_db_path;
2706                $self->rebuild_from_rev_db($rev_db);
2707                if ($self->use_svm_props) {
2708                        my $svm_rev_db = $self->rev_db_path($self->svm_uuid);
2709                        $self->rebuild_from_rev_db($svm_rev_db);
2710                }
2711                $self->unlink_rev_db_symlink;
2712                return;
2713        }
2714        print "Rebuilding $map_path ...\n" if (!$partial);
2715        my ($base_rev, $head) = ($partial ? $self->rev_map_max_norebuild(1) :
2716                (undef, undef));
2717        my ($log, $ctx) =
2718            command_output_pipe(qw/rev-list --pretty=raw --no-color --reverse/,
2719                                ($head ? "$head.." : "") . $self->refname,
2720                                '--');
2721        my $metadata_url = $self->metadata_url;
2722        remove_username($metadata_url);
2723        my $svn_uuid = $self->ra_uuid;
2724        my $c;
2725        while (<$log>) {
2726                if ( m{^commit ($::sha1)$} ) {
2727                        $c = $1;
2728                        next;
2729                }
2730                next unless s{^\s*(git-svn-id:)}{$1};
2731                my ($url, $rev, $uuid) = ::extract_metadata($_);
2732                remove_username($url);
2733
2734                # ignore merges (from set-tree)
2735                next if (!defined $rev || !$uuid);
2736
2737                # if we merged or otherwise started elsewhere, this is
2738                # how we break out of it
2739                if (($uuid ne $svn_uuid) ||
2740                    ($metadata_url && $url && ($url ne $metadata_url))) {
2741                        next;
2742                }
2743                if ($partial && $head) {
2744                        print "Partial-rebuilding $map_path ...\n";
2745                        print "Currently at $base_rev = $head\n";
2746                        $head = undef;
2747                }
2748
2749                $self->rev_map_set($rev, $c);
2750                print "r$rev = $c\n";
2751        }
2752        command_close_pipe($log, $ctx);
2753        print "Done rebuilding $map_path\n" if (!$partial || !$head);
2754        my $rev_db_path = $self->rev_db_path;
2755        if (-f $self->rev_db_path) {
2756                unlink $self->rev_db_path or croak "unlink: $!";
2757        }
2758        $self->unlink_rev_db_symlink;
2759}
2760
2761# rev_map:
2762# Tie::File seems to be prone to offset errors if revisions get sparse,
2763# it's not that fast, either.  Tie::File is also not in Perl 5.6.  So
2764# one of my favorite modules is out :<  Next up would be one of the DBM
2765# modules, but I'm not sure which is most portable...
2766#
2767# This is the replacement for the rev_db format, which was too big
2768# and inefficient for large repositories with a lot of sparse history
2769# (mainly tags)
2770#
2771# The format is this:
2772#   - 24 bytes for every record,
2773#     * 4 bytes for the integer representing an SVN revision number
2774#     * 20 bytes representing the sha1 of a git commit
2775#   - No empty padding records like the old format
2776#     (except the last record, which can be overwritten)
2777#   - new records are written append-only since SVN revision numbers
2778#     increase monotonically
2779#   - lookups on SVN revision number are done via a binary search
2780#   - Piping the file to xxd -c24 is a good way of dumping it for
2781#     viewing or editing (piped back through xxd -r), should the need
2782#     ever arise.
2783#   - The last record can be padding revision with an all-zero sha1
2784#     This is used to optimize fetch performance when using multiple
2785#     "fetch" directives in .git/config
2786#
2787# These files are disposable unless noMetadata or useSvmProps is set
2788
2789sub _rev_map_set {
2790        my ($fh, $rev, $commit) = @_;
2791
2792        binmode $fh or croak "binmode: $!";
2793        my $size = (stat($fh))[7];
2794        ($size % 24) == 0 or croak "inconsistent size: $size";
2795
2796        my $wr_offset = 0;
2797        if ($size > 0) {
2798                sysseek($fh, -24, SEEK_END) or croak "seek: $!";
2799                my $read = sysread($fh, my $buf, 24) or croak "read: $!";
2800                $read == 24 or croak "read only $read bytes (!= 24)";
2801                my ($last_rev, $last_commit) = unpack(rev_map_fmt, $buf);
2802                if ($last_commit eq ('0' x40)) {
2803                        if ($size >= 48) {
2804                                sysseek($fh, -48, SEEK_END) or croak "seek: $!";
2805                                $read = sysread($fh, $buf, 24) or
2806                                    croak "read: $!";
2807                                $read == 24 or
2808                                    croak "read only $read bytes (!= 24)";
2809                                ($last_rev, $last_commit) =
2810                                    unpack(rev_map_fmt, $buf);
2811                                if ($last_commit eq ('0' x40)) {
2812                                        croak "inconsistent .rev_map\n";
2813                                }
2814                        }
2815                        if ($last_rev >= $rev) {
2816                                croak "last_rev is higher!: $last_rev >= $rev";
2817                        }
2818                        $wr_offset = -24;
2819                }
2820        }
2821        sysseek($fh, $wr_offset, SEEK_END) or croak "seek: $!";
2822        syswrite($fh, pack(rev_map_fmt, $rev, $commit), 24) == 24 or
2823          croak "write: $!";
2824}
2825
2826sub mkfile {
2827        my ($path) = @_;
2828        unless (-e $path) {
2829                my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
2830                mkpath([$dir]) unless -d $dir;
2831                open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
2832                close $fh or die "Couldn't close (create) $path: $!\n";
2833        }
2834}
2835
2836sub rev_map_set {
2837        my ($self, $rev, $commit, $update_ref, $uuid) = @_;
2838        length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
2839        my $db = $self->map_path($uuid);
2840        my $db_lock = "$db.lock";
2841        my $sig;
2842        if ($update_ref) {
2843                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
2844                            $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
2845        }
2846        mkfile($db);
2847
2848        $LOCKFILES{$db_lock} = 1;
2849        my $sync;
2850        # both of these options make our .rev_db file very, very important
2851        # and we can't afford to lose it because rebuild() won't work
2852        if ($self->use_svm_props || $self->no_metadata) {
2853                $sync = 1;
2854                copy($db, $db_lock) or die "rev_map_set(@_): ",
2855                                           "Failed to copy: ",
2856                                           "$db => $db_lock ($!)\n";
2857        } else {
2858                rename $db, $db_lock or die "rev_map_set(@_): ",
2859                                            "Failed to rename: ",
2860                                            "$db => $db_lock ($!)\n";
2861        }
2862
2863        sysopen(my $fh, $db_lock, O_RDWR | O_CREAT)
2864             or croak "Couldn't open $db_lock: $!\n";
2865        _rev_map_set($fh, $rev, $commit);
2866        if ($sync) {
2867                $fh->flush or die "Couldn't flush $db_lock: $!\n";
2868                $fh->sync or die "Couldn't sync $db_lock: $!\n";
2869        }
2870        close $fh or croak $!;
2871        if ($update_ref) {
2872                $_head = $self;
2873                command_noisy('update-ref', '-m', "r$rev",
2874                              $self->refname, $commit);
2875        }
2876        rename $db_lock, $db or die "rev_map_set(@_): ", "Failed to rename: ",
2877                                    "$db_lock => $db ($!)\n";
2878        delete $LOCKFILES{$db_lock};
2879        if ($update_ref) {
2880                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
2881                            $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
2882                kill $sig, $$ if defined $sig;
2883        }
2884}
2885
2886# If want_commit, this will return an array of (rev, commit) where
2887# commit _must_ be a valid commit in the archive.
2888# Otherwise, it'll return the max revision (whether or not the
2889# commit is valid or just a 0x40 placeholder).
2890sub rev_map_max {
2891        my ($self, $want_commit) = @_;
2892        $self->rebuild;
2893        my ($r, $c) = $self->rev_map_max_norebuild($want_commit);
2894        $want_commit ? ($r, $c) : $r;
2895}
2896
2897sub rev_map_max_norebuild {
2898        my ($self, $want_commit) = @_;
2899        my $map_path = $self->map_path;
2900        stat $map_path or return $want_commit ? (0, undef) : 0;
2901        sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
2902        binmode $fh or croak "binmode: $!";
2903        my $size = (stat($fh))[7];
2904        ($size % 24) == 0 or croak "inconsistent size: $size";
2905
2906        if ($size == 0) {
2907                close $fh or croak "close: $!";
2908                return $want_commit ? (0, undef) : 0;
2909        }
2910
2911        sysseek($fh, -24, SEEK_END) or croak "seek: $!";
2912        sysread($fh, my $buf, 24) == 24 or croak "read: $!";
2913        my ($r, $c) = unpack(rev_map_fmt, $buf);
2914        if ($want_commit && $c eq ('0' x40)) {
2915                if ($size < 48) {
2916                        return $want_commit ? (0, undef) : 0;
2917                }
2918                sysseek($fh, -48, SEEK_END) or croak "seek: $!";
2919                sysread($fh, $buf, 24) == 24 or croak "read: $!";
2920                ($r, $c) = unpack(rev_map_fmt, $buf);
2921                if ($c eq ('0'x40)) {
2922                        croak "Penultimate record is all-zeroes in $map_path";
2923                }
2924        }
2925        close $fh or croak "close: $!";
2926        $want_commit ? ($r, $c) : $r;
2927}
2928
2929sub rev_map_get {
2930        my ($self, $rev, $uuid) = @_;
2931        my $map_path = $self->map_path($uuid);
2932        return undef unless -e $map_path;
2933
2934        sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
2935        binmode $fh or croak "binmode: $!";
2936        my $size = (stat($fh))[7];
2937        ($size % 24) == 0 or croak "inconsistent size: $size";
2938
2939        if ($size == 0) {
2940                close $fh or croak "close: $fh";
2941                return undef;
2942        }
2943
2944        my ($l, $u) = (0, $size - 24);
2945        my ($r, $c, $buf);
2946
2947        while ($l <= $u) {
2948                my $i = int(($l/24 + $u/24) / 2) * 24;
2949                sysseek($fh, $i, SEEK_SET) or croak "seek: $!";
2950                sysread($fh, my $buf, 24) == 24 or croak "read: $!";
2951                my ($r, $c) = unpack('NH40', $buf);
2952
2953                if ($r < $rev) {
2954                        $l = $i + 24;
2955                } elsif ($r > $rev) {
2956                        $u = $i - 24;
2957                } else { # $r == $rev
2958                        close($fh) or croak "close: $!";
2959                        return $c eq ('0' x 40) ? undef : $c;
2960                }
2961        }
2962        close($fh) or croak "close: $!";
2963        undef;
2964}
2965
2966# Finds the first svn revision that exists on (if $eq_ok is true) or
2967# before $rev for the current branch.  It will not search any lower
2968# than $min_rev.  Returns the git commit hash and svn revision number
2969# if found, else (undef, undef).
2970sub find_rev_before {
2971        my ($self, $rev, $eq_ok, $min_rev) = @_;
2972        --$rev unless $eq_ok;
2973        $min_rev ||= 1;
2974        while ($rev >= $min_rev) {
2975                if (my $c = $self->rev_map_get($rev)) {
2976                        return ($rev, $c);
2977                }
2978                --$rev;
2979        }
2980        return (undef, undef);
2981}
2982
2983# Finds the first svn revision that exists on (if $eq_ok is true) or
2984# after $rev for the current branch.  It will not search any higher
2985# than $max_rev.  Returns the git commit hash and svn revision number
2986# if found, else (undef, undef).
2987sub find_rev_after {
2988        my ($self, $rev, $eq_ok, $max_rev) = @_;
2989        ++$rev unless $eq_ok;
2990        $max_rev ||= $self->rev_map_max;
2991        while ($rev <= $max_rev) {
2992                if (my $c = $self->rev_map_get($rev)) {
2993                        return ($rev, $c);
2994                }
2995                ++$rev;
2996        }
2997        return (undef, undef);
2998}
2999
3000sub _new {
3001        my ($class, $repo_id, $ref_id, $path) = @_;
3002        unless (defined $repo_id && length $repo_id) {
3003                $repo_id = $Git::SVN::default_repo_id;
3004        }
3005        unless (defined $ref_id && length $ref_id) {
3006                $_[2] = $ref_id = $Git::SVN::default_ref_id;
3007        }
3008        $_[1] = $repo_id;
3009        my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
3010        $_[3] = $path = '' unless (defined $path);
3011        mkpath(["$ENV{GIT_DIR}/svn"]);
3012        bless {
3013                ref_id => $ref_id, dir => $dir, index => "$dir/index",
3014                path => $path, config => "$ENV{GIT_DIR}/svn/config",
3015                map_root => "$dir/.rev_map", repo_id => $repo_id }, $class;
3016}
3017
3018# for read-only access of old .rev_db formats
3019sub unlink_rev_db_symlink {
3020        my ($self) = @_;
3021        my $link = $self->rev_db_path;
3022        $link =~ s/\.[\w-]+$// or croak "missing UUID at the end of $link";
3023        if (-l $link) {
3024                unlink $link or croak "unlink: $link failed!";
3025        }
3026}
3027
3028sub rev_db_path {
3029        my ($self, $uuid) = @_;
3030        my $db_path = $self->map_path($uuid);
3031        $db_path =~ s{/\.rev_map\.}{/\.rev_db\.}
3032            or croak "map_path: $db_path does not contain '/.rev_map.' !";
3033        $db_path;
3034}
3035
3036# the new replacement for .rev_db
3037sub map_path {
3038        my ($self, $uuid) = @_;
3039        $uuid ||= $self->ra_uuid;
3040        "$self->{map_root}.$uuid";
3041}
3042
3043sub uri_encode {
3044        my ($f) = @_;
3045        $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
3046        $f
3047}
3048
3049sub remove_username {
3050        $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
3051}
3052
3053package Git::SVN::Prompt;
3054use strict;
3055use warnings;
3056require SVN::Core;
3057use vars qw/$_no_auth_cache $_username/;
3058
3059sub simple {
3060        my ($cred, $realm, $default_username, $may_save, $pool) = @_;
3061        $may_save = undef if $_no_auth_cache;
3062        $default_username = $_username if defined $_username;
3063        if (defined $default_username && length $default_username) {
3064                if (defined $realm && length $realm) {
3065                        print STDERR "Authentication realm: $realm\n";
3066                        STDERR->flush;
3067                }
3068                $cred->username($default_username);
3069        } else {
3070                username($cred, $realm, $may_save, $pool);
3071        }
3072        $cred->password(_read_password("Password for '" .
3073                                       $cred->username . "': ", $realm));
3074        $cred->may_save($may_save);
3075        $SVN::_Core::SVN_NO_ERROR;
3076}
3077
3078sub ssl_server_trust {
3079        my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
3080        $may_save = undef if $_no_auth_cache;
3081        print STDERR "Error validating server certificate for '$realm':\n";
3082        {
3083                no warnings 'once';
3084                # All variables SVN::Auth::SSL::* are used only once,
3085                # so we're shutting up Perl warnings about this.
3086                if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
3087                        print STDERR " - The certificate is not issued ",
3088                            "by a trusted authority. Use the\n",
3089                            "   fingerprint to validate ",
3090                            "the certificate manually!\n";
3091                }
3092                if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
3093                        print STDERR " - The certificate hostname ",
3094                            "does not match.\n";
3095                }
3096                if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
3097                        print STDERR " - The certificate is not yet valid.\n";
3098                }
3099                if ($failures & $SVN::Auth::SSL::EXPIRED) {
3100                        print STDERR " - The certificate has expired.\n";
3101                }
3102                if ($failures & $SVN::Auth::SSL::OTHER) {
3103                        print STDERR " - The certificate has ",
3104                            "an unknown error.\n";
3105                }
3106        } # no warnings 'once'
3107        printf STDERR
3108                "Certificate information:\n".
3109                " - Hostname: %s\n".
3110                " - Valid: from %s until %s\n".
3111                " - Issuer: %s\n".
3112                " - Fingerprint: %s\n",
3113                map $cert_info->$_, qw(hostname valid_from valid_until
3114                                       issuer_dname fingerprint);
3115        my $choice;
3116prompt:
3117        print STDERR $may_save ?
3118              "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
3119              "(R)eject or accept (t)emporarily? ";
3120        STDERR->flush;
3121        $choice = lc(substr(<STDIN> || 'R', 0, 1));
3122        if ($choice =~ /^t$/i) {
3123                $cred->may_save(undef);
3124        } elsif ($choice =~ /^r$/i) {
3125                return -1;
3126        } elsif ($may_save && $choice =~ /^p$/i) {
3127                $cred->may_save($may_save);
3128        } else {
3129                goto prompt;
3130        }
3131        $cred->accepted_failures($failures);
3132        $SVN::_Core::SVN_NO_ERROR;
3133}
3134
3135sub ssl_client_cert {
3136        my ($cred, $realm, $may_save, $pool) = @_;
3137        $may_save = undef if $_no_auth_cache;
3138        print STDERR "Client certificate filename: ";
3139        STDERR->flush;
3140        chomp(my $filename = <STDIN>);
3141        $cred->cert_file($filename);
3142        $cred->may_save($may_save);
3143        $SVN::_Core::SVN_NO_ERROR;
3144}
3145
3146sub ssl_client_cert_pw {
3147        my ($cred, $realm, $may_save, $pool) = @_;
3148        $may_save = undef if $_no_auth_cache;
3149        $cred->password(_read_password("Password: ", $realm));
3150        $cred->may_save($may_save);
3151        $SVN::_Core::SVN_NO_ERROR;
3152}
3153
3154sub username {
3155        my ($cred, $realm, $may_save, $pool) = @_;
3156        $may_save = undef if $_no_auth_cache;
3157        if (defined $realm && length $realm) {
3158                print STDERR "Authentication realm: $realm\n";
3159        }
3160        my $username;
3161        if (defined $_username) {
3162                $username = $_username;
3163        } else {
3164                print STDERR "Username: ";
3165                STDERR->flush;
3166                chomp($username = <STDIN>);
3167        }
3168        $cred->username($username);
3169        $cred->may_save($may_save);
3170        $SVN::_Core::SVN_NO_ERROR;
3171}
3172
3173sub _read_password {
3174        my ($prompt, $realm) = @_;
3175        print STDERR $prompt;
3176        STDERR->flush;
3177        require Term::ReadKey;
3178        Term::ReadKey::ReadMode('noecho');
3179        my $password = '';
3180        while (defined(my $key = Term::ReadKey::ReadKey(0))) {
3181                last if $key =~ /[\012\015]/; # \n\r
3182                $password .= $key;
3183        }
3184        Term::ReadKey::ReadMode('restore');
3185        print STDERR "\n";
3186        STDERR->flush;
3187        $password;
3188}
3189
3190package SVN::Git::Fetcher;
3191use vars qw/@ISA/;
3192use strict;
3193use warnings;
3194use Carp qw/croak/;
3195use File::Temp qw/tempfile/;
3196use IO::File qw//;
3197
3198# file baton members: path, mode_a, mode_b, pool, fh, blob, base
3199sub new {
3200        my ($class, $git_svn) = @_;
3201        my $self = SVN::Delta::Editor->new;
3202        bless $self, $class;
3203        $self->{c} = $git_svn->{last_commit} if exists $git_svn->{last_commit};
3204        $self->{empty} = {};
3205        $self->{dir_prop} = {};
3206        $self->{file_prop} = {};
3207        $self->{absent_dir} = {};
3208        $self->{absent_file} = {};
3209        $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
3210        $self;
3211}
3212
3213sub set_path_strip {
3214        my ($self, $path) = @_;
3215        $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
3216}
3217
3218sub open_root {
3219        { path => '' };
3220}
3221
3222sub open_directory {
3223        my ($self, $path, $pb, $rev) = @_;
3224        { path => $path };
3225}
3226
3227sub git_path {
3228        my ($self, $path) = @_;
3229        if ($self->{path_strip}) {
3230                $path =~ s!$self->{path_strip}!! or
3231                  die "Failed to strip path '$path' ($self->{path_strip})\n";
3232        }
3233        $path;
3234}
3235
3236sub delete_entry {
3237        my ($self, $path, $rev, $pb) = @_;
3238
3239        my $gpath = $self->git_path($path);
3240        return undef if ($gpath eq '');
3241
3242        # remove entire directories.
3243        if (command('ls-tree', $self->{c}, '--', $gpath) =~ /^040000 tree/) {
3244                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3245                                                     -r --name-only -z/,
3246                                                     $self->{c}, '--', $gpath);
3247                local $/ = "\0";
3248                while (<$ls>) {
3249                        chomp;
3250                        $self->{gii}->remove($_);
3251                        print "\tD\t$_\n" unless $::_q;
3252                }
3253                print "\tD\t$gpath/\n" unless $::_q;
3254                command_close_pipe($ls, $ctx);
3255                $self->{empty}->{$path} = 0
3256        } else {
3257                $self->{gii}->remove($gpath);
3258                print "\tD\t$gpath\n" unless $::_q;
3259        }
3260        undef;
3261}
3262
3263sub open_file {
3264        my ($self, $path, $pb, $rev) = @_;
3265        my $gpath = $self->git_path($path);
3266        my ($mode, $blob) = (command('ls-tree', $self->{c}, '--', $gpath)
3267                             =~ /^(\d{6}) blob ([a-f\d]{40})\t/);
3268        unless (defined $mode && defined $blob) {
3269                die "$path was not found in commit $self->{c} (r$rev)\n";
3270        }
3271        { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
3272          pool => SVN::Pool->new, action => 'M' };
3273}
3274
3275sub add_file {
3276        my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
3277        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3278        delete $self->{empty}->{$dir};
3279        { path => $path, mode_a => 100644, mode_b => 100644,
3280          pool => SVN::Pool->new, action => 'A' };
3281}
3282
3283sub add_directory {
3284        my ($self, $path, $cp_path, $cp_rev) = @_;
3285        my $gpath = $self->git_path($path);
3286        if ($gpath eq '') {
3287                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3288                                                     -r --name-only -z/,
3289                                                     $self->{c});
3290                local $/ = "\0";
3291                while (<$ls>) {
3292                        chomp;
3293                        $self->{gii}->remove($_);
3294                        print "\tD\t$_\n" unless $::_q;
3295                }
3296                command_close_pipe($ls, $ctx);
3297                $self->{empty}->{$path} = 0;
3298        }
3299        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3300        delete $self->{empty}->{$dir};
3301        $self->{empty}->{$path} = 1;
3302        { path => $path };
3303}
3304
3305sub change_dir_prop {
3306        my ($self, $db, $prop, $value) = @_;
3307        $self->{dir_prop}->{$db->{path}} ||= {};
3308        $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
3309        undef;
3310}
3311
3312sub absent_directory {
3313        my ($self, $path, $pb) = @_;
3314        $self->{absent_dir}->{$pb->{path}} ||= [];
3315        push @{$self->{absent_dir}->{$pb->{path}}}, $path;
3316        undef;
3317}
3318
3319sub absent_file {
3320        my ($self, $path, $pb) = @_;
3321        $self->{absent_file}->{$pb->{path}} ||= [];
3322        push @{$self->{absent_file}->{$pb->{path}}}, $path;
3323        undef;
3324}
3325
3326sub change_file_prop {
3327        my ($self, $fb, $prop, $value) = @_;
3328        if ($prop eq 'svn:executable') {
3329                if ($fb->{mode_b} != 120000) {
3330                        $fb->{mode_b} = defined $value ? 100755 : 100644;
3331                }
3332        } elsif ($prop eq 'svn:special') {
3333                $fb->{mode_b} = defined $value ? 120000 : 100644;
3334        } else {
3335                $self->{file_prop}->{$fb->{path}} ||= {};
3336                $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
3337        }
3338        undef;
3339}
3340
3341sub apply_textdelta {
3342        my ($self, $fb, $exp) = @_;
3343        my $fh = $::_repository->temp_acquire('svn_delta');
3344        # $fh gets auto-closed() by SVN::TxDelta::apply(),
3345        # (but $base does not,) so dup() it for reading in close_file
3346        open my $dup, '<&', $fh or croak $!;
3347        my $base = $::_repository->temp_acquire('git_blob');
3348        if ($fb->{blob}) {
3349                print $base 'link ' if ($fb->{mode_a} == 120000);
3350                my $size = $::_repository->cat_blob($fb->{blob}, $base);
3351                die "Failed to read object $fb->{blob}" if ($size < 0);
3352
3353                if (defined $exp) {
3354                        seek $base, 0, 0 or croak $!;
3355                        my $got = ::md5sum($base);
3356                        die "Checksum mismatch: $fb->{path} $fb->{blob}\n",
3357                            "expected: $exp\n",
3358                            "     got: $got\n" if ($got ne $exp);
3359                }
3360        }
3361        seek $base, 0, 0 or croak $!;
3362        $fb->{fh} = $fh;
3363        $fb->{base} = $base;
3364        [ SVN::TxDelta::apply($base, $dup, undef, $fb->{path}, $fb->{pool}) ];
3365}
3366
3367sub close_file {
3368        my ($self, $fb, $exp) = @_;
3369        my $hash;
3370        my $path = $self->git_path($fb->{path});
3371        if (my $fh = $fb->{fh}) {
3372                if (defined $exp) {
3373                        seek($fh, 0, 0) or croak $!;
3374                        my $got = ::md5sum($fh);
3375                        if ($got ne $exp) {
3376                                die "Checksum mismatch: $path\n",
3377                                    "expected: $exp\n    got: $got\n";
3378                        }
3379                }
3380                if ($fb->{mode_b} == 120000) {
3381                        sysseek($fh, 0, 0) or croak $!;
3382                        sysread($fh, my $buf, 5) == 5 or croak $!;
3383
3384                        unless ($buf eq 'link ') {
3385                                warn "$path has mode 120000",
3386                                                " but is not a link\n";
3387                        } else {
3388                                my $tmp_fh = $::_repository->temp_acquire(
3389                                        'svn_hash');
3390                                my $res;
3391                                while ($res = sysread($fh, my $str, 1024)) {
3392                                        my $out = syswrite($tmp_fh, $str, $res);
3393                                        defined($out) && $out == $res
3394                                                or croak("write ",
3395                                                        Git::temp_path($tmp_fh),
3396                                                        ": $!\n");
3397                                }
3398                                defined $res or croak $!;
3399
3400                                ($fh, $tmp_fh) = ($tmp_fh, $fh);
3401                                Git::temp_release($tmp_fh, 1);
3402                        }
3403                }
3404
3405                $hash = $::_repository->hash_and_insert_object(
3406                                Git::temp_path($fh));
3407                $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
3408
3409                Git::temp_release($fb->{base}, 1);
3410                Git::temp_release($fh, 1);
3411        } else {
3412                $hash = $fb->{blob} or die "no blob information\n";
3413        }
3414        $fb->{pool}->clear;
3415        $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
3416        print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
3417        undef;
3418}
3419
3420sub abort_edit {
3421        my $self = shift;
3422        $self->{nr} = $self->{gii}->{nr};
3423        delete $self->{gii};
3424        $self->SUPER::abort_edit(@_);
3425}
3426
3427sub close_edit {
3428        my $self = shift;
3429        $self->{git_commit_ok} = 1;
3430        $self->{nr} = $self->{gii}->{nr};
3431        delete $self->{gii};
3432        $self->SUPER::close_edit(@_);
3433}
3434
3435package SVN::Git::Editor;
3436use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
3437use strict;
3438use warnings;
3439use Carp qw/croak/;
3440use IO::File;
3441
3442sub new {
3443        my ($class, $opts) = @_;
3444        foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
3445                die "$_ required!\n" unless (defined $opts->{$_});
3446        }
3447
3448        my $pool = SVN::Pool->new;
3449        my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
3450        my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
3451                                     $opts->{r}, $mods);
3452
3453        # $opts->{ra} functions should not be used after this:
3454        my @ce  = $opts->{ra}->get_commit_editor($opts->{log},
3455                                                $opts->{editor_cb}, $pool);
3456        my $self = SVN::Delta::Editor->new(@ce, $pool);
3457        bless $self, $class;
3458        foreach (qw/svn_path r tree_a tree_b/) {
3459                $self->{$_} = $opts->{$_};
3460        }
3461        $self->{url} = $opts->{ra}->{url};
3462        $self->{mods} = $mods;
3463        $self->{types} = $types;
3464        $self->{pool} = $pool;
3465        $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
3466        $self->{rm} = { };
3467        $self->{path_prefix} = length $self->{svn_path} ?
3468                               "$self->{svn_path}/" : '';
3469        $self->{config} = $opts->{config};
3470        return $self;
3471}
3472
3473sub generate_diff {
3474        my ($tree_a, $tree_b) = @_;
3475        my @diff_tree = qw(diff-tree -z -r);
3476        if ($_cp_similarity) {
3477                push @diff_tree, "-C$_cp_similarity";
3478        } else {
3479                push @diff_tree, '-C';
3480        }
3481        push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
3482        push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
3483        push @diff_tree, $tree_a, $tree_b;
3484        my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
3485        local $/ = "\0";
3486        my $state = 'meta';
3487        my @mods;
3488        while (<$diff_fh>) {
3489                chomp $_; # this gets rid of the trailing "\0"
3490                if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
3491                                        ($::sha1)\s($::sha1)\s
3492                                        ([MTCRAD])\d*$/xo) {
3493                        push @mods, {   mode_a => $1, mode_b => $2,
3494                                        sha1_a => $3, sha1_b => $4,
3495                                        chg => $5 };
3496                        if ($5 =~ /^(?:C|R)$/) {
3497                                $state = 'file_a';
3498                        } else {
3499                                $state = 'file_b';
3500                        }
3501                } elsif ($state eq 'file_a') {
3502                        my $x = $mods[$#mods] or croak "Empty array\n";
3503                        if ($x->{chg} !~ /^(?:C|R)$/) {
3504                                croak "Error parsing $_, $x->{chg}\n";
3505                        }
3506                        $x->{file_a} = $_;
3507                        $state = 'file_b';
3508                } elsif ($state eq 'file_b') {
3509                        my $x = $mods[$#mods] or croak "Empty array\n";
3510                        if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
3511                                croak "Error parsing $_, $x->{chg}\n";
3512                        }
3513                        if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
3514                                croak "Error parsing $_, $x->{chg}\n";
3515                        }
3516                        $x->{file_b} = $_;
3517                        $state = 'meta';
3518                } else {
3519                        croak "Error parsing $_\n";
3520                }
3521        }
3522        command_close_pipe($diff_fh, $ctx);
3523        \@mods;
3524}
3525
3526sub check_diff_paths {
3527        my ($ra, $pfx, $rev, $mods) = @_;
3528        my %types;
3529        $pfx .= '/' if length $pfx;
3530
3531        sub type_diff_paths {
3532                my ($ra, $types, $path, $rev) = @_;
3533                my @p = split m#/+#, $path;
3534                my $c = shift @p;
3535                unless (defined $types->{$c}) {
3536                        $types->{$c} = $ra->check_path($c, $rev);
3537                }
3538                while (@p) {
3539                        $c .= '/' . shift @p;
3540                        next if defined $types->{$c};
3541                        $types->{$c} = $ra->check_path($c, $rev);
3542                }
3543        }
3544
3545        foreach my $m (@$mods) {
3546                foreach my $f (qw/file_a file_b/) {
3547                        next unless defined $m->{$f};
3548                        my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
3549                        if (length $pfx.$dir && ! defined $types{$dir}) {
3550                                type_diff_paths($ra, \%types, $pfx.$dir, $rev);
3551                        }
3552                }
3553        }
3554        \%types;
3555}
3556
3557sub split_path {
3558        return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
3559}
3560
3561sub repo_path {
3562        my ($self, $path) = @_;
3563        $self->{path_prefix}.(defined $path ? $path : '');
3564}
3565
3566sub url_path {
3567        my ($self, $path) = @_;
3568        if ($self->{url} =~ m#^https?://#) {
3569                $path =~ s/([^~a-zA-Z0-9_.-])/uc sprintf("%%%02x",ord($1))/eg;
3570        }
3571        $self->{url} . '/' . $self->repo_path($path);
3572}
3573
3574sub rmdirs {
3575        my ($self) = @_;
3576        my $rm = $self->{rm};
3577        delete $rm->{''}; # we never delete the url we're tracking
3578        return unless %$rm;
3579
3580        foreach (keys %$rm) {
3581                my @d = split m#/#, $_;
3582                my $c = shift @d;
3583                $rm->{$c} = 1;
3584                while (@d) {
3585                        $c .= '/' . shift @d;
3586                        $rm->{$c} = 1;
3587                }
3588        }
3589        delete $rm->{$self->{svn_path}};
3590        delete $rm->{''}; # we never delete the url we're tracking
3591        return unless %$rm;
3592
3593        my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
3594                                             $self->{tree_b});
3595        local $/ = "\0";
3596        while (<$fh>) {
3597                chomp;
3598                my @dn = split m#/#, $_;
3599                while (pop @dn) {
3600                        delete $rm->{join '/', @dn};
3601                }
3602                unless (%$rm) {
3603                        close $fh;
3604                        return;
3605                }
3606        }
3607        command_close_pipe($fh, $ctx);
3608
3609        my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
3610        foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
3611                $self->close_directory($bat->{$d}, $p);
3612                my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
3613                print "\tD+\t$d/\n" unless $::_q;
3614                $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
3615                delete $bat->{$d};
3616        }
3617}
3618
3619sub open_or_add_dir {
3620        my ($self, $full_path, $baton) = @_;
3621        my $t = $self->{types}->{$full_path};
3622        if (!defined $t) {
3623                die "$full_path not known in r$self->{r} or we have a bug!\n";
3624        }
3625        {
3626                no warnings 'once';
3627                # SVN::Node::none and SVN::Node::file are used only once,
3628                # so we're shutting up Perl's warnings about them.
3629                if ($t == $SVN::Node::none) {
3630                        return $self->add_directory($full_path, $baton,
3631                            undef, -1, $self->{pool});
3632                } elsif ($t == $SVN::Node::dir) {
3633                        return $self->open_directory($full_path, $baton,
3634                            $self->{r}, $self->{pool});
3635                } # no warnings 'once'
3636                print STDERR "$full_path already exists in repository at ",
3637                    "r$self->{r} and it is not a directory (",
3638                    ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
3639        } # no warnings 'once'
3640        exit 1;
3641}
3642
3643sub ensure_path {
3644        my ($self, $path) = @_;
3645        my $bat = $self->{bat};
3646        my $repo_path = $self->repo_path($path);
3647        return $bat->{''} unless (length $repo_path);
3648        my @p = split m#/+#, $repo_path;
3649        my $c = shift @p;
3650        $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
3651        while (@p) {
3652                my $c0 = $c;
3653                $c .= '/' . shift @p;
3654                $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
3655        }
3656        return $bat->{$c};
3657}
3658
3659# Subroutine to convert a globbing pattern to a regular expression.
3660# From perl cookbook.
3661sub glob2pat {
3662        my $globstr = shift;
3663        my %patmap = ('*' => '.*', '?' => '.', '[' => '[', ']' => ']');
3664        $globstr =~ s{(.)} { $patmap{$1} || "\Q$1" }ge;
3665        return '^' . $globstr . '$';
3666}
3667
3668sub check_autoprop {
3669        my ($self, $pattern, $properties, $file, $fbat) = @_;
3670        # Convert the globbing pattern to a regular expression.
3671        my $regex = glob2pat($pattern);
3672        # Check if the pattern matches the file name.
3673        if($file =~ m/($regex)/) {
3674                # Parse the list of properties to set.
3675                my @props = split(/;/, $properties);
3676                foreach my $prop (@props) {
3677                        # Parse 'name=value' syntax and set the property.
3678                        if ($prop =~ /([^=]+)=(.*)/) {
3679                                my ($n,$v) = ($1,$2);
3680                                for ($n, $v) {
3681                                        s/^\s+//; s/\s+$//;
3682                                }
3683                                $self->change_file_prop($fbat, $n, $v);
3684                        }
3685                }
3686        }
3687}
3688
3689sub apply_autoprops {
3690        my ($self, $file, $fbat) = @_;
3691        my $conf_t = ${$self->{config}}{'config'};
3692        no warnings 'once';
3693        # Check [miscellany]/enable-auto-props in svn configuration.
3694        if (SVN::_Core::svn_config_get_bool(
3695                $conf_t,
3696                $SVN::_Core::SVN_CONFIG_SECTION_MISCELLANY,
3697                $SVN::_Core::SVN_CONFIG_OPTION_ENABLE_AUTO_PROPS,
3698                0)) {
3699                # Auto-props are enabled.  Enumerate them to look for matches.
3700                my $callback = sub {
3701                        $self->check_autoprop($_[0], $_[1], $file, $fbat);
3702                };
3703                SVN::_Core::svn_config_enumerate(
3704                        $conf_t,
3705                        $SVN::_Core::SVN_CONFIG_SECTION_AUTO_PROPS,
3706                        $callback);
3707        }
3708}
3709
3710sub A {
3711        my ($self, $m) = @_;
3712        my ($dir, $file) = split_path($m->{file_b});
3713        my $pbat = $self->ensure_path($dir);
3714        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3715                                        undef, -1);
3716        print "\tA\t$m->{file_b}\n" unless $::_q;
3717        $self->apply_autoprops($file, $fbat);
3718        $self->chg_file($fbat, $m);
3719        $self->close_file($fbat,undef,$self->{pool});
3720}
3721
3722sub C {
3723        my ($self, $m) = @_;
3724        my ($dir, $file) = split_path($m->{file_b});
3725        my $pbat = $self->ensure_path($dir);
3726        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3727                                $self->url_path($m->{file_a}), $self->{r});
3728        print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
3729        $self->chg_file($fbat, $m);
3730        $self->close_file($fbat,undef,$self->{pool});
3731}
3732
3733sub delete_entry {
3734        my ($self, $path, $pbat) = @_;
3735        my $rpath = $self->repo_path($path);
3736        my ($dir, $file) = split_path($rpath);
3737        $self->{rm}->{$dir} = 1;
3738        $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
3739}
3740
3741sub R {
3742        my ($self, $m) = @_;
3743        my ($dir, $file) = split_path($m->{file_b});
3744        my $pbat = $self->ensure_path($dir);
3745        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3746                                $self->url_path($m->{file_a}), $self->{r});
3747        print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
3748        $self->apply_autoprops($file, $fbat);
3749        $self->chg_file($fbat, $m);
3750        $self->close_file($fbat,undef,$self->{pool});
3751
3752        ($dir, $file) = split_path($m->{file_a});
3753        $pbat = $self->ensure_path($dir);
3754        $self->delete_entry($m->{file_a}, $pbat);
3755}
3756
3757sub M {
3758        my ($self, $m) = @_;
3759        my ($dir, $file) = split_path($m->{file_b});
3760        my $pbat = $self->ensure_path($dir);
3761        my $fbat = $self->open_file($self->repo_path($m->{file_b}),
3762                                $pbat,$self->{r},$self->{pool});
3763        print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
3764        $self->chg_file($fbat, $m);
3765        $self->close_file($fbat,undef,$self->{pool});
3766}
3767
3768sub T { shift->M(@_) }
3769
3770sub change_file_prop {
3771        my ($self, $fbat, $pname, $pval) = @_;
3772        $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
3773}
3774
3775sub _chg_file_get_blob ($$$$) {
3776        my ($self, $fbat, $m, $which) = @_;
3777        my $fh = $::_repository->temp_acquire("git_blob_$which");
3778        if ($m->{"mode_$which"} =~ /^120/) {
3779                print $fh 'link ' or croak $!;
3780                $self->change_file_prop($fbat,'svn:special','*');
3781        } elsif ($m->{mode_a} =~ /^120/ && $m->{"mode_$which"} !~ /^120/) {
3782                $self->change_file_prop($fbat,'svn:special',undef);
3783        }
3784        my $blob = $m->{"sha1_$which"};
3785        return ($fh,) if ($blob =~ /^0{40}$/);
3786        my $size = $::_repository->cat_blob($blob, $fh);
3787        croak "Failed to read object $blob" if ($size < 0);
3788        $fh->flush == 0 or croak $!;
3789        seek $fh, 0, 0 or croak $!;
3790
3791        my $exp = ::md5sum($fh);
3792        seek $fh, 0, 0 or croak $!;
3793        return ($fh, $exp);
3794}
3795
3796sub chg_file {
3797        my ($self, $fbat, $m) = @_;
3798        if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
3799                $self->change_file_prop($fbat,'svn:executable','*');
3800        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
3801                $self->change_file_prop($fbat,'svn:executable',undef);
3802        }
3803        my ($fh_a, $exp_a) = _chg_file_get_blob $self, $fbat, $m, 'a';
3804        my ($fh_b, $exp_b) = _chg_file_get_blob $self, $fbat, $m, 'b';
3805        my $pool = SVN::Pool->new;
3806        my $atd = $self->apply_textdelta($fbat, $exp_a, $pool);
3807        if (-s $fh_a) {
3808                my $txstream = SVN::TxDelta::new ($fh_a, $fh_b, $pool);
3809                my $res = SVN::TxDelta::send_txstream($txstream, @$atd, $pool);
3810                if (defined $res) {
3811                        die "Unexpected result from send_txstream: $res\n",
3812                            "(SVN::Core::VERSION: $SVN::Core::VERSION)\n";
3813                }
3814        } else {
3815                my $got = SVN::TxDelta::send_stream($fh_b, @$atd, $pool);
3816                die "Checksum mismatch\nexpected: $exp_b\ngot: $got\n"
3817                    if ($got ne $exp_b);
3818        }
3819        Git::temp_release($fh_b, 1);
3820        Git::temp_release($fh_a, 1);
3821        $pool->clear;
3822}
3823
3824sub D {
3825        my ($self, $m) = @_;
3826        my ($dir, $file) = split_path($m->{file_b});
3827        my $pbat = $self->ensure_path($dir);
3828        print "\tD\t$m->{file_b}\n" unless $::_q;
3829        $self->delete_entry($m->{file_b}, $pbat);
3830}
3831
3832sub close_edit {
3833        my ($self) = @_;
3834        my ($p,$bat) = ($self->{pool}, $self->{bat});
3835        foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
3836                next if $_ eq '';
3837                $self->close_directory($bat->{$_}, $p);
3838        }
3839        $self->close_directory($bat->{''}, $p);
3840        $self->SUPER::close_edit($p);
3841        $p->clear;
3842}
3843
3844sub abort_edit {
3845        my ($self) = @_;
3846        $self->SUPER::abort_edit($self->{pool});
3847}
3848
3849sub DESTROY {
3850        my $self = shift;
3851        $self->SUPER::DESTROY(@_);
3852        $self->{pool}->clear;
3853}
3854
3855# this drives the editor
3856sub apply_diff {
3857        my ($self) = @_;
3858        my $mods = $self->{mods};
3859        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
3860        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
3861                my $f = $m->{chg};
3862                if (defined $o{$f}) {
3863                        $self->$f($m);
3864                } else {
3865                        fatal("Invalid change type: $f");
3866                }
3867        }
3868        $self->rmdirs if $_rmdir;
3869        if (@$mods == 0) {
3870                $self->abort_edit;
3871        } else {
3872                $self->close_edit;
3873        }
3874        return scalar @$mods;
3875}
3876
3877package Git::SVN::Ra;
3878use vars qw/@ISA $config_dir $_log_window_size/;
3879use strict;
3880use warnings;
3881my ($ra_invalid, $can_do_switch, %ignored_err, $RA);
3882
3883BEGIN {
3884        # enforce temporary pool usage for some simple functions
3885        no strict 'refs';
3886        for my $f (qw/rev_proplist get_latest_revnum get_uuid get_repos_root/) {
3887                my $SUPER = "SUPER::$f";
3888                *$f = sub {
3889                        my $self = shift;
3890                        my $pool = SVN::Pool->new;
3891                        my @ret = $self->$SUPER(@_,$pool);
3892                        $pool->clear;
3893                        wantarray ? @ret : $ret[0];
3894                };
3895        }
3896}
3897
3898sub _auth_providers () {
3899        [
3900          SVN::Client::get_simple_provider(),
3901          SVN::Client::get_ssl_server_trust_file_provider(),
3902          SVN::Client::get_simple_prompt_provider(
3903            \&Git::SVN::Prompt::simple, 2),
3904          SVN::Client::get_ssl_client_cert_file_provider(),
3905          SVN::Client::get_ssl_client_cert_prompt_provider(
3906            \&Git::SVN::Prompt::ssl_client_cert, 2),
3907          SVN::Client::get_ssl_client_cert_pw_file_provider(),
3908          SVN::Client::get_ssl_client_cert_pw_prompt_provider(
3909            \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
3910          SVN::Client::get_username_provider(),
3911          SVN::Client::get_ssl_server_trust_prompt_provider(
3912            \&Git::SVN::Prompt::ssl_server_trust),
3913          SVN::Client::get_username_prompt_provider(
3914            \&Git::SVN::Prompt::username, 2)
3915        ]
3916}
3917
3918sub escape_uri_only {
3919        my ($uri) = @_;
3920        my @tmp;
3921        foreach (split m{/}, $uri) {
3922                s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
3923                push @tmp, $_;
3924        }
3925        join('/', @tmp);
3926}
3927
3928sub escape_url {
3929        my ($url) = @_;
3930        if ($url =~ m#^(https?)://([^/]+)(.*)$#) {
3931                my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
3932                $url = "$scheme://$domain$uri";
3933        }
3934        $url;
3935}
3936
3937sub new {
3938        my ($class, $url) = @_;
3939        $url =~ s!/+$!!;
3940        return $RA if ($RA && $RA->{url} eq $url);
3941
3942        SVN::_Core::svn_config_ensure($config_dir, undef);
3943        my ($baton, $callbacks) = SVN::Core::auth_open_helper(_auth_providers);
3944        my $config = SVN::Core::config_get_config($config_dir);
3945        $RA = undef;
3946        my $dont_store_passwords = 1;
3947        my $conf_t = ${$config}{'config'};
3948        {
3949                no warnings 'once';
3950                # The usage of $SVN::_Core::SVN_CONFIG_* variables
3951                # produces warnings that variables are used only once.
3952                # I had not found the better way to shut them up, so
3953                # the warnings of type 'once' are disabled in this block.
3954                if (SVN::_Core::svn_config_get_bool($conf_t,
3955                    $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
3956                    $SVN::_Core::SVN_CONFIG_OPTION_STORE_PASSWORDS,
3957                    1) == 0) {
3958                        SVN::_Core::svn_auth_set_parameter($baton,
3959                            $SVN::_Core::SVN_AUTH_PARAM_DONT_STORE_PASSWORDS,
3960                            bless (\$dont_store_passwords, "_p_void"));
3961                }
3962                if (SVN::_Core::svn_config_get_bool($conf_t,
3963                    $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
3964                    $SVN::_Core::SVN_CONFIG_OPTION_STORE_AUTH_CREDS,
3965                    1) == 0) {
3966                        $Git::SVN::Prompt::_no_auth_cache = 1;
3967                }
3968        } # no warnings 'once'
3969        my $self = SVN::Ra->new(url => escape_url($url), auth => $baton,
3970                              config => $config,
3971                              pool => SVN::Pool->new,
3972                              auth_provider_callbacks => $callbacks);
3973        $self->{url} = $url;
3974        $self->{svn_path} = $url;
3975        $self->{repos_root} = $self->get_repos_root;
3976        $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
3977        $self->{cache} = { check_path => { r => 0, data => {} },
3978                           get_dir => { r => 0, data => {} } };
3979        $RA = bless $self, $class;
3980}
3981
3982sub check_path {
3983        my ($self, $path, $r) = @_;
3984        my $cache = $self->{cache}->{check_path};
3985        if ($r == $cache->{r} && exists $cache->{data}->{$path}) {
3986                return $cache->{data}->{$path};
3987        }
3988        my $pool = SVN::Pool->new;
3989        my $t = $self->SUPER::check_path($path, $r, $pool);
3990        $pool->clear;
3991        if ($r != $cache->{r}) {
3992                %{$cache->{data}} = ();
3993                $cache->{r} = $r;
3994        }
3995        $cache->{data}->{$path} = $t;
3996}
3997
3998sub get_dir {
3999        my ($self, $dir, $r) = @_;
4000        my $cache = $self->{cache}->{get_dir};
4001        if ($r == $cache->{r}) {
4002                if (my $x = $cache->{data}->{$dir}) {
4003                        return wantarray ? @$x : $x->[0];
4004                }
4005        }
4006        my $pool = SVN::Pool->new;
4007        my ($d, undef, $props) = $self->SUPER::get_dir($dir, $r, $pool);
4008        my %dirents = map { $_ => { kind => $d->{$_}->kind } } keys %$d;
4009        $pool->clear;
4010        if ($r != $cache->{r}) {
4011                %{$cache->{data}} = ();
4012                $cache->{r} = $r;
4013        }
4014        $cache->{data}->{$dir} = [ \%dirents, $r, $props ];
4015        wantarray ? (\%dirents, $r, $props) : \%dirents;
4016}
4017
4018sub DESTROY {
4019        # do not call the real DESTROY since we store ourselves in $RA
4020}
4021
4022sub get_log {
4023        my ($self, @args) = @_;
4024        my $pool = SVN::Pool->new;
4025        splice(@args, 3, 1) if ($SVN::Core::VERSION le '1.2.0');
4026        my $ret = $self->SUPER::get_log(@args, $pool);
4027        $pool->clear;
4028        $ret;
4029}
4030
4031sub trees_match {
4032        my ($self, $url1, $rev1, $url2, $rev2) = @_;
4033        my $ctx = SVN::Client->new(auth => _auth_providers);
4034        my $out = IO::File->new_tmpfile;
4035
4036        # older SVN (1.1.x) doesn't take $pool as the last parameter for
4037        # $ctx->diff(), so we'll create a default one
4038        my $pool = SVN::Pool->new_default_sub;
4039
4040        $ra_invalid = 1; # this will open a new SVN::Ra connection to $url1
4041        $ctx->diff([], $url1, $rev1, $url2, $rev2, 1, 1, 0, $out, $out);
4042        $out->flush;
4043        my $ret = (($out->stat)[7] == 0);
4044        close $out or croak $!;
4045
4046        $ret;
4047}
4048
4049sub get_commit_editor {
4050        my ($self, $log, $cb, $pool) = @_;
4051        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
4052        $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
4053}
4054
4055sub gs_do_update {
4056        my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
4057        my $new = ($rev_a == $rev_b);
4058        my $path = $gs->{path};
4059
4060        if ($new && -e $gs->{index}) {
4061                unlink $gs->{index} or die
4062                  "Couldn't unlink index: $gs->{index}: $!\n";
4063        }
4064        my $pool = SVN::Pool->new;
4065        $editor->set_path_strip($path);
4066        my (@pc) = split m#/#, $path;
4067        my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
4068                                        1, $editor, $pool);
4069        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
4070
4071        # Since we can't rely on svn_ra_reparent being available, we'll
4072        # just have to do some magic with set_path to make it so
4073        # we only want a partial path.
4074        my $sp = '';
4075        my $final = join('/', @pc);
4076        while (@pc) {
4077                $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
4078                $sp .= '/' if length $sp;
4079                $sp .= shift @pc;
4080        }
4081        die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
4082
4083        $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
4084
4085        $reporter->finish_report($pool);
4086        $pool->clear;
4087        $editor->{git_commit_ok};
4088}
4089
4090# this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
4091# svn_ra_reparent didn't work before 1.4)
4092sub gs_do_switch {
4093        my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
4094        my $path = $gs->{path};
4095        my $pool = SVN::Pool->new;
4096
4097        my $full_url = $self->{url};
4098        my $old_url = $full_url;
4099        $full_url .= '/' . escape_uri_only($path) if length $path;
4100        my ($ra, $reparented);
4101
4102        if ($old_url =~ m#^svn(\+ssh)?://#) {
4103                $_[0] = undef;
4104                $self = undef;
4105                $RA = undef;
4106                $ra = Git::SVN::Ra->new($full_url);
4107                $ra_invalid = 1;
4108        } elsif ($old_url ne $full_url) {
4109                SVN::_Ra::svn_ra_reparent($self->{session}, $full_url, $pool);
4110                $self->{url} = $full_url;
4111                $reparented = 1;
4112        }
4113
4114        $ra ||= $self;
4115        $url_b = escape_url($url_b);
4116        my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
4117        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
4118        $reporter->set_path('', $rev_a, 0, @lock, $pool);
4119        $reporter->finish_report($pool);
4120
4121        if ($reparented) {
4122                SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
4123                $self->{url} = $old_url;
4124        }
4125
4126        $pool->clear;
4127        $editor->{git_commit_ok};
4128}
4129
4130sub longest_common_path {
4131        my ($gsv, $globs) = @_;
4132        my %common;
4133        my $common_max = scalar @$gsv;
4134
4135        foreach my $gs (@$gsv) {
4136                my @tmp = split m#/#, $gs->{path};
4137                my $p = '';
4138                foreach (@tmp) {
4139                        $p .= length($p) ? "/$_" : $_;
4140                        $common{$p} ||= 0;
4141                        $common{$p}++;
4142                }
4143        }
4144        $globs ||= [];
4145        $common_max += scalar @$globs;
4146        foreach my $glob (@$globs) {
4147                my @tmp = split m#/#, $glob->{path}->{left};
4148                my $p = '';
4149                foreach (@tmp) {
4150                        $p .= length($p) ? "/$_" : $_;
4151                        $common{$p} ||= 0;
4152                        $common{$p}++;
4153                }
4154        }
4155
4156        my $longest_path = '';
4157        foreach (sort {length $b <=> length $a} keys %common) {
4158                if ($common{$_} == $common_max) {
4159                        $longest_path = $_;
4160                        last;
4161                }
4162        }
4163        $longest_path;
4164}
4165
4166sub gs_fetch_loop_common {
4167        my ($self, $base, $head, $gsv, $globs) = @_;
4168        return if ($base > $head);
4169        my $inc = $_log_window_size;
4170        my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
4171        my $longest_path = longest_common_path($gsv, $globs);
4172        my $ra_url = $self->{url};
4173        while (1) {
4174                my %revs;
4175                my $err;
4176                my $err_handler = $SVN::Error::handler;
4177                $SVN::Error::handler = sub {
4178                        ($err) = @_;
4179                        skip_unknown_revs($err);
4180                };
4181                sub _cb {
4182                        my ($paths, $r, $author, $date, $log) = @_;
4183                        [ dup_changed_paths($paths),
4184                          { author => $author, date => $date, log => $log } ];
4185                }
4186                $self->get_log([$longest_path], $min, $max, 0, 1, 1,
4187                               sub { $revs{$_[1]} = _cb(@_) });
4188                if ($err && $max >= $head) {
4189                        print STDERR "Path '$longest_path' ",
4190                                     "was probably deleted:\n",
4191                                     $err->expanded_message,
4192                                     "\nWill attempt to follow ",
4193                                     "revisions r$min .. r$max ",
4194                                     "committed before the deletion\n";
4195                        my $hi = $max;
4196                        while (--$hi >= $min) {
4197                                my $ok;
4198                                $self->get_log([$longest_path], $min, $hi,
4199                                               0, 1, 1, sub {
4200                                               $ok ||= $_[1];
4201                                               $revs{$_[1]} = _cb(@_) });
4202                                if ($ok) {
4203                                        print STDERR "r$min .. r$ok OK\n";
4204                                        last;
4205                                }
4206                        }
4207                }
4208                $SVN::Error::handler = $err_handler;
4209
4210                my %exists = map { $_->{path} => $_ } @$gsv;
4211                foreach my $r (sort {$a <=> $b} keys %revs) {
4212                        my ($paths, $logged) = @{$revs{$r}};
4213
4214                        foreach my $gs ($self->match_globs(\%exists, $paths,
4215                                                           $globs, $r)) {
4216                                if ($gs->rev_map_max >= $r) {
4217                                        next;
4218                                }
4219                                next unless $gs->match_paths($paths, $r);
4220                                $gs->{logged_rev_props} = $logged;
4221                                if (my $last_commit = $gs->last_commit) {
4222                                        $gs->assert_index_clean($last_commit);
4223                                }
4224                                my $log_entry = $gs->do_fetch($paths, $r);
4225                                if ($log_entry) {
4226                                        $gs->do_git_commit($log_entry);
4227                                }
4228                                $INDEX_FILES{$gs->{index}} = 1;
4229                        }
4230                        foreach my $g (@$globs) {
4231                                my $k = "svn-remote.$g->{remote}." .
4232                                        "$g->{t}-maxRev";
4233                                Git::SVN::tmp_config($k, $r);
4234                        }
4235                        if ($ra_invalid) {
4236                                $_[0] = undef;
4237                                $self = undef;
4238                                $RA = undef;
4239                                $self = Git::SVN::Ra->new($ra_url);
4240                                $ra_invalid = undef;
4241                        }
4242                }
4243                # pre-fill the .rev_db since it'll eventually get filled in
4244                # with '0' x40 if something new gets committed
4245                foreach my $gs (@$gsv) {
4246                        next if $gs->rev_map_max >= $max;
4247                        next if defined $gs->rev_map_get($max);
4248                        $gs->rev_map_set($max, 0 x40);
4249                }
4250                foreach my $g (@$globs) {
4251                        my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
4252                        Git::SVN::tmp_config($k, $max);
4253                }
4254                last if $max >= $head;
4255                $min = $max + 1;
4256                $max += $inc;
4257                $max = $head if ($max > $head);
4258        }
4259        Git::SVN::gc();
4260}
4261
4262sub get_dir_globbed {
4263        my ($self, $left, $depth, $r) = @_;
4264
4265        my @x = eval { $self->get_dir($left, $r) };
4266        return unless scalar @x == 3;
4267        my $dirents = $x[0];
4268        my @finalents;
4269        foreach my $de (keys %$dirents) {
4270                next if $dirents->{$de}->{kind} != $SVN::Node::dir;
4271                if ($depth > 1) {
4272                        my @args = ("$left/$de", $depth - 1, $r);
4273                        foreach my $dir ($self->get_dir_globbed(@args)) {
4274                                push @finalents, "$de/$dir";
4275                        }
4276                } else {
4277                        push @finalents, $de;
4278                }
4279        }
4280        @finalents;
4281}
4282
4283sub match_globs {
4284        my ($self, $exists, $paths, $globs, $r) = @_;
4285
4286        sub get_dir_check {
4287                my ($self, $exists, $g, $r) = @_;
4288
4289                my @dirs = $self->get_dir_globbed($g->{path}->{left},
4290                                                  $g->{path}->{depth},
4291                                                  $r);
4292
4293                foreach my $de (@dirs) {
4294                        my $p = $g->{path}->full_path($de);
4295                        next if $exists->{$p};
4296                        next if (length $g->{path}->{right} &&
4297                                 ($self->check_path($p, $r) !=
4298                                  $SVN::Node::dir));
4299                        $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
4300                                         $g->{ref}->full_path($de), 1);
4301                }
4302        }
4303        foreach my $g (@$globs) {
4304                if (my $path = $paths->{"/$g->{path}->{left}"}) {
4305                        if ($path->{action} =~ /^[AR]$/) {
4306                                get_dir_check($self, $exists, $g, $r);
4307                        }
4308                }
4309                foreach (keys %$paths) {
4310                        if (/$g->{path}->{left_regex}/ &&
4311                            !/$g->{path}->{regex}/) {
4312                                next if $paths->{$_}->{action} !~ /^[AR]$/;
4313                                get_dir_check($self, $exists, $g, $r);
4314                        }
4315                        next unless /$g->{path}->{regex}/;
4316                        my $p = $1;
4317                        my $pathname = $g->{path}->full_path($p);
4318                        next if $exists->{$pathname};
4319                        next if ($self->check_path($pathname, $r) !=
4320                                 $SVN::Node::dir);
4321                        $exists->{$pathname} = Git::SVN->init(
4322                                              $self->{url}, $pathname, undef,
4323                                              $g->{ref}->full_path($p), 1);
4324                }
4325                my $c = '';
4326                foreach (split m#/#, $g->{path}->{left}) {
4327                        $c .= "/$_";
4328                        next unless ($paths->{$c} &&
4329                                     ($paths->{$c}->{action} =~ /^[AR]$/));
4330                        get_dir_check($self, $exists, $g, $r);
4331                }
4332        }
4333        values %$exists;
4334}
4335
4336sub minimize_url {
4337        my ($self) = @_;
4338        return $self->{url} if ($self->{url} eq $self->{repos_root});
4339        my $url = $self->{repos_root};
4340        my @components = split(m!/!, $self->{svn_path});
4341        my $c = '';
4342        do {
4343                $url .= "/$c" if length $c;
4344                eval { (ref $self)->new($url)->get_latest_revnum };
4345        } while ($@ && ($c = shift @components));
4346        $url;
4347}
4348
4349sub can_do_switch {
4350        my $self = shift;
4351        unless (defined $can_do_switch) {
4352                my $pool = SVN::Pool->new;
4353                my $rep = eval {
4354                        $self->do_switch(1, '', 0, $self->{url},
4355                                         SVN::Delta::Editor->new, $pool);
4356                };
4357                if ($@) {
4358                        $can_do_switch = 0;
4359                } else {
4360                        $rep->abort_report($pool);
4361                        $can_do_switch = 1;
4362                }
4363                $pool->clear;
4364        }
4365        $can_do_switch;
4366}
4367
4368sub skip_unknown_revs {
4369        my ($err) = @_;
4370        my $errno = $err->apr_err();
4371        # Maybe the branch we're tracking didn't
4372        # exist when the repo started, so it's
4373        # not an error if it doesn't, just continue
4374        #
4375        # Wonderfully consistent library, eh?
4376        # 160013 - svn:// and file://
4377        # 175002 - http(s)://
4378        # 175007 - http(s):// (this repo required authorization, too...)
4379        #   More codes may be discovered later...
4380        if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
4381                my $err_key = $err->expanded_message;
4382                # revision numbers change every time, filter them out
4383                $err_key =~ s/\d+/\0/g;
4384                $err_key = "$errno\0$err_key";
4385                unless ($ignored_err{$err_key}) {
4386                        warn "W: Ignoring error from SVN, path probably ",
4387                             "does not exist: ($errno): ",
4388                             $err->expanded_message,"\n";
4389                        warn "W: Do not be alarmed at the above message ",
4390                             "git-svn is just searching aggressively for ",
4391                             "old history.\n",
4392                             "This may take a while on large repositories\n";
4393                        $ignored_err{$err_key} = 1;
4394                }
4395                return;
4396        }
4397        die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
4398}
4399
4400# svn_log_changed_path_t objects passed to get_log are likely to be
4401# overwritten even if only the refs are copied to an external variable,
4402# so we should dup the structures in their entirety.  Using an externally
4403# passed pool (instead of our temporary and quickly cleared pool in
4404# Git::SVN::Ra) does not help matters at all...
4405sub dup_changed_paths {
4406        my ($paths) = @_;
4407        return undef unless $paths;
4408        my %ret;
4409        foreach my $p (keys %$paths) {
4410                my $i = $paths->{$p};
4411                my %s = map { $_ => $i->$_ }
4412                              qw/copyfrom_path copyfrom_rev action/;
4413                $ret{$p} = \%s;
4414        }
4415        \%ret;
4416}
4417
4418package Git::SVN::Log;
4419use strict;
4420use warnings;
4421use POSIX qw/strftime/;
4422use constant commit_log_separator => ('-' x 72) . "\n";
4423use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
4424            %rusers $show_commit $incremental/;
4425my $l_fmt;
4426
4427sub cmt_showable {
4428        my ($c) = @_;
4429        return 1 if defined $c->{r};
4430
4431        # big commit message got truncated by the 16k pretty buffer in rev-list
4432        if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
4433                                $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
4434                @{$c->{l}} = ();
4435                my @log = command(qw/cat-file commit/, $c->{c});
4436
4437                # shift off the headers
4438                shift @log while ($log[0] ne '');
4439                shift @log;
4440
4441                # TODO: make $c->{l} not have a trailing newline in the future
4442                @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
4443
4444                (undef, $c->{r}, undef) = ::extract_metadata(
4445                                (grep(/^git-svn-id: /, @log))[-1]);
4446        }
4447        return defined $c->{r};
4448}
4449
4450sub log_use_color {
4451        return $color || Git->repository->get_colorbool('color.diff');
4452}
4453
4454sub git_svn_log_cmd {
4455        my ($r_min, $r_max, @args) = @_;
4456        my $head = 'HEAD';
4457        my (@files, @log_opts);
4458        foreach my $x (@args) {
4459                if ($x eq '--' || @files) {
4460                        push @files, $x;
4461                } else {
4462                        if (::verify_ref("$x^0")) {
4463                                $head = $x;
4464                        } else {
4465                                push @log_opts, $x;
4466                        }
4467                }
4468        }
4469
4470        my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
4471        $gs ||= Git::SVN->_new;
4472        my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
4473                   $gs->refname);
4474        push @cmd, '-r' unless $non_recursive;
4475        push @cmd, qw/--raw --name-status/ if $verbose;
4476        push @cmd, '--color' if log_use_color();
4477        push @cmd, @log_opts;
4478        if (defined $r_max && $r_max == $r_min) {
4479                push @cmd, '--max-count=1';
4480                if (my $c = $gs->rev_map_get($r_max)) {
4481                        push @cmd, $c;
4482                }
4483        } elsif (defined $r_max) {
4484                if ($r_max < $r_min) {
4485                        ($r_min, $r_max) = ($r_max, $r_min);
4486                }
4487                my (undef, $c_max) = $gs->find_rev_before($r_max, 1, $r_min);
4488                my (undef, $c_min) = $gs->find_rev_after($r_min, 1, $r_max);
4489                # If there are no commits in the range, both $c_max and $c_min
4490                # will be undefined.  If there is at least 1 commit in the
4491                # range, both will be defined.
4492                return () if !defined $c_min || !defined $c_max;
4493                if ($c_min eq $c_max) {
4494                        push @cmd, '--max-count=1', $c_min;
4495                } else {
4496                        push @cmd, '--boundary', "$c_min..$c_max";
4497                }
4498        }
4499        return (@cmd, @files);
4500}
4501
4502# adapted from pager.c
4503sub config_pager {
4504        $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
4505        if (!defined $pager) {
4506                $pager = 'less';
4507        } elsif (length $pager == 0 || $pager eq 'cat') {
4508                $pager = undef;
4509        }
4510        $ENV{GIT_PAGER_IN_USE} = defined($pager);
4511}
4512
4513sub run_pager {
4514        return unless -t *STDOUT && defined $pager;
4515        pipe my ($rfd, $wfd) or return;
4516        defined(my $pid = fork) or ::fatal "Can't fork: $!";
4517        if (!$pid) {
4518                open STDOUT, '>&', $wfd or
4519                                     ::fatal "Can't redirect to stdout: $!";
4520                return;
4521        }
4522        open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!";
4523        $ENV{LESS} ||= 'FRSX';
4524        exec $pager or ::fatal "Can't run pager: $! ($pager)";
4525}
4526
4527sub format_svn_date {
4528        return strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)", localtime(shift));
4529}
4530
4531sub parse_git_date {
4532        my ($t, $tz) = @_;
4533        # Date::Parse isn't in the standard Perl distro :(
4534        if ($tz =~ s/^\+//) {
4535                $t += tz_to_s_offset($tz);
4536        } elsif ($tz =~ s/^\-//) {
4537                $t -= tz_to_s_offset($tz);
4538        }
4539        return $t;
4540}
4541
4542sub set_local_timezone {
4543        if (defined $TZ) {
4544                $ENV{TZ} = $TZ;
4545        } else {
4546                delete $ENV{TZ};
4547        }
4548}
4549
4550sub tz_to_s_offset {
4551        my ($tz) = @_;
4552        $tz =~ s/(\d\d)$//;
4553        return ($1 * 60) + ($tz * 3600);
4554}
4555
4556sub get_author_info {
4557        my ($dest, $author, $t, $tz) = @_;
4558        $author =~ s/(?:^\s*|\s*$)//g;
4559        $dest->{a_raw} = $author;
4560        my $au;
4561        if ($::_authors) {
4562                $au = $rusers{$author} || undef;
4563        }
4564        if (!$au) {
4565                ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
4566        }
4567        $dest->{t} = $t;
4568        $dest->{tz} = $tz;
4569        $dest->{a} = $au;
4570        $dest->{t_utc} = parse_git_date($t, $tz);
4571}
4572
4573sub process_commit {
4574        my ($c, $r_min, $r_max, $defer) = @_;
4575        if (defined $r_min && defined $r_max) {
4576                if ($r_min == $c->{r} && $r_min == $r_max) {
4577                        show_commit($c);
4578                        return 0;
4579                }
4580                return 1 if $r_min == $r_max;
4581                if ($r_min < $r_max) {
4582                        # we need to reverse the print order
4583                        return 0 if (defined $limit && --$limit < 0);
4584                        push @$defer, $c;
4585                        return 1;
4586                }
4587                if ($r_min != $r_max) {
4588                        return 1 if ($r_min < $c->{r});
4589                        return 1 if ($r_max > $c->{r});
4590                }
4591        }
4592        return 0 if (defined $limit && --$limit < 0);
4593        show_commit($c);
4594        return 1;
4595}
4596
4597sub show_commit {
4598        my $c = shift;
4599        if ($oneline) {
4600                my $x = "\n";
4601                if (my $l = $c->{l}) {
4602                        while ($l->[0] =~ /^\s*$/) { shift @$l }
4603                        $x = $l->[0];
4604                }
4605                $l_fmt ||= 'A' . length($c->{r});
4606                print 'r',pack($l_fmt, $c->{r}),' | ';
4607                print "$c->{c} | " if $show_commit;
4608                print $x;
4609        } else {
4610                show_commit_normal($c);
4611        }
4612}
4613
4614sub show_commit_changed_paths {
4615        my ($c) = @_;
4616        return unless $c->{changed};
4617        print "Changed paths:\n", @{$c->{changed}};
4618}
4619
4620sub show_commit_normal {
4621        my ($c) = @_;
4622        print commit_log_separator, "r$c->{r} | ";
4623        print "$c->{c} | " if $show_commit;
4624        print "$c->{a} | ", format_svn_date($c->{t_utc}), ' | ';
4625        my $nr_line = 0;
4626
4627        if (my $l = $c->{l}) {
4628                while ($l->[$#$l] eq "\n" && $#$l > 0
4629                                          && $l->[($#$l - 1)] eq "\n") {
4630                        pop @$l;
4631                }
4632                $nr_line = scalar @$l;
4633                if (!$nr_line) {
4634                        print "1 line\n\n\n";
4635                } else {
4636                        if ($nr_line == 1) {
4637                                $nr_line = '1 line';
4638                        } else {
4639                                $nr_line .= ' lines';
4640                        }
4641                        print $nr_line, "\n";
4642                        show_commit_changed_paths($c);
4643                        print "\n";
4644                        print $_ foreach @$l;
4645                }
4646        } else {
4647                print "1 line\n";
4648                show_commit_changed_paths($c);
4649                print "\n";
4650
4651        }
4652        foreach my $x (qw/raw stat diff/) {
4653                if ($c->{$x}) {
4654                        print "\n";
4655                        print $_ foreach @{$c->{$x}}
4656                }
4657        }
4658}
4659
4660sub cmd_show_log {
4661        my (@args) = @_;
4662        my ($r_min, $r_max);
4663        my $r_last = -1; # prevent dupes
4664        set_local_timezone();
4665        if (defined $::_revision) {
4666                if ($::_revision =~ /^(\d+):(\d+)$/) {
4667                        ($r_min, $r_max) = ($1, $2);
4668                } elsif ($::_revision =~ /^\d+$/) {
4669                        $r_min = $r_max = $::_revision;
4670                } else {
4671                        ::fatal "-r$::_revision is not supported, use ",
4672                                "standard 'git log' arguments instead";
4673                }
4674        }
4675
4676        config_pager();
4677        @args = git_svn_log_cmd($r_min, $r_max, @args);
4678        if (!@args) {
4679                print commit_log_separator unless $incremental || $oneline;
4680                return;
4681        }
4682        my $log = command_output_pipe(@args);
4683        run_pager();
4684        my (@k, $c, $d, $stat);
4685        my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
4686        while (<$log>) {
4687                if (/^${esc_color}commit -?($::sha1_short)/o) {
4688                        my $cmt = $1;
4689                        if ($c && cmt_showable($c) && $c->{r} != $r_last) {
4690                                $r_last = $c->{r};
4691                                process_commit($c, $r_min, $r_max, \@k) or
4692                                                                goto out;
4693                        }
4694                        $d = undef;
4695                        $c = { c => $cmt };
4696                } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
4697                        get_author_info($c, $1, $2, $3);
4698                } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
4699                        # ignore
4700                } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
4701                        push @{$c->{raw}}, $_;
4702                } elsif (/^${esc_color}[ACRMDT]\t/) {
4703                        # we could add $SVN->{svn_path} here, but that requires
4704                        # remote access at the moment (repo_path_split)...
4705                        s#^(${esc_color})([ACRMDT])\t#$1   $2 #o;
4706                        push @{$c->{changed}}, $_;
4707                } elsif (/^${esc_color}diff /o) {
4708                        $d = 1;
4709                        push @{$c->{diff}}, $_;
4710                } elsif ($d) {
4711                        push @{$c->{diff}}, $_;
4712                } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
4713                          $esc_color*[\+\-]*$esc_color$/x) {
4714                        $stat = 1;
4715                        push @{$c->{stat}}, $_;
4716                } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
4717                        push @{$c->{stat}}, $_;
4718                        $stat = undef;
4719                } elsif (/^${esc_color}    (git-svn-id:.+)$/o) {
4720                        ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
4721                } elsif (s/^${esc_color}    //o) {
4722                        push @{$c->{l}}, $_;
4723                }
4724        }
4725        if ($c && defined $c->{r} && $c->{r} != $r_last) {
4726                $r_last = $c->{r};
4727                process_commit($c, $r_min, $r_max, \@k);
4728        }
4729        if (@k) {
4730                ($r_min, $r_max) = ($r_max, $r_min);
4731                process_commit($_, $r_min, $r_max) foreach reverse @k;
4732        }
4733out:
4734        close $log;
4735        print commit_log_separator unless $incremental || $oneline;
4736}
4737
4738sub cmd_blame {
4739        my $path = pop;
4740
4741        config_pager();
4742        run_pager();
4743
4744        my ($fh, $ctx, $rev);
4745
4746        if ($_git_format) {
4747                ($fh, $ctx) = command_output_pipe('blame', @_, $path);
4748                while (my $line = <$fh>) {
4749                        if ($line =~ /^\^?([[:xdigit:]]+)\s/) {
4750                                # Uncommitted edits show up as a rev ID of
4751                                # all zeros, which we can't look up with
4752                                # cmt_metadata
4753                                if ($1 !~ /^0+$/) {
4754                                        (undef, $rev, undef) =
4755                                                ::cmt_metadata($1);
4756                                        $rev = '0' if (!$rev);
4757                                } else {
4758                                        $rev = '0';
4759                                }
4760                                $rev = sprintf('%-10s', $rev);
4761                                $line =~ s/^\^?[[:xdigit:]]+(\s)/$rev$1/;
4762                        }
4763                        print $line;
4764                }
4765        } else {
4766                ($fh, $ctx) = command_output_pipe('blame', '-p', @_, 'HEAD',
4767                                                  '--', $path);
4768                my ($sha1);
4769                my %authors;
4770                while (my $line = <$fh>) {
4771                        if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
4772                                $sha1 = $1;
4773                                (undef, $rev, undef) = ::cmt_metadata($1);
4774                                $rev = '0' if (!$rev);
4775                        }
4776                        elsif ($line =~ /^author (.*)/) {
4777                                $authors{$rev} = $1;
4778                                $authors{$rev} =~ s/\s/_/g;
4779                        }
4780                        elsif ($line =~ /^\t(.*)$/) {
4781                                printf("%6s %10s %s\n", $rev, $authors{$rev}, $1);
4782                        }
4783                }
4784        }
4785        command_close_pipe($fh, $ctx);
4786}
4787
4788package Git::SVN::Migration;
4789# these version numbers do NOT correspond to actual version numbers
4790# of git nor git-svn.  They are just relative.
4791#
4792# v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
4793#
4794# v1 layout: .git/$id/info/url, refs/remotes/$id
4795#
4796# v2 layout: .git/svn/$id/info/url, refs/remotes/$id
4797#
4798# v3 layout: .git/svn/$id, refs/remotes/$id
4799#            - info/url may remain for backwards compatibility
4800#            - this is what we migrate up to this layout automatically,
4801#            - this will be used by git svn init on single branches
4802# v3.1 layout (auto migrated):
4803#            - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
4804#              for backwards compatibility
4805#
4806# v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
4807#            - this is only created for newly multi-init-ed
4808#              repositories.  Similar in spirit to the
4809#              --use-separate-remotes option in git-clone (now default)
4810#            - we do not automatically migrate to this (following
4811#              the example set by core git)
4812#
4813# v5 layout: .rev_db.$UUID => .rev_map.$UUID
4814#            - newer, more-efficient format that uses 24-bytes per record
4815#              with no filler space.
4816#            - use xxd -c24 < .rev_map.$UUID to view and debug
4817#            - This is a one-way migration, repositories updated to the
4818#              new format will not be able to use old git-svn without
4819#              rebuilding the .rev_db.  Rebuilding the rev_db is not
4820#              possible if noMetadata or useSvmProps are set; but should
4821#              be no problem for users that use the (sensible) defaults.
4822use strict;
4823use warnings;
4824use Carp qw/croak/;
4825use File::Path qw/mkpath/;
4826use File::Basename qw/dirname basename/;
4827use vars qw/$_minimize/;
4828
4829sub migrate_from_v0 {
4830        my $git_dir = $ENV{GIT_DIR};
4831        return undef unless -d $git_dir;
4832        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
4833        my $migrated = 0;
4834        while (<$fh>) {
4835                chomp;
4836                my ($id, $orig_ref) = ($_, $_);
4837                next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
4838                next unless -f "$git_dir/$id/info/url";
4839                my $new_ref = "refs/remotes/$id";
4840                if (::verify_ref("$new_ref^0")) {
4841                        print STDERR "W: $orig_ref is probably an old ",
4842                                     "branch used by an ancient version of ",
4843                                     "git-svn.\n",
4844                                     "However, $new_ref also exists.\n",
4845                                     "We will not be able ",
4846                                     "to use this branch until this ",
4847                                     "ambiguity is resolved.\n";
4848                        next;
4849                }
4850                print STDERR "Migrating from v0 layout...\n" if !$migrated;
4851                print STDERR "Renaming ref: $orig_ref => $new_ref\n";
4852                command_noisy('update-ref', $new_ref, $orig_ref);
4853                command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
4854                $migrated++;
4855        }
4856        command_close_pipe($fh, $ctx);
4857        print STDERR "Done migrating from v0 layout...\n" if $migrated;
4858        $migrated;
4859}
4860
4861sub migrate_from_v1 {
4862        my $git_dir = $ENV{GIT_DIR};
4863        my $migrated = 0;
4864        return $migrated unless -d $git_dir;
4865        my $svn_dir = "$git_dir/svn";
4866
4867        # just in case somebody used 'svn' as their $id at some point...
4868        return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
4869
4870        print STDERR "Migrating from a git-svn v1 layout...\n";
4871        mkpath([$svn_dir]);
4872        print STDERR "Data from a previous version of git-svn exists, but\n\t",
4873                     "$svn_dir\n\t(required for this version ",
4874                     "($::VERSION) of git-svn) does not exist.\n";
4875        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
4876        while (<$fh>) {
4877                my $x = $_;
4878                next unless $x =~ s#^refs/remotes/##;
4879                chomp $x;
4880                next unless -f "$git_dir/$x/info/url";
4881                my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
4882                next unless $u;
4883                my $dn = dirname("$git_dir/svn/$x");
4884                mkpath([$dn]) unless -d $dn;
4885                if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
4886                        mkpath(["$git_dir/svn/svn"]);
4887                        print STDERR " - $git_dir/$x/info => ",
4888                                        "$git_dir/svn/$x/info\n";
4889                        rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
4890                               croak "$!: $x";
4891                        # don't worry too much about these, they probably
4892                        # don't exist with repos this old (save for index,
4893                        # and we can easily regenerate that)
4894                        foreach my $f (qw/unhandled.log index .rev_db/) {
4895                                rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
4896                        }
4897                } else {
4898                        print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
4899                        rename "$git_dir/$x", "$git_dir/svn/$x" or
4900                               croak "$!: $x";
4901                }
4902                $migrated++;
4903        }
4904        command_close_pipe($fh, $ctx);
4905        print STDERR "Done migrating from a git-svn v1 layout\n";
4906        $migrated;
4907}
4908
4909sub read_old_urls {
4910        my ($l_map, $pfx, $path) = @_;
4911        my @dir;
4912        foreach (<$path/*>) {
4913                if (-r "$_/info/url") {
4914                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
4915                        my $ref_id = $pfx . basename $_;
4916                        my $url = ::file_to_s("$_/info/url");
4917                        $l_map->{$ref_id} = $url;
4918                } elsif (-d $_) {
4919                        push @dir, $_;
4920                }
4921        }
4922        foreach (@dir) {
4923                my $x = $_;
4924                $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
4925                read_old_urls($l_map, $x, $_);
4926        }
4927}
4928
4929sub migrate_from_v2 {
4930        my @cfg = command(qw/config -l/);
4931        return if grep /^svn-remote\..+\.url=/, @cfg;
4932        my %l_map;
4933        read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
4934        my $migrated = 0;
4935
4936        foreach my $ref_id (sort keys %l_map) {
4937                eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
4938                if ($@) {
4939                        Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
4940                }
4941                $migrated++;
4942        }
4943        $migrated;
4944}
4945
4946sub minimize_connections {
4947        my $r = Git::SVN::read_all_remotes();
4948        my $new_urls = {};
4949        my $root_repos = {};
4950        foreach my $repo_id (keys %$r) {
4951                my $url = $r->{$repo_id}->{url} or next;
4952                my $fetch = $r->{$repo_id}->{fetch} or next;
4953                my $ra = Git::SVN::Ra->new($url);
4954
4955                # skip existing cases where we already connect to the root
4956                if (($ra->{url} eq $ra->{repos_root}) ||
4957                    ($ra->{repos_root} eq $repo_id)) {
4958                        $root_repos->{$ra->{url}} = $repo_id;
4959                        next;
4960                }
4961
4962                my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
4963                my $root_path = $ra->{url};
4964                $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
4965                foreach my $path (keys %$fetch) {
4966                        my $ref_id = $fetch->{$path};
4967                        my $gs = Git::SVN->new($ref_id, $repo_id, $path);
4968
4969                        # make sure we can read when connecting to
4970                        # a higher level of a repository
4971                        my ($last_rev, undef) = $gs->last_rev_commit;
4972                        if (!defined $last_rev) {
4973                                $last_rev = eval {
4974                                        $root_ra->get_latest_revnum;
4975                                };
4976                                next if $@;
4977                        }
4978                        my $new = $root_path;
4979                        $new .= length $path ? "/$path" : '';
4980                        eval {
4981                                $root_ra->get_log([$new], $last_rev, $last_rev,
4982                                                  0, 0, 1, sub { });
4983                        };
4984                        next if $@;
4985                        $new_urls->{$ra->{repos_root}}->{$new} =
4986                                { ref_id => $ref_id,
4987                                  old_repo_id => $repo_id,
4988                                  old_path => $path };
4989                }
4990        }
4991
4992        my @emptied;
4993        foreach my $url (keys %$new_urls) {
4994                # see if we can re-use an existing [svn-remote "repo_id"]
4995                # instead of creating a(n ugly) new section:
4996                my $repo_id = $root_repos->{$url} || $url;
4997
4998                my $fetch = $new_urls->{$url};
4999                foreach my $path (keys %$fetch) {
5000                        my $x = $fetch->{$path};
5001                        Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
5002                        my $pfx = "svn-remote.$x->{old_repo_id}";
5003
5004                        my $old_fetch = quotemeta("$x->{old_path}:".
5005                                                  "refs/remotes/$x->{ref_id}");
5006                        command_noisy(qw/config --unset/,
5007                                      "$pfx.fetch", '^'. $old_fetch . '$');
5008                        delete $r->{$x->{old_repo_id}}->
5009                               {fetch}->{$x->{old_path}};
5010                        if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
5011                                command_noisy(qw/config --unset/,
5012                                              "$pfx.url");
5013                                push @emptied, $x->{old_repo_id}
5014                        }
5015                }
5016        }
5017        if (@emptied) {
5018                my $file = $ENV{GIT_CONFIG} || $ENV{GIT_CONFIG_LOCAL} ||
5019                           "$ENV{GIT_DIR}/config";
5020                print STDERR <<EOF;
5021The following [svn-remote] sections in your config file ($file) are empty
5022and can be safely removed:
5023EOF
5024                print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
5025        }
5026}
5027
5028sub migration_check {
5029        migrate_from_v0();
5030        migrate_from_v1();
5031        migrate_from_v2();
5032        minimize_connections() if $_minimize;
5033}
5034
5035package Git::IndexInfo;
5036use strict;
5037use warnings;
5038use Git qw/command_input_pipe command_close_pipe/;
5039
5040sub new {
5041        my ($class) = @_;
5042        my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
5043        bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
5044}
5045
5046sub remove {
5047        my ($self, $path) = @_;
5048        if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
5049                return ++$self->{nr};
5050        }
5051        undef;
5052}
5053
5054sub update {
5055        my ($self, $mode, $hash, $path) = @_;
5056        if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
5057                return ++$self->{nr};
5058        }
5059        undef;
5060}
5061
5062sub DESTROY {
5063        my ($self) = @_;
5064        command_close_pipe($self->{gui}, $self->{ctx});
5065}
5066
5067package Git::SVN::GlobSpec;
5068use strict;
5069use warnings;
5070
5071sub new {
5072        my ($class, $glob) = @_;
5073        my $re = $glob;
5074        $re =~ s!/+$!!g; # no need for trailing slashes
5075        $re =~ m!^([^*]*)(\*(?:/\*)*)([^*]*)$!;
5076        my $temp = $re;
5077        my ($left, $right) = ($1, $3);
5078        $re = $2;
5079        my $depth = $re =~ tr/*/*/;
5080        if ($depth != $temp =~ tr/*/*/) {
5081                die "Only one set of wildcard directories " .
5082                        "(e.g. '*' or '*/*/*') is supported: '$glob'\n";
5083        }
5084        if ($depth == 0) {
5085                die "One '*' is needed for glob: '$glob'\n";
5086        }
5087        $re =~ s!\*!\[^/\]*!g;
5088        $re = quotemeta($left) . "($re)" . quotemeta($right);
5089        if (length $left && !($left =~ s!/+$!!g)) {
5090                die "Missing trailing '/' on left side of: '$glob' ($left)\n";
5091        }
5092        if (length $right && !($right =~ s!^/+!!g)) {
5093                die "Missing leading '/' on right side of: '$glob' ($right)\n";
5094        }
5095        my $left_re = qr/^\/\Q$left\E(\/|$)/;
5096        bless { left => $left, right => $right, left_regex => $left_re,
5097                regex => qr/$re/, glob => $glob, depth => $depth }, $class;
5098}
5099
5100sub full_path {
5101        my ($self, $path) = @_;
5102        return (length $self->{left} ? "$self->{left}/" : '') .
5103               $path . (length $self->{right} ? "/$self->{right}" : '');
5104}
5105
5106__END__
5107
5108Data structures:
5109
5110
5111$remotes = { # returned by read_all_remotes()
5112        'svn' => {
5113                # svn-remote.svn.url=https://svn.musicpd.org
5114                url => 'https://svn.musicpd.org',
5115                # svn-remote.svn.fetch=mpd/trunk:trunk
5116                fetch => {
5117                        'mpd/trunk' => 'trunk',
5118                },
5119                # svn-remote.svn.tags=mpd/tags/*:tags/*
5120                tags => {
5121                        path => {
5122                                left => 'mpd/tags',
5123                                right => '',
5124                                regex => qr!mpd/tags/([^/]+)$!,
5125                                glob => 'tags/*',
5126                        },
5127                        ref => {
5128                                left => 'tags',
5129                                right => '',
5130                                regex => qr!tags/([^/]+)$!,
5131                                glob => 'tags/*',
5132                        },
5133                }
5134        }
5135};
5136
5137$log_entry hashref as returned by libsvn_log_entry()
5138{
5139        log => 'whitespace-formatted log entry
5140',                                              # trailing newline is preserved
5141        revision => '8',                        # integer
5142        date => '2004-02-24T17:01:44.108345Z',  # commit date
5143        author => 'committer name'
5144};
5145
5146
5147# this is generated by generate_diff();
5148@mods = array of diff-index line hashes, each element represents one line
5149        of diff-index output
5150
5151diff-index line ($m hash)
5152{
5153        mode_a => first column of diff-index output, no leading ':',
5154        mode_b => second column of diff-index output,
5155        sha1_b => sha1sum of the final blob,
5156        chg => change type [MCRADT],
5157        file_a => original file name of a file (iff chg is 'C' or 'R')
5158        file_b => new/current file name of a file (any chg)
5159}
5160;
5161
5162# retval of read_url_paths{,_all}();
5163$l_map = {
5164        # repository root url
5165        'https://svn.musicpd.org' => {
5166                # repository path               # GIT_SVN_ID
5167                'mpd/trunk'             =>      'trunk',
5168                'mpd/tags/0.11.5'       =>      'tags/0.11.5',
5169        },
5170}
5171
5172Notes:
5173        I don't trust the each() function on unless I created %hash myself
5174        because the internal iterator may not have started at base.