git-svn.perlon commit builtin-clone: use strbuf in clone_local() and copy_or_link_directory() (b9e125e)
   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()->{svn};
 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        if (!defined $r0 || !defined $parent) {
2407                my ($base, $head) = parse_revision_argument(0, $r);
2408                if ($base <= $r) {
2409                        $gs->fetch($base, $r);
2410                }
2411                ($r0, $parent) = $gs->last_rev_commit;
2412        }
2413        if (defined $r0 && defined $parent) {
2414                print STDERR "Found branch parent: ($self->{ref_id}) $parent\n";
2415                my $ed;
2416                if ($self->ra->can_do_switch) {
2417                        $self->assert_index_clean($parent);
2418                        print STDERR "Following parent with do_switch\n";
2419                        # do_switch works with svn/trunk >= r22312, but that
2420                        # is not included with SVN 1.4.3 (the latest version
2421                        # at the moment), so we can't rely on it
2422                        $self->{last_commit} = $parent;
2423                        $ed = SVN::Git::Fetcher->new($self);
2424                        $gs->ra->gs_do_switch($r0, $rev, $gs,
2425                                              $self->full_url, $ed)
2426                          or die "SVN connection failed somewhere...\n";
2427                } elsif ($self->ra->trees_match($new_url, $r0,
2428                                                $self->full_url, $rev)) {
2429                        print STDERR "Trees match:\n",
2430                                     "  $new_url\@$r0\n",
2431                                     "  ${\$self->full_url}\@$rev\n",
2432                                     "Following parent with no changes\n";
2433                        $self->tmp_index_do(sub {
2434                            command_noisy('read-tree', $parent);
2435                        });
2436                        $self->{last_commit} = $parent;
2437                } else {
2438                        print STDERR "Following parent with do_update\n";
2439                        $ed = SVN::Git::Fetcher->new($self);
2440                        $self->ra->gs_do_update($rev, $rev, $self, $ed)
2441                          or die "SVN connection failed somewhere...\n";
2442                }
2443                print STDERR "Successfully followed parent\n";
2444                return $self->make_log_entry($rev, [$parent], $ed);
2445        }
2446        return undef;
2447}
2448
2449sub do_fetch {
2450        my ($self, $paths, $rev) = @_;
2451        my $ed;
2452        my ($last_rev, @parents);
2453        if (my $lc = $self->last_commit) {
2454                # we can have a branch that was deleted, then re-added
2455                # under the same name but copied from another path, in
2456                # which case we'll have multiple parents (we don't
2457                # want to break the original ref, nor lose copypath info):
2458                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2459                        push @{$log_entry->{parents}}, $lc;
2460                        return $log_entry;
2461                }
2462                $ed = SVN::Git::Fetcher->new($self);
2463                $last_rev = $self->{last_rev};
2464                $ed->{c} = $lc;
2465                @parents = ($lc);
2466        } else {
2467                $last_rev = $rev;
2468                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2469                        return $log_entry;
2470                }
2471                $ed = SVN::Git::Fetcher->new($self);
2472        }
2473        unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
2474                die "SVN connection failed somewhere...\n";
2475        }
2476        $self->make_log_entry($rev, \@parents, $ed);
2477}
2478
2479sub get_untracked {
2480        my ($self, $ed) = @_;
2481        my @out;
2482        my $h = $ed->{empty};
2483        foreach (sort keys %$h) {
2484                my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
2485                push @out, "  $act: " . uri_encode($_);
2486                warn "W: $act: $_\n";
2487        }
2488        foreach my $t (qw/dir_prop file_prop/) {
2489                $h = $ed->{$t} or next;
2490                foreach my $path (sort keys %$h) {
2491                        my $ppath = $path eq '' ? '.' : $path;
2492                        foreach my $prop (sort keys %{$h->{$path}}) {
2493                                next if $SKIP_PROP{$prop};
2494                                my $v = $h->{$path}->{$prop};
2495                                my $t_ppath_prop = "$t: " .
2496                                                    uri_encode($ppath) . ' ' .
2497                                                    uri_encode($prop);
2498                                if (defined $v) {
2499                                        push @out, "  +$t_ppath_prop " .
2500                                                   uri_encode($v);
2501                                } else {
2502                                        push @out, "  -$t_ppath_prop";
2503                                }
2504                        }
2505                }
2506        }
2507        foreach my $t (qw/absent_file absent_directory/) {
2508                $h = $ed->{$t} or next;
2509                foreach my $parent (sort keys %$h) {
2510                        foreach my $path (sort @{$h->{$parent}}) {
2511                                push @out, "  $t: " .
2512                                           uri_encode("$parent/$path");
2513                                warn "W: $t: $parent/$path ",
2514                                     "Insufficient permissions?\n";
2515                        }
2516                }
2517        }
2518        \@out;
2519}
2520
2521sub parse_svn_date {
2522        my $date = shift || return '+0000 1970-01-01 00:00:00';
2523        my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2524                                            (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x) or
2525                                         croak "Unable to parse date: $date\n";
2526        "+0000 $Y-$m-$d $H:$M:$S";
2527}
2528
2529sub check_author {
2530        my ($author) = @_;
2531        if (!defined $author || length $author == 0) {
2532                $author = '(no author)';
2533        } elsif (defined $::_authors && ! defined $::users{$author}) {
2534                die "Author: $author not defined in $::_authors file\n";
2535        }
2536        $author;
2537}
2538
2539sub make_log_entry {
2540        my ($self, $rev, $parents, $ed) = @_;
2541        my $untracked = $self->get_untracked($ed);
2542
2543        open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
2544        print $un "r$rev\n" or croak $!;
2545        print $un $_, "\n" foreach @$untracked;
2546        my %log_entry = ( parents => $parents || [], revision => $rev,
2547                          log => '');
2548
2549        my $headrev;
2550        my $logged = delete $self->{logged_rev_props};
2551        if (!$logged || $self->{-want_revprops}) {
2552                my $rp = $self->ra->rev_proplist($rev);
2553                foreach (sort keys %$rp) {
2554                        my $v = $rp->{$_};
2555                        if (/^svn:(author|date|log)$/) {
2556                                $log_entry{$1} = $v;
2557                        } elsif ($_ eq 'svm:headrev') {
2558                                $headrev = $v;
2559                        } else {
2560                                print $un "  rev_prop: ", uri_encode($_), ' ',
2561                                          uri_encode($v), "\n";
2562                        }
2563                }
2564        } else {
2565                map { $log_entry{$_} = $logged->{$_} } keys %$logged;
2566        }
2567        close $un or croak $!;
2568
2569        $log_entry{date} = parse_svn_date($log_entry{date});
2570        $log_entry{log} .= "\n";
2571        my $author = $log_entry{author} = check_author($log_entry{author});
2572        my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
2573                                                       : ($author, undef);
2574
2575        my ($commit_name, $commit_email) = ($name, $email);
2576        if ($_use_log_author) {
2577                my $name_field;
2578                if ($log_entry{log} =~ /From:\s+(.*\S)\s*\n/i) {
2579                        $name_field = $1;
2580                } elsif ($log_entry{log} =~ /Signed-off-by:\s+(.*\S)\s*\n/i) {
2581                        $name_field = $1;
2582                }
2583                if (!defined $name_field) {
2584                        if (!defined $email) {
2585                                $email = $name;
2586                        }
2587                } elsif ($name_field =~ /(.*?)\s+<(.*)>/) {
2588                        ($name, $email) = ($1, $2);
2589                } elsif ($name_field =~ /(.*)@/) {
2590                        ($name, $email) = ($1, $name_field);
2591                } else {
2592                        ($name, $email) = ($name_field, $name_field);
2593                }
2594        }
2595        if (defined $headrev && $self->use_svm_props) {
2596                if ($self->rewrite_root) {
2597                        die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
2598                            "options set!\n";
2599                }
2600                my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$};
2601                # we don't want "SVM: initializing mirror for junk" ...
2602                return undef if $r == 0;
2603                my $svm = $self->svm;
2604                if ($uuid ne $svm->{uuid}) {
2605                        die "UUID mismatch on SVM path:\n",
2606                            "expected: $svm->{uuid}\n",
2607                            "     got: $uuid\n";
2608                }
2609                my $full_url = $self->full_url;
2610                $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
2611                             die "Failed to replace '$svm->{replace}' with ",
2612                                 "'$svm->{source}' in $full_url\n";
2613                # throw away username for storing in records
2614                remove_username($full_url);
2615                $log_entry{metadata} = "$full_url\@$r $uuid";
2616                $log_entry{svm_revision} = $r;
2617                $email ||= "$author\@$uuid";
2618                $commit_email ||= "$author\@$uuid";
2619        } elsif ($self->use_svnsync_props) {
2620                my $full_url = $self->svnsync->{url};
2621                $full_url .= "/$self->{path}" if length $self->{path};
2622                remove_username($full_url);
2623                my $uuid = $self->svnsync->{uuid};
2624                $log_entry{metadata} = "$full_url\@$rev $uuid";
2625                $email ||= "$author\@$uuid";
2626                $commit_email ||= "$author\@$uuid";
2627        } else {
2628                my $url = $self->metadata_url;
2629                remove_username($url);
2630                $log_entry{metadata} = "$url\@$rev " .
2631                                       $self->ra->get_uuid;
2632                $email ||= "$author\@" . $self->ra->get_uuid;
2633                $commit_email ||= "$author\@" . $self->ra->get_uuid;
2634        }
2635        $log_entry{name} = $name;
2636        $log_entry{email} = $email;
2637        $log_entry{commit_name} = $commit_name;
2638        $log_entry{commit_email} = $commit_email;
2639        \%log_entry;
2640}
2641
2642sub fetch {
2643        my ($self, $min_rev, $max_rev, @parents) = @_;
2644        my ($last_rev, $last_commit) = $self->last_rev_commit;
2645        my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
2646        $self->ra->gs_fetch_loop_common($base, $head, [$self]);
2647}
2648
2649sub set_tree_cb {
2650        my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
2651        $self->{inject_parents} = { $rev => $tree };
2652        $self->fetch(undef, undef);
2653}
2654
2655sub set_tree {
2656        my ($self, $tree) = (shift, shift);
2657        my $log_entry = ::get_commit_entry($tree);
2658        unless ($self->{last_rev}) {
2659                ::fatal("Must have an existing revision to commit");
2660        }
2661        my %ed_opts = ( r => $self->{last_rev},
2662                        log => $log_entry->{log},
2663                        ra => $self->ra,
2664                        tree_a => $self->{last_commit},
2665                        tree_b => $tree,
2666                        editor_cb => sub {
2667                               $self->set_tree_cb($log_entry, $tree, @_) },
2668                        svn_path => $self->{path} );
2669        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
2670                print "No changes\nr$self->{last_rev} = $tree\n";
2671        }
2672}
2673
2674sub rebuild_from_rev_db {
2675        my ($self, $path) = @_;
2676        my $r = -1;
2677        open my $fh, '<', $path or croak "open: $!";
2678        binmode $fh or croak "binmode: $!";
2679        while (<$fh>) {
2680                length($_) == 41 or croak "inconsistent size in ($_) != 41";
2681                chomp($_);
2682                ++$r;
2683                next if $_ eq ('0' x 40);
2684                $self->rev_map_set($r, $_);
2685                print "r$r = $_\n";
2686        }
2687        close $fh or croak "close: $!";
2688        unlink $path or croak "unlink: $!";
2689}
2690
2691sub rebuild {
2692        my ($self) = @_;
2693        my $map_path = $self->map_path;
2694        my $partial = (-e $map_path && ! -z $map_path);
2695        return unless ::verify_ref($self->refname.'^0');
2696        if (!$partial && ($self->use_svm_props || $self->no_metadata)) {
2697                my $rev_db = $self->rev_db_path;
2698                $self->rebuild_from_rev_db($rev_db);
2699                if ($self->use_svm_props) {
2700                        my $svm_rev_db = $self->rev_db_path($self->svm_uuid);
2701                        $self->rebuild_from_rev_db($svm_rev_db);
2702                }
2703                $self->unlink_rev_db_symlink;
2704                return;
2705        }
2706        print "Rebuilding $map_path ...\n" if (!$partial);
2707        my ($base_rev, $head) = ($partial ? $self->rev_map_max_norebuild(1) :
2708                (undef, undef));
2709        my ($log, $ctx) =
2710            command_output_pipe(qw/rev-list --pretty=raw --no-color --reverse/,
2711                                ($head ? "$head.." : "") . $self->refname,
2712                                '--');
2713        my $metadata_url = $self->metadata_url;
2714        remove_username($metadata_url);
2715        my $svn_uuid = $self->ra_uuid;
2716        my $c;
2717        while (<$log>) {
2718                if ( m{^commit ($::sha1)$} ) {
2719                        $c = $1;
2720                        next;
2721                }
2722                next unless s{^\s*(git-svn-id:)}{$1};
2723                my ($url, $rev, $uuid) = ::extract_metadata($_);
2724                remove_username($url);
2725
2726                # ignore merges (from set-tree)
2727                next if (!defined $rev || !$uuid);
2728
2729                # if we merged or otherwise started elsewhere, this is
2730                # how we break out of it
2731                if (($uuid ne $svn_uuid) ||
2732                    ($metadata_url && $url && ($url ne $metadata_url))) {
2733                        next;
2734                }
2735                if ($partial && $head) {
2736                        print "Partial-rebuilding $map_path ...\n";
2737                        print "Currently at $base_rev = $head\n";
2738                        $head = undef;
2739                }
2740
2741                $self->rev_map_set($rev, $c);
2742                print "r$rev = $c\n";
2743        }
2744        command_close_pipe($log, $ctx);
2745        print "Done rebuilding $map_path\n" if (!$partial || !$head);
2746        my $rev_db_path = $self->rev_db_path;
2747        if (-f $self->rev_db_path) {
2748                unlink $self->rev_db_path or croak "unlink: $!";
2749        }
2750        $self->unlink_rev_db_symlink;
2751}
2752
2753# rev_map:
2754# Tie::File seems to be prone to offset errors if revisions get sparse,
2755# it's not that fast, either.  Tie::File is also not in Perl 5.6.  So
2756# one of my favorite modules is out :<  Next up would be one of the DBM
2757# modules, but I'm not sure which is most portable...
2758#
2759# This is the replacement for the rev_db format, which was too big
2760# and inefficient for large repositories with a lot of sparse history
2761# (mainly tags)
2762#
2763# The format is this:
2764#   - 24 bytes for every record,
2765#     * 4 bytes for the integer representing an SVN revision number
2766#     * 20 bytes representing the sha1 of a git commit
2767#   - No empty padding records like the old format
2768#     (except the last record, which can be overwritten)
2769#   - new records are written append-only since SVN revision numbers
2770#     increase monotonically
2771#   - lookups on SVN revision number are done via a binary search
2772#   - Piping the file to xxd -c24 is a good way of dumping it for
2773#     viewing or editing (piped back through xxd -r), should the need
2774#     ever arise.
2775#   - The last record can be padding revision with an all-zero sha1
2776#     This is used to optimize fetch performance when using multiple
2777#     "fetch" directives in .git/config
2778#
2779# These files are disposable unless noMetadata or useSvmProps is set
2780
2781sub _rev_map_set {
2782        my ($fh, $rev, $commit) = @_;
2783
2784        binmode $fh or croak "binmode: $!";
2785        my $size = (stat($fh))[7];
2786        ($size % 24) == 0 or croak "inconsistent size: $size";
2787
2788        my $wr_offset = 0;
2789        if ($size > 0) {
2790                sysseek($fh, -24, SEEK_END) or croak "seek: $!";
2791                my $read = sysread($fh, my $buf, 24) or croak "read: $!";
2792                $read == 24 or croak "read only $read bytes (!= 24)";
2793                my ($last_rev, $last_commit) = unpack(rev_map_fmt, $buf);
2794                if ($last_commit eq ('0' x40)) {
2795                        if ($size >= 48) {
2796                                sysseek($fh, -48, SEEK_END) or croak "seek: $!";
2797                                $read = sysread($fh, $buf, 24) or
2798                                    croak "read: $!";
2799                                $read == 24 or
2800                                    croak "read only $read bytes (!= 24)";
2801                                ($last_rev, $last_commit) =
2802                                    unpack(rev_map_fmt, $buf);
2803                                if ($last_commit eq ('0' x40)) {
2804                                        croak "inconsistent .rev_map\n";
2805                                }
2806                        }
2807                        if ($last_rev >= $rev) {
2808                                croak "last_rev is higher!: $last_rev >= $rev";
2809                        }
2810                        $wr_offset = -24;
2811                }
2812        }
2813        sysseek($fh, $wr_offset, SEEK_END) or croak "seek: $!";
2814        syswrite($fh, pack(rev_map_fmt, $rev, $commit), 24) == 24 or
2815          croak "write: $!";
2816}
2817
2818sub mkfile {
2819        my ($path) = @_;
2820        unless (-e $path) {
2821                my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
2822                mkpath([$dir]) unless -d $dir;
2823                open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
2824                close $fh or die "Couldn't close (create) $path: $!\n";
2825        }
2826}
2827
2828sub rev_map_set {
2829        my ($self, $rev, $commit, $update_ref, $uuid) = @_;
2830        length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
2831        my $db = $self->map_path($uuid);
2832        my $db_lock = "$db.lock";
2833        my $sig;
2834        if ($update_ref) {
2835                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
2836                            $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
2837        }
2838        mkfile($db);
2839
2840        $LOCKFILES{$db_lock} = 1;
2841        my $sync;
2842        # both of these options make our .rev_db file very, very important
2843        # and we can't afford to lose it because rebuild() won't work
2844        if ($self->use_svm_props || $self->no_metadata) {
2845                $sync = 1;
2846                copy($db, $db_lock) or die "rev_map_set(@_): ",
2847                                           "Failed to copy: ",
2848                                           "$db => $db_lock ($!)\n";
2849        } else {
2850                rename $db, $db_lock or die "rev_map_set(@_): ",
2851                                            "Failed to rename: ",
2852                                            "$db => $db_lock ($!)\n";
2853        }
2854
2855        sysopen(my $fh, $db_lock, O_RDWR | O_CREAT)
2856             or croak "Couldn't open $db_lock: $!\n";
2857        _rev_map_set($fh, $rev, $commit);
2858        if ($sync) {
2859                $fh->flush or die "Couldn't flush $db_lock: $!\n";
2860                $fh->sync or die "Couldn't sync $db_lock: $!\n";
2861        }
2862        close $fh or croak $!;
2863        if ($update_ref) {
2864                $_head = $self;
2865                command_noisy('update-ref', '-m', "r$rev",
2866                              $self->refname, $commit);
2867        }
2868        rename $db_lock, $db or die "rev_map_set(@_): ", "Failed to rename: ",
2869                                    "$db_lock => $db ($!)\n";
2870        delete $LOCKFILES{$db_lock};
2871        if ($update_ref) {
2872                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
2873                            $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
2874                kill $sig, $$ if defined $sig;
2875        }
2876}
2877
2878# If want_commit, this will return an array of (rev, commit) where
2879# commit _must_ be a valid commit in the archive.
2880# Otherwise, it'll return the max revision (whether or not the
2881# commit is valid or just a 0x40 placeholder).
2882sub rev_map_max {
2883        my ($self, $want_commit) = @_;
2884        $self->rebuild;
2885        my ($r, $c) = $self->rev_map_max_norebuild($want_commit);
2886        $want_commit ? ($r, $c) : $r;
2887}
2888
2889sub rev_map_max_norebuild {
2890        my ($self, $want_commit) = @_;
2891        my $map_path = $self->map_path;
2892        stat $map_path or return $want_commit ? (0, undef) : 0;
2893        sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
2894        binmode $fh or croak "binmode: $!";
2895        my $size = (stat($fh))[7];
2896        ($size % 24) == 0 or croak "inconsistent size: $size";
2897
2898        if ($size == 0) {
2899                close $fh or croak "close: $!";
2900                return $want_commit ? (0, undef) : 0;
2901        }
2902
2903        sysseek($fh, -24, SEEK_END) or croak "seek: $!";
2904        sysread($fh, my $buf, 24) == 24 or croak "read: $!";
2905        my ($r, $c) = unpack(rev_map_fmt, $buf);
2906        if ($want_commit && $c eq ('0' x40)) {
2907                if ($size < 48) {
2908                        return $want_commit ? (0, undef) : 0;
2909                }
2910                sysseek($fh, -48, SEEK_END) or croak "seek: $!";
2911                sysread($fh, $buf, 24) == 24 or croak "read: $!";
2912                ($r, $c) = unpack(rev_map_fmt, $buf);
2913                if ($c eq ('0'x40)) {
2914                        croak "Penultimate record is all-zeroes in $map_path";
2915                }
2916        }
2917        close $fh or croak "close: $!";
2918        $want_commit ? ($r, $c) : $r;
2919}
2920
2921sub rev_map_get {
2922        my ($self, $rev, $uuid) = @_;
2923        my $map_path = $self->map_path($uuid);
2924        return undef unless -e $map_path;
2925
2926        sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
2927        binmode $fh or croak "binmode: $!";
2928        my $size = (stat($fh))[7];
2929        ($size % 24) == 0 or croak "inconsistent size: $size";
2930
2931        if ($size == 0) {
2932                close $fh or croak "close: $fh";
2933                return undef;
2934        }
2935
2936        my ($l, $u) = (0, $size - 24);
2937        my ($r, $c, $buf);
2938
2939        while ($l <= $u) {
2940                my $i = int(($l/24 + $u/24) / 2) * 24;
2941                sysseek($fh, $i, SEEK_SET) or croak "seek: $!";
2942                sysread($fh, my $buf, 24) == 24 or croak "read: $!";
2943                my ($r, $c) = unpack('NH40', $buf);
2944
2945                if ($r < $rev) {
2946                        $l = $i + 24;
2947                } elsif ($r > $rev) {
2948                        $u = $i - 24;
2949                } else { # $r == $rev
2950                        close($fh) or croak "close: $!";
2951                        return $c eq ('0' x 40) ? undef : $c;
2952                }
2953        }
2954        close($fh) or croak "close: $!";
2955        undef;
2956}
2957
2958# Finds the first svn revision that exists on (if $eq_ok is true) or
2959# before $rev for the current branch.  It will not search any lower
2960# than $min_rev.  Returns the git commit hash and svn revision number
2961# if found, else (undef, undef).
2962sub find_rev_before {
2963        my ($self, $rev, $eq_ok, $min_rev) = @_;
2964        --$rev unless $eq_ok;
2965        $min_rev ||= 1;
2966        while ($rev >= $min_rev) {
2967                if (my $c = $self->rev_map_get($rev)) {
2968                        return ($rev, $c);
2969                }
2970                --$rev;
2971        }
2972        return (undef, undef);
2973}
2974
2975# Finds the first svn revision that exists on (if $eq_ok is true) or
2976# after $rev for the current branch.  It will not search any higher
2977# than $max_rev.  Returns the git commit hash and svn revision number
2978# if found, else (undef, undef).
2979sub find_rev_after {
2980        my ($self, $rev, $eq_ok, $max_rev) = @_;
2981        ++$rev unless $eq_ok;
2982        $max_rev ||= $self->rev_map_max;
2983        while ($rev <= $max_rev) {
2984                if (my $c = $self->rev_map_get($rev)) {
2985                        return ($rev, $c);
2986                }
2987                ++$rev;
2988        }
2989        return (undef, undef);
2990}
2991
2992sub _new {
2993        my ($class, $repo_id, $ref_id, $path) = @_;
2994        unless (defined $repo_id && length $repo_id) {
2995                $repo_id = $Git::SVN::default_repo_id;
2996        }
2997        unless (defined $ref_id && length $ref_id) {
2998                $_[2] = $ref_id = $Git::SVN::default_ref_id;
2999        }
3000        $_[1] = $repo_id;
3001        my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
3002        $_[3] = $path = '' unless (defined $path);
3003        mkpath(["$ENV{GIT_DIR}/svn"]);
3004        bless {
3005                ref_id => $ref_id, dir => $dir, index => "$dir/index",
3006                path => $path, config => "$ENV{GIT_DIR}/svn/config",
3007                map_root => "$dir/.rev_map", repo_id => $repo_id }, $class;
3008}
3009
3010# for read-only access of old .rev_db formats
3011sub unlink_rev_db_symlink {
3012        my ($self) = @_;
3013        my $link = $self->rev_db_path;
3014        $link =~ s/\.[\w-]+$// or croak "missing UUID at the end of $link";
3015        if (-l $link) {
3016                unlink $link or croak "unlink: $link failed!";
3017        }
3018}
3019
3020sub rev_db_path {
3021        my ($self, $uuid) = @_;
3022        my $db_path = $self->map_path($uuid);
3023        $db_path =~ s{/\.rev_map\.}{/\.rev_db\.}
3024            or croak "map_path: $db_path does not contain '/.rev_map.' !";
3025        $db_path;
3026}
3027
3028# the new replacement for .rev_db
3029sub map_path {
3030        my ($self, $uuid) = @_;
3031        $uuid ||= $self->ra_uuid;
3032        "$self->{map_root}.$uuid";
3033}
3034
3035sub uri_encode {
3036        my ($f) = @_;
3037        $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
3038        $f
3039}
3040
3041sub remove_username {
3042        $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
3043}
3044
3045package Git::SVN::Prompt;
3046use strict;
3047use warnings;
3048require SVN::Core;
3049use vars qw/$_no_auth_cache $_username/;
3050
3051sub simple {
3052        my ($cred, $realm, $default_username, $may_save, $pool) = @_;
3053        $may_save = undef if $_no_auth_cache;
3054        $default_username = $_username if defined $_username;
3055        if (defined $default_username && length $default_username) {
3056                if (defined $realm && length $realm) {
3057                        print STDERR "Authentication realm: $realm\n";
3058                        STDERR->flush;
3059                }
3060                $cred->username($default_username);
3061        } else {
3062                username($cred, $realm, $may_save, $pool);
3063        }
3064        $cred->password(_read_password("Password for '" .
3065                                       $cred->username . "': ", $realm));
3066        $cred->may_save($may_save);
3067        $SVN::_Core::SVN_NO_ERROR;
3068}
3069
3070sub ssl_server_trust {
3071        my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
3072        $may_save = undef if $_no_auth_cache;
3073        print STDERR "Error validating server certificate for '$realm':\n";
3074        {
3075                no warnings 'once';
3076                # All variables SVN::Auth::SSL::* are used only once,
3077                # so we're shutting up Perl warnings about this.
3078                if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
3079                        print STDERR " - The certificate is not issued ",
3080                            "by a trusted authority. Use the\n",
3081                            "   fingerprint to validate ",
3082                            "the certificate manually!\n";
3083                }
3084                if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
3085                        print STDERR " - The certificate hostname ",
3086                            "does not match.\n";
3087                }
3088                if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
3089                        print STDERR " - The certificate is not yet valid.\n";
3090                }
3091                if ($failures & $SVN::Auth::SSL::EXPIRED) {
3092                        print STDERR " - The certificate has expired.\n";
3093                }
3094                if ($failures & $SVN::Auth::SSL::OTHER) {
3095                        print STDERR " - The certificate has ",
3096                            "an unknown error.\n";
3097                }
3098        } # no warnings 'once'
3099        printf STDERR
3100                "Certificate information:\n".
3101                " - Hostname: %s\n".
3102                " - Valid: from %s until %s\n".
3103                " - Issuer: %s\n".
3104                " - Fingerprint: %s\n",
3105                map $cert_info->$_, qw(hostname valid_from valid_until
3106                                       issuer_dname fingerprint);
3107        my $choice;
3108prompt:
3109        print STDERR $may_save ?
3110              "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
3111              "(R)eject or accept (t)emporarily? ";
3112        STDERR->flush;
3113        $choice = lc(substr(<STDIN> || 'R', 0, 1));
3114        if ($choice =~ /^t$/i) {
3115                $cred->may_save(undef);
3116        } elsif ($choice =~ /^r$/i) {
3117                return -1;
3118        } elsif ($may_save && $choice =~ /^p$/i) {
3119                $cred->may_save($may_save);
3120        } else {
3121                goto prompt;
3122        }
3123        $cred->accepted_failures($failures);
3124        $SVN::_Core::SVN_NO_ERROR;
3125}
3126
3127sub ssl_client_cert {
3128        my ($cred, $realm, $may_save, $pool) = @_;
3129        $may_save = undef if $_no_auth_cache;
3130        print STDERR "Client certificate filename: ";
3131        STDERR->flush;
3132        chomp(my $filename = <STDIN>);
3133        $cred->cert_file($filename);
3134        $cred->may_save($may_save);
3135        $SVN::_Core::SVN_NO_ERROR;
3136}
3137
3138sub ssl_client_cert_pw {
3139        my ($cred, $realm, $may_save, $pool) = @_;
3140        $may_save = undef if $_no_auth_cache;
3141        $cred->password(_read_password("Password: ", $realm));
3142        $cred->may_save($may_save);
3143        $SVN::_Core::SVN_NO_ERROR;
3144}
3145
3146sub username {
3147        my ($cred, $realm, $may_save, $pool) = @_;
3148        $may_save = undef if $_no_auth_cache;
3149        if (defined $realm && length $realm) {
3150                print STDERR "Authentication realm: $realm\n";
3151        }
3152        my $username;
3153        if (defined $_username) {
3154                $username = $_username;
3155        } else {
3156                print STDERR "Username: ";
3157                STDERR->flush;
3158                chomp($username = <STDIN>);
3159        }
3160        $cred->username($username);
3161        $cred->may_save($may_save);
3162        $SVN::_Core::SVN_NO_ERROR;
3163}
3164
3165sub _read_password {
3166        my ($prompt, $realm) = @_;
3167        print STDERR $prompt;
3168        STDERR->flush;
3169        require Term::ReadKey;
3170        Term::ReadKey::ReadMode('noecho');
3171        my $password = '';
3172        while (defined(my $key = Term::ReadKey::ReadKey(0))) {
3173                last if $key =~ /[\012\015]/; # \n\r
3174                $password .= $key;
3175        }
3176        Term::ReadKey::ReadMode('restore');
3177        print STDERR "\n";
3178        STDERR->flush;
3179        $password;
3180}
3181
3182package SVN::Git::Fetcher;
3183use vars qw/@ISA/;
3184use strict;
3185use warnings;
3186use Carp qw/croak/;
3187use File::Temp qw/tempfile/;
3188use IO::File qw//;
3189
3190# file baton members: path, mode_a, mode_b, pool, fh, blob, base
3191sub new {
3192        my ($class, $git_svn) = @_;
3193        my $self = SVN::Delta::Editor->new;
3194        bless $self, $class;
3195        $self->{c} = $git_svn->{last_commit} if exists $git_svn->{last_commit};
3196        $self->{empty} = {};
3197        $self->{dir_prop} = {};
3198        $self->{file_prop} = {};
3199        $self->{absent_dir} = {};
3200        $self->{absent_file} = {};
3201        $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
3202        $self;
3203}
3204
3205sub set_path_strip {
3206        my ($self, $path) = @_;
3207        $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
3208}
3209
3210sub open_root {
3211        { path => '' };
3212}
3213
3214sub open_directory {
3215        my ($self, $path, $pb, $rev) = @_;
3216        { path => $path };
3217}
3218
3219sub git_path {
3220        my ($self, $path) = @_;
3221        if ($self->{path_strip}) {
3222                $path =~ s!$self->{path_strip}!! or
3223                  die "Failed to strip path '$path' ($self->{path_strip})\n";
3224        }
3225        $path;
3226}
3227
3228sub delete_entry {
3229        my ($self, $path, $rev, $pb) = @_;
3230
3231        my $gpath = $self->git_path($path);
3232        return undef if ($gpath eq '');
3233
3234        # remove entire directories.
3235        if (command('ls-tree', $self->{c}, '--', $gpath) =~ /^040000 tree/) {
3236                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3237                                                     -r --name-only -z/,
3238                                                     $self->{c}, '--', $gpath);
3239                local $/ = "\0";
3240                while (<$ls>) {
3241                        chomp;
3242                        $self->{gii}->remove($_);
3243                        print "\tD\t$_\n" unless $::_q;
3244                }
3245                print "\tD\t$gpath/\n" unless $::_q;
3246                command_close_pipe($ls, $ctx);
3247                $self->{empty}->{$path} = 0
3248        } else {
3249                $self->{gii}->remove($gpath);
3250                print "\tD\t$gpath\n" unless $::_q;
3251        }
3252        undef;
3253}
3254
3255sub open_file {
3256        my ($self, $path, $pb, $rev) = @_;
3257        my $gpath = $self->git_path($path);
3258        my ($mode, $blob) = (command('ls-tree', $self->{c}, '--', $gpath)
3259                             =~ /^(\d{6}) blob ([a-f\d]{40})\t/);
3260        unless (defined $mode && defined $blob) {
3261                die "$path was not found in commit $self->{c} (r$rev)\n";
3262        }
3263        { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
3264          pool => SVN::Pool->new, action => 'M' };
3265}
3266
3267sub add_file {
3268        my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
3269        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3270        delete $self->{empty}->{$dir};
3271        { path => $path, mode_a => 100644, mode_b => 100644,
3272          pool => SVN::Pool->new, action => 'A' };
3273}
3274
3275sub add_directory {
3276        my ($self, $path, $cp_path, $cp_rev) = @_;
3277        my $gpath = $self->git_path($path);
3278        if ($gpath eq '') {
3279                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3280                                                     -r --name-only -z/,
3281                                                     $self->{c});
3282                local $/ = "\0";
3283                while (<$ls>) {
3284                        chomp;
3285                        $self->{gii}->remove($_);
3286                        print "\tD\t$_\n" unless $::_q;
3287                }
3288                command_close_pipe($ls, $ctx);
3289                $self->{empty}->{$path} = 0;
3290        }
3291        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3292        delete $self->{empty}->{$dir};
3293        $self->{empty}->{$path} = 1;
3294        { path => $path };
3295}
3296
3297sub change_dir_prop {
3298        my ($self, $db, $prop, $value) = @_;
3299        $self->{dir_prop}->{$db->{path}} ||= {};
3300        $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
3301        undef;
3302}
3303
3304sub absent_directory {
3305        my ($self, $path, $pb) = @_;
3306        $self->{absent_dir}->{$pb->{path}} ||= [];
3307        push @{$self->{absent_dir}->{$pb->{path}}}, $path;
3308        undef;
3309}
3310
3311sub absent_file {
3312        my ($self, $path, $pb) = @_;
3313        $self->{absent_file}->{$pb->{path}} ||= [];
3314        push @{$self->{absent_file}->{$pb->{path}}}, $path;
3315        undef;
3316}
3317
3318sub change_file_prop {
3319        my ($self, $fb, $prop, $value) = @_;
3320        if ($prop eq 'svn:executable') {
3321                if ($fb->{mode_b} != 120000) {
3322                        $fb->{mode_b} = defined $value ? 100755 : 100644;
3323                }
3324        } elsif ($prop eq 'svn:special') {
3325                $fb->{mode_b} = defined $value ? 120000 : 100644;
3326        } else {
3327                $self->{file_prop}->{$fb->{path}} ||= {};
3328                $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
3329        }
3330        undef;
3331}
3332
3333sub apply_textdelta {
3334        my ($self, $fb, $exp) = @_;
3335        my $fh = $::_repository->temp_acquire('svn_delta');
3336        # $fh gets auto-closed() by SVN::TxDelta::apply(),
3337        # (but $base does not,) so dup() it for reading in close_file
3338        open my $dup, '<&', $fh or croak $!;
3339        my $base = $::_repository->temp_acquire('git_blob');
3340        if ($fb->{blob}) {
3341                print $base 'link ' if ($fb->{mode_a} == 120000);
3342                my $size = $::_repository->cat_blob($fb->{blob}, $base);
3343                die "Failed to read object $fb->{blob}" if ($size < 0);
3344
3345                if (defined $exp) {
3346                        seek $base, 0, 0 or croak $!;
3347                        my $got = ::md5sum($base);
3348                        die "Checksum mismatch: $fb->{path} $fb->{blob}\n",
3349                            "expected: $exp\n",
3350                            "     got: $got\n" if ($got ne $exp);
3351                }
3352        }
3353        seek $base, 0, 0 or croak $!;
3354        $fb->{fh} = $fh;
3355        $fb->{base} = $base;
3356        [ SVN::TxDelta::apply($base, $dup, undef, $fb->{path}, $fb->{pool}) ];
3357}
3358
3359sub close_file {
3360        my ($self, $fb, $exp) = @_;
3361        my $hash;
3362        my $path = $self->git_path($fb->{path});
3363        if (my $fh = $fb->{fh}) {
3364                if (defined $exp) {
3365                        seek($fh, 0, 0) or croak $!;
3366                        my $got = ::md5sum($fh);
3367                        if ($got ne $exp) {
3368                                die "Checksum mismatch: $path\n",
3369                                    "expected: $exp\n    got: $got\n";
3370                        }
3371                }
3372                if ($fb->{mode_b} == 120000) {
3373                        sysseek($fh, 0, 0) or croak $!;
3374                        sysread($fh, my $buf, 5) == 5 or croak $!;
3375
3376                        unless ($buf eq 'link ') {
3377                                warn "$path has mode 120000",
3378                                                " but is not a link\n";
3379                        } else {
3380                                my $tmp_fh = $::_repository->temp_acquire(
3381                                        'svn_hash');
3382                                my $res;
3383                                while ($res = sysread($fh, my $str, 1024)) {
3384                                        my $out = syswrite($tmp_fh, $str, $res);
3385                                        defined($out) && $out == $res
3386                                                or croak("write ",
3387                                                        Git::temp_path($tmp_fh),
3388                                                        ": $!\n");
3389                                }
3390                                defined $res or croak $!;
3391
3392                                ($fh, $tmp_fh) = ($tmp_fh, $fh);
3393                                Git::temp_release($tmp_fh, 1);
3394                        }
3395                }
3396
3397                $hash = $::_repository->hash_and_insert_object(
3398                                Git::temp_path($fh));
3399                $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
3400
3401                Git::temp_release($fb->{base}, 1);
3402                Git::temp_release($fh, 1);
3403        } else {
3404                $hash = $fb->{blob} or die "no blob information\n";
3405        }
3406        $fb->{pool}->clear;
3407        $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
3408        print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
3409        undef;
3410}
3411
3412sub abort_edit {
3413        my $self = shift;
3414        $self->{nr} = $self->{gii}->{nr};
3415        delete $self->{gii};
3416        $self->SUPER::abort_edit(@_);
3417}
3418
3419sub close_edit {
3420        my $self = shift;
3421        $self->{git_commit_ok} = 1;
3422        $self->{nr} = $self->{gii}->{nr};
3423        delete $self->{gii};
3424        $self->SUPER::close_edit(@_);
3425}
3426
3427package SVN::Git::Editor;
3428use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
3429use strict;
3430use warnings;
3431use Carp qw/croak/;
3432use IO::File;
3433
3434sub new {
3435        my ($class, $opts) = @_;
3436        foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
3437                die "$_ required!\n" unless (defined $opts->{$_});
3438        }
3439
3440        my $pool = SVN::Pool->new;
3441        my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
3442        my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
3443                                     $opts->{r}, $mods);
3444
3445        # $opts->{ra} functions should not be used after this:
3446        my @ce  = $opts->{ra}->get_commit_editor($opts->{log},
3447                                                $opts->{editor_cb}, $pool);
3448        my $self = SVN::Delta::Editor->new(@ce, $pool);
3449        bless $self, $class;
3450        foreach (qw/svn_path r tree_a tree_b/) {
3451                $self->{$_} = $opts->{$_};
3452        }
3453        $self->{url} = $opts->{ra}->{url};
3454        $self->{mods} = $mods;
3455        $self->{types} = $types;
3456        $self->{pool} = $pool;
3457        $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
3458        $self->{rm} = { };
3459        $self->{path_prefix} = length $self->{svn_path} ?
3460                               "$self->{svn_path}/" : '';
3461        $self->{config} = $opts->{config};
3462        return $self;
3463}
3464
3465sub generate_diff {
3466        my ($tree_a, $tree_b) = @_;
3467        my @diff_tree = qw(diff-tree -z -r);
3468        if ($_cp_similarity) {
3469                push @diff_tree, "-C$_cp_similarity";
3470        } else {
3471                push @diff_tree, '-C';
3472        }
3473        push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
3474        push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
3475        push @diff_tree, $tree_a, $tree_b;
3476        my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
3477        local $/ = "\0";
3478        my $state = 'meta';
3479        my @mods;
3480        while (<$diff_fh>) {
3481                chomp $_; # this gets rid of the trailing "\0"
3482                if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
3483                                        ($::sha1)\s($::sha1)\s
3484                                        ([MTCRAD])\d*$/xo) {
3485                        push @mods, {   mode_a => $1, mode_b => $2,
3486                                        sha1_a => $3, sha1_b => $4,
3487                                        chg => $5 };
3488                        if ($5 =~ /^(?:C|R)$/) {
3489                                $state = 'file_a';
3490                        } else {
3491                                $state = 'file_b';
3492                        }
3493                } elsif ($state eq 'file_a') {
3494                        my $x = $mods[$#mods] or croak "Empty array\n";
3495                        if ($x->{chg} !~ /^(?:C|R)$/) {
3496                                croak "Error parsing $_, $x->{chg}\n";
3497                        }
3498                        $x->{file_a} = $_;
3499                        $state = 'file_b';
3500                } elsif ($state eq 'file_b') {
3501                        my $x = $mods[$#mods] or croak "Empty array\n";
3502                        if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
3503                                croak "Error parsing $_, $x->{chg}\n";
3504                        }
3505                        if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
3506                                croak "Error parsing $_, $x->{chg}\n";
3507                        }
3508                        $x->{file_b} = $_;
3509                        $state = 'meta';
3510                } else {
3511                        croak "Error parsing $_\n";
3512                }
3513        }
3514        command_close_pipe($diff_fh, $ctx);
3515        \@mods;
3516}
3517
3518sub check_diff_paths {
3519        my ($ra, $pfx, $rev, $mods) = @_;
3520        my %types;
3521        $pfx .= '/' if length $pfx;
3522
3523        sub type_diff_paths {
3524                my ($ra, $types, $path, $rev) = @_;
3525                my @p = split m#/+#, $path;
3526                my $c = shift @p;
3527                unless (defined $types->{$c}) {
3528                        $types->{$c} = $ra->check_path($c, $rev);
3529                }
3530                while (@p) {
3531                        $c .= '/' . shift @p;
3532                        next if defined $types->{$c};
3533                        $types->{$c} = $ra->check_path($c, $rev);
3534                }
3535        }
3536
3537        foreach my $m (@$mods) {
3538                foreach my $f (qw/file_a file_b/) {
3539                        next unless defined $m->{$f};
3540                        my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
3541                        if (length $pfx.$dir && ! defined $types{$dir}) {
3542                                type_diff_paths($ra, \%types, $pfx.$dir, $rev);
3543                        }
3544                }
3545        }
3546        \%types;
3547}
3548
3549sub split_path {
3550        return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
3551}
3552
3553sub repo_path {
3554        my ($self, $path) = @_;
3555        $self->{path_prefix}.(defined $path ? $path : '');
3556}
3557
3558sub url_path {
3559        my ($self, $path) = @_;
3560        if ($self->{url} =~ m#^https?://#) {
3561                $path =~ s/([^~a-zA-Z0-9_.-])/uc sprintf("%%%02x",ord($1))/eg;
3562        }
3563        $self->{url} . '/' . $self->repo_path($path);
3564}
3565
3566sub rmdirs {
3567        my ($self) = @_;
3568        my $rm = $self->{rm};
3569        delete $rm->{''}; # we never delete the url we're tracking
3570        return unless %$rm;
3571
3572        foreach (keys %$rm) {
3573                my @d = split m#/#, $_;
3574                my $c = shift @d;
3575                $rm->{$c} = 1;
3576                while (@d) {
3577                        $c .= '/' . shift @d;
3578                        $rm->{$c} = 1;
3579                }
3580        }
3581        delete $rm->{$self->{svn_path}};
3582        delete $rm->{''}; # we never delete the url we're tracking
3583        return unless %$rm;
3584
3585        my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
3586                                             $self->{tree_b});
3587        local $/ = "\0";
3588        while (<$fh>) {
3589                chomp;
3590                my @dn = split m#/#, $_;
3591                while (pop @dn) {
3592                        delete $rm->{join '/', @dn};
3593                }
3594                unless (%$rm) {
3595                        close $fh;
3596                        return;
3597                }
3598        }
3599        command_close_pipe($fh, $ctx);
3600
3601        my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
3602        foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
3603                $self->close_directory($bat->{$d}, $p);
3604                my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
3605                print "\tD+\t$d/\n" unless $::_q;
3606                $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
3607                delete $bat->{$d};
3608        }
3609}
3610
3611sub open_or_add_dir {
3612        my ($self, $full_path, $baton) = @_;
3613        my $t = $self->{types}->{$full_path};
3614        if (!defined $t) {
3615                die "$full_path not known in r$self->{r} or we have a bug!\n";
3616        }
3617        {
3618                no warnings 'once';
3619                # SVN::Node::none and SVN::Node::file are used only once,
3620                # so we're shutting up Perl's warnings about them.
3621                if ($t == $SVN::Node::none) {
3622                        return $self->add_directory($full_path, $baton,
3623                            undef, -1, $self->{pool});
3624                } elsif ($t == $SVN::Node::dir) {
3625                        return $self->open_directory($full_path, $baton,
3626                            $self->{r}, $self->{pool});
3627                } # no warnings 'once'
3628                print STDERR "$full_path already exists in repository at ",
3629                    "r$self->{r} and it is not a directory (",
3630                    ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
3631        } # no warnings 'once'
3632        exit 1;
3633}
3634
3635sub ensure_path {
3636        my ($self, $path) = @_;
3637        my $bat = $self->{bat};
3638        my $repo_path = $self->repo_path($path);
3639        return $bat->{''} unless (length $repo_path);
3640        my @p = split m#/+#, $repo_path;
3641        my $c = shift @p;
3642        $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
3643        while (@p) {
3644                my $c0 = $c;
3645                $c .= '/' . shift @p;
3646                $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
3647        }
3648        return $bat->{$c};
3649}
3650
3651# Subroutine to convert a globbing pattern to a regular expression.
3652# From perl cookbook.
3653sub glob2pat {
3654        my $globstr = shift;
3655        my %patmap = ('*' => '.*', '?' => '.', '[' => '[', ']' => ']');
3656        $globstr =~ s{(.)} { $patmap{$1} || "\Q$1" }ge;
3657        return '^' . $globstr . '$';
3658}
3659
3660sub check_autoprop {
3661        my ($self, $pattern, $properties, $file, $fbat) = @_;
3662        # Convert the globbing pattern to a regular expression.
3663        my $regex = glob2pat($pattern);
3664        # Check if the pattern matches the file name.
3665        if($file =~ m/($regex)/) {
3666                # Parse the list of properties to set.
3667                my @props = split(/;/, $properties);
3668                foreach my $prop (@props) {
3669                        # Parse 'name=value' syntax and set the property.
3670                        if ($prop =~ /([^=]+)=(.*)/) {
3671                                my ($n,$v) = ($1,$2);
3672                                for ($n, $v) {
3673                                        s/^\s+//; s/\s+$//;
3674                                }
3675                                $self->change_file_prop($fbat, $n, $v);
3676                        }
3677                }
3678        }
3679}
3680
3681sub apply_autoprops {
3682        my ($self, $file, $fbat) = @_;
3683        my $conf_t = ${$self->{config}}{'config'};
3684        no warnings 'once';
3685        # Check [miscellany]/enable-auto-props in svn configuration.
3686        if (SVN::_Core::svn_config_get_bool(
3687                $conf_t,
3688                $SVN::_Core::SVN_CONFIG_SECTION_MISCELLANY,
3689                $SVN::_Core::SVN_CONFIG_OPTION_ENABLE_AUTO_PROPS,
3690                0)) {
3691                # Auto-props are enabled.  Enumerate them to look for matches.
3692                my $callback = sub {
3693                        $self->check_autoprop($_[0], $_[1], $file, $fbat);
3694                };
3695                SVN::_Core::svn_config_enumerate(
3696                        $conf_t,
3697                        $SVN::_Core::SVN_CONFIG_SECTION_AUTO_PROPS,
3698                        $callback);
3699        }
3700}
3701
3702sub A {
3703        my ($self, $m) = @_;
3704        my ($dir, $file) = split_path($m->{file_b});
3705        my $pbat = $self->ensure_path($dir);
3706        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3707                                        undef, -1);
3708        print "\tA\t$m->{file_b}\n" unless $::_q;
3709        $self->apply_autoprops($file, $fbat);
3710        $self->chg_file($fbat, $m);
3711        $self->close_file($fbat,undef,$self->{pool});
3712}
3713
3714sub C {
3715        my ($self, $m) = @_;
3716        my ($dir, $file) = split_path($m->{file_b});
3717        my $pbat = $self->ensure_path($dir);
3718        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3719                                $self->url_path($m->{file_a}), $self->{r});
3720        print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
3721        $self->chg_file($fbat, $m);
3722        $self->close_file($fbat,undef,$self->{pool});
3723}
3724
3725sub delete_entry {
3726        my ($self, $path, $pbat) = @_;
3727        my $rpath = $self->repo_path($path);
3728        my ($dir, $file) = split_path($rpath);
3729        $self->{rm}->{$dir} = 1;
3730        $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
3731}
3732
3733sub R {
3734        my ($self, $m) = @_;
3735        my ($dir, $file) = split_path($m->{file_b});
3736        my $pbat = $self->ensure_path($dir);
3737        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3738                                $self->url_path($m->{file_a}), $self->{r});
3739        print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
3740        $self->apply_autoprops($file, $fbat);
3741        $self->chg_file($fbat, $m);
3742        $self->close_file($fbat,undef,$self->{pool});
3743
3744        ($dir, $file) = split_path($m->{file_a});
3745        $pbat = $self->ensure_path($dir);
3746        $self->delete_entry($m->{file_a}, $pbat);
3747}
3748
3749sub M {
3750        my ($self, $m) = @_;
3751        my ($dir, $file) = split_path($m->{file_b});
3752        my $pbat = $self->ensure_path($dir);
3753        my $fbat = $self->open_file($self->repo_path($m->{file_b}),
3754                                $pbat,$self->{r},$self->{pool});
3755        print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
3756        $self->chg_file($fbat, $m);
3757        $self->close_file($fbat,undef,$self->{pool});
3758}
3759
3760sub T { shift->M(@_) }
3761
3762sub change_file_prop {
3763        my ($self, $fbat, $pname, $pval) = @_;
3764        $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
3765}
3766
3767sub _chg_file_get_blob ($$$$) {
3768        my ($self, $fbat, $m, $which) = @_;
3769        my $fh = $::_repository->temp_acquire("git_blob_$which");
3770        if ($m->{"mode_$which"} =~ /^120/) {
3771                print $fh 'link ' or croak $!;
3772                $self->change_file_prop($fbat,'svn:special','*');
3773        } elsif ($m->{mode_a} =~ /^120/ && $m->{"mode_$which"} !~ /^120/) {
3774                $self->change_file_prop($fbat,'svn:special',undef);
3775        }
3776        my $blob = $m->{"sha1_$which"};
3777        return ($fh,) if ($blob =~ /^0{40}$/);
3778        my $size = $::_repository->cat_blob($blob, $fh);
3779        croak "Failed to read object $blob" if ($size < 0);
3780        $fh->flush == 0 or croak $!;
3781        seek $fh, 0, 0 or croak $!;
3782
3783        my $exp = ::md5sum($fh);
3784        seek $fh, 0, 0 or croak $!;
3785        return ($fh, $exp);
3786}
3787
3788sub chg_file {
3789        my ($self, $fbat, $m) = @_;
3790        if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
3791                $self->change_file_prop($fbat,'svn:executable','*');
3792        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
3793                $self->change_file_prop($fbat,'svn:executable',undef);
3794        }
3795        my ($fh_a, $exp_a) = _chg_file_get_blob $self, $fbat, $m, 'a';
3796        my ($fh_b, $exp_b) = _chg_file_get_blob $self, $fbat, $m, 'b';
3797        my $pool = SVN::Pool->new;
3798        my $atd = $self->apply_textdelta($fbat, $exp_a, $pool);
3799        if (-s $fh_a) {
3800                my $txstream = SVN::TxDelta::new ($fh_a, $fh_b, $pool);
3801                my $res = SVN::TxDelta::send_txstream($txstream, @$atd, $pool);
3802                if (defined $res) {
3803                        die "Unexpected result from send_txstream: $res\n",
3804                            "(SVN::Core::VERSION: $SVN::Core::VERSION)\n";
3805                }
3806        } else {
3807                my $got = SVN::TxDelta::send_stream($fh_b, @$atd, $pool);
3808                die "Checksum mismatch\nexpected: $exp_b\ngot: $got\n"
3809                    if ($got ne $exp_b);
3810        }
3811        Git::temp_release($fh_b, 1);
3812        Git::temp_release($fh_a, 1);
3813        $pool->clear;
3814}
3815
3816sub D {
3817        my ($self, $m) = @_;
3818        my ($dir, $file) = split_path($m->{file_b});
3819        my $pbat = $self->ensure_path($dir);
3820        print "\tD\t$m->{file_b}\n" unless $::_q;
3821        $self->delete_entry($m->{file_b}, $pbat);
3822}
3823
3824sub close_edit {
3825        my ($self) = @_;
3826        my ($p,$bat) = ($self->{pool}, $self->{bat});
3827        foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
3828                next if $_ eq '';
3829                $self->close_directory($bat->{$_}, $p);
3830        }
3831        $self->close_directory($bat->{''}, $p);
3832        $self->SUPER::close_edit($p);
3833        $p->clear;
3834}
3835
3836sub abort_edit {
3837        my ($self) = @_;
3838        $self->SUPER::abort_edit($self->{pool});
3839}
3840
3841sub DESTROY {
3842        my $self = shift;
3843        $self->SUPER::DESTROY(@_);
3844        $self->{pool}->clear;
3845}
3846
3847# this drives the editor
3848sub apply_diff {
3849        my ($self) = @_;
3850        my $mods = $self->{mods};
3851        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
3852        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
3853                my $f = $m->{chg};
3854                if (defined $o{$f}) {
3855                        $self->$f($m);
3856                } else {
3857                        fatal("Invalid change type: $f");
3858                }
3859        }
3860        $self->rmdirs if $_rmdir;
3861        if (@$mods == 0) {
3862                $self->abort_edit;
3863        } else {
3864                $self->close_edit;
3865        }
3866        return scalar @$mods;
3867}
3868
3869package Git::SVN::Ra;
3870use vars qw/@ISA $config_dir $_log_window_size/;
3871use strict;
3872use warnings;
3873my ($ra_invalid, $can_do_switch, %ignored_err, $RA);
3874
3875BEGIN {
3876        # enforce temporary pool usage for some simple functions
3877        no strict 'refs';
3878        for my $f (qw/rev_proplist get_latest_revnum get_uuid get_repos_root/) {
3879                my $SUPER = "SUPER::$f";
3880                *$f = sub {
3881                        my $self = shift;
3882                        my $pool = SVN::Pool->new;
3883                        my @ret = $self->$SUPER(@_,$pool);
3884                        $pool->clear;
3885                        wantarray ? @ret : $ret[0];
3886                };
3887        }
3888}
3889
3890sub _auth_providers () {
3891        [
3892          SVN::Client::get_simple_provider(),
3893          SVN::Client::get_ssl_server_trust_file_provider(),
3894          SVN::Client::get_simple_prompt_provider(
3895            \&Git::SVN::Prompt::simple, 2),
3896          SVN::Client::get_ssl_client_cert_file_provider(),
3897          SVN::Client::get_ssl_client_cert_prompt_provider(
3898            \&Git::SVN::Prompt::ssl_client_cert, 2),
3899          SVN::Client::get_ssl_client_cert_pw_file_provider(),
3900          SVN::Client::get_ssl_client_cert_pw_prompt_provider(
3901            \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
3902          SVN::Client::get_username_provider(),
3903          SVN::Client::get_ssl_server_trust_prompt_provider(
3904            \&Git::SVN::Prompt::ssl_server_trust),
3905          SVN::Client::get_username_prompt_provider(
3906            \&Git::SVN::Prompt::username, 2)
3907        ]
3908}
3909
3910sub escape_uri_only {
3911        my ($uri) = @_;
3912        my @tmp;
3913        foreach (split m{/}, $uri) {
3914                s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
3915                push @tmp, $_;
3916        }
3917        join('/', @tmp);
3918}
3919
3920sub escape_url {
3921        my ($url) = @_;
3922        if ($url =~ m#^(https?)://([^/]+)(.*)$#) {
3923                my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
3924                $url = "$scheme://$domain$uri";
3925        }
3926        $url;
3927}
3928
3929sub new {
3930        my ($class, $url) = @_;
3931        $url =~ s!/+$!!;
3932        return $RA if ($RA && $RA->{url} eq $url);
3933
3934        SVN::_Core::svn_config_ensure($config_dir, undef);
3935        my ($baton, $callbacks) = SVN::Core::auth_open_helper(_auth_providers);
3936        my $config = SVN::Core::config_get_config($config_dir);
3937        $RA = undef;
3938        my $dont_store_passwords = 1;
3939        my $conf_t = ${$config}{'config'};
3940        {
3941                no warnings 'once';
3942                # The usage of $SVN::_Core::SVN_CONFIG_* variables
3943                # produces warnings that variables are used only once.
3944                # I had not found the better way to shut them up, so
3945                # the warnings of type 'once' are disabled in this block.
3946                if (SVN::_Core::svn_config_get_bool($conf_t,
3947                    $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
3948                    $SVN::_Core::SVN_CONFIG_OPTION_STORE_PASSWORDS,
3949                    1) == 0) {
3950                        SVN::_Core::svn_auth_set_parameter($baton,
3951                            $SVN::_Core::SVN_AUTH_PARAM_DONT_STORE_PASSWORDS,
3952                            bless (\$dont_store_passwords, "_p_void"));
3953                }
3954                if (SVN::_Core::svn_config_get_bool($conf_t,
3955                    $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
3956                    $SVN::_Core::SVN_CONFIG_OPTION_STORE_AUTH_CREDS,
3957                    1) == 0) {
3958                        $Git::SVN::Prompt::_no_auth_cache = 1;
3959                }
3960        } # no warnings 'once'
3961        my $self = SVN::Ra->new(url => escape_url($url), auth => $baton,
3962                              config => $config,
3963                              pool => SVN::Pool->new,
3964                              auth_provider_callbacks => $callbacks);
3965        $self->{url} = $url;
3966        $self->{svn_path} = $url;
3967        $self->{repos_root} = $self->get_repos_root;
3968        $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
3969        $self->{cache} = { check_path => { r => 0, data => {} },
3970                           get_dir => { r => 0, data => {} } };
3971        $RA = bless $self, $class;
3972}
3973
3974sub check_path {
3975        my ($self, $path, $r) = @_;
3976        my $cache = $self->{cache}->{check_path};
3977        if ($r == $cache->{r} && exists $cache->{data}->{$path}) {
3978                return $cache->{data}->{$path};
3979        }
3980        my $pool = SVN::Pool->new;
3981        my $t = $self->SUPER::check_path($path, $r, $pool);
3982        $pool->clear;
3983        if ($r != $cache->{r}) {
3984                %{$cache->{data}} = ();
3985                $cache->{r} = $r;
3986        }
3987        $cache->{data}->{$path} = $t;
3988}
3989
3990sub get_dir {
3991        my ($self, $dir, $r) = @_;
3992        my $cache = $self->{cache}->{get_dir};
3993        if ($r == $cache->{r}) {
3994                if (my $x = $cache->{data}->{$dir}) {
3995                        return wantarray ? @$x : $x->[0];
3996                }
3997        }
3998        my $pool = SVN::Pool->new;
3999        my ($d, undef, $props) = $self->SUPER::get_dir($dir, $r, $pool);
4000        my %dirents = map { $_ => { kind => $d->{$_}->kind } } keys %$d;
4001        $pool->clear;
4002        if ($r != $cache->{r}) {
4003                %{$cache->{data}} = ();
4004                $cache->{r} = $r;
4005        }
4006        $cache->{data}->{$dir} = [ \%dirents, $r, $props ];
4007        wantarray ? (\%dirents, $r, $props) : \%dirents;
4008}
4009
4010sub DESTROY {
4011        # do not call the real DESTROY since we store ourselves in $RA
4012}
4013
4014sub get_log {
4015        my ($self, @args) = @_;
4016        my $pool = SVN::Pool->new;
4017        splice(@args, 3, 1) if ($SVN::Core::VERSION le '1.2.0');
4018        my $ret = $self->SUPER::get_log(@args, $pool);
4019        $pool->clear;
4020        $ret;
4021}
4022
4023sub trees_match {
4024        my ($self, $url1, $rev1, $url2, $rev2) = @_;
4025        my $ctx = SVN::Client->new(auth => _auth_providers);
4026        my $out = IO::File->new_tmpfile;
4027
4028        # older SVN (1.1.x) doesn't take $pool as the last parameter for
4029        # $ctx->diff(), so we'll create a default one
4030        my $pool = SVN::Pool->new_default_sub;
4031
4032        $ra_invalid = 1; # this will open a new SVN::Ra connection to $url1
4033        $ctx->diff([], $url1, $rev1, $url2, $rev2, 1, 1, 0, $out, $out);
4034        $out->flush;
4035        my $ret = (($out->stat)[7] == 0);
4036        close $out or croak $!;
4037
4038        $ret;
4039}
4040
4041sub get_commit_editor {
4042        my ($self, $log, $cb, $pool) = @_;
4043        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
4044        $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
4045}
4046
4047sub gs_do_update {
4048        my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
4049        my $new = ($rev_a == $rev_b);
4050        my $path = $gs->{path};
4051
4052        if ($new && -e $gs->{index}) {
4053                unlink $gs->{index} or die
4054                  "Couldn't unlink index: $gs->{index}: $!\n";
4055        }
4056        my $pool = SVN::Pool->new;
4057        $editor->set_path_strip($path);
4058        my (@pc) = split m#/#, $path;
4059        my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
4060                                        1, $editor, $pool);
4061        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
4062
4063        # Since we can't rely on svn_ra_reparent being available, we'll
4064        # just have to do some magic with set_path to make it so
4065        # we only want a partial path.
4066        my $sp = '';
4067        my $final = join('/', @pc);
4068        while (@pc) {
4069                $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
4070                $sp .= '/' if length $sp;
4071                $sp .= shift @pc;
4072        }
4073        die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
4074
4075        $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
4076
4077        $reporter->finish_report($pool);
4078        $pool->clear;
4079        $editor->{git_commit_ok};
4080}
4081
4082# this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
4083# svn_ra_reparent didn't work before 1.4)
4084sub gs_do_switch {
4085        my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
4086        my $path = $gs->{path};
4087        my $pool = SVN::Pool->new;
4088
4089        my $full_url = $self->{url};
4090        my $old_url = $full_url;
4091        $full_url .= '/' . escape_uri_only($path) if length $path;
4092        my ($ra, $reparented);
4093
4094        if ($old_url =~ m#^svn(\+ssh)?://#) {
4095                $_[0] = undef;
4096                $self = undef;
4097                $RA = undef;
4098                $ra = Git::SVN::Ra->new($full_url);
4099                $ra_invalid = 1;
4100        } elsif ($old_url ne $full_url) {
4101                SVN::_Ra::svn_ra_reparent($self->{session}, $full_url, $pool);
4102                $self->{url} = $full_url;
4103                $reparented = 1;
4104        }
4105
4106        $ra ||= $self;
4107        $url_b = escape_url($url_b);
4108        my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
4109        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
4110        $reporter->set_path('', $rev_a, 0, @lock, $pool);
4111        $reporter->finish_report($pool);
4112
4113        if ($reparented) {
4114                SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
4115                $self->{url} = $old_url;
4116        }
4117
4118        $pool->clear;
4119        $editor->{git_commit_ok};
4120}
4121
4122sub longest_common_path {
4123        my ($gsv, $globs) = @_;
4124        my %common;
4125        my $common_max = scalar @$gsv;
4126
4127        foreach my $gs (@$gsv) {
4128                my @tmp = split m#/#, $gs->{path};
4129                my $p = '';
4130                foreach (@tmp) {
4131                        $p .= length($p) ? "/$_" : $_;
4132                        $common{$p} ||= 0;
4133                        $common{$p}++;
4134                }
4135        }
4136        $globs ||= [];
4137        $common_max += scalar @$globs;
4138        foreach my $glob (@$globs) {
4139                my @tmp = split m#/#, $glob->{path}->{left};
4140                my $p = '';
4141                foreach (@tmp) {
4142                        $p .= length($p) ? "/$_" : $_;
4143                        $common{$p} ||= 0;
4144                        $common{$p}++;
4145                }
4146        }
4147
4148        my $longest_path = '';
4149        foreach (sort {length $b <=> length $a} keys %common) {
4150                if ($common{$_} == $common_max) {
4151                        $longest_path = $_;
4152                        last;
4153                }
4154        }
4155        $longest_path;
4156}
4157
4158sub gs_fetch_loop_common {
4159        my ($self, $base, $head, $gsv, $globs) = @_;
4160        return if ($base > $head);
4161        my $inc = $_log_window_size;
4162        my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
4163        my $longest_path = longest_common_path($gsv, $globs);
4164        my $ra_url = $self->{url};
4165        while (1) {
4166                my %revs;
4167                my $err;
4168                my $err_handler = $SVN::Error::handler;
4169                $SVN::Error::handler = sub {
4170                        ($err) = @_;
4171                        skip_unknown_revs($err);
4172                };
4173                sub _cb {
4174                        my ($paths, $r, $author, $date, $log) = @_;
4175                        [ dup_changed_paths($paths),
4176                          { author => $author, date => $date, log => $log } ];
4177                }
4178                $self->get_log([$longest_path], $min, $max, 0, 1, 1,
4179                               sub { $revs{$_[1]} = _cb(@_) });
4180                if ($err && $max >= $head) {
4181                        print STDERR "Path '$longest_path' ",
4182                                     "was probably deleted:\n",
4183                                     $err->expanded_message,
4184                                     "\nWill attempt to follow ",
4185                                     "revisions r$min .. r$max ",
4186                                     "committed before the deletion\n";
4187                        my $hi = $max;
4188                        while (--$hi >= $min) {
4189                                my $ok;
4190                                $self->get_log([$longest_path], $min, $hi,
4191                                               0, 1, 1, sub {
4192                                               $ok ||= $_[1];
4193                                               $revs{$_[1]} = _cb(@_) });
4194                                if ($ok) {
4195                                        print STDERR "r$min .. r$ok OK\n";
4196                                        last;
4197                                }
4198                        }
4199                }
4200                $SVN::Error::handler = $err_handler;
4201
4202                my %exists = map { $_->{path} => $_ } @$gsv;
4203                foreach my $r (sort {$a <=> $b} keys %revs) {
4204                        my ($paths, $logged) = @{$revs{$r}};
4205
4206                        foreach my $gs ($self->match_globs(\%exists, $paths,
4207                                                           $globs, $r)) {
4208                                if ($gs->rev_map_max >= $r) {
4209                                        next;
4210                                }
4211                                next unless $gs->match_paths($paths, $r);
4212                                $gs->{logged_rev_props} = $logged;
4213                                if (my $last_commit = $gs->last_commit) {
4214                                        $gs->assert_index_clean($last_commit);
4215                                }
4216                                my $log_entry = $gs->do_fetch($paths, $r);
4217                                if ($log_entry) {
4218                                        $gs->do_git_commit($log_entry);
4219                                }
4220                                $INDEX_FILES{$gs->{index}} = 1;
4221                        }
4222                        foreach my $g (@$globs) {
4223                                my $k = "svn-remote.$g->{remote}." .
4224                                        "$g->{t}-maxRev";
4225                                Git::SVN::tmp_config($k, $r);
4226                        }
4227                        if ($ra_invalid) {
4228                                $_[0] = undef;
4229                                $self = undef;
4230                                $RA = undef;
4231                                $self = Git::SVN::Ra->new($ra_url);
4232                                $ra_invalid = undef;
4233                        }
4234                }
4235                # pre-fill the .rev_db since it'll eventually get filled in
4236                # with '0' x40 if something new gets committed
4237                foreach my $gs (@$gsv) {
4238                        next if $gs->rev_map_max >= $max;
4239                        next if defined $gs->rev_map_get($max);
4240                        $gs->rev_map_set($max, 0 x40);
4241                }
4242                foreach my $g (@$globs) {
4243                        my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
4244                        Git::SVN::tmp_config($k, $max);
4245                }
4246                last if $max >= $head;
4247                $min = $max + 1;
4248                $max += $inc;
4249                $max = $head if ($max > $head);
4250        }
4251        Git::SVN::gc();
4252}
4253
4254sub get_dir_globbed {
4255        my ($self, $left, $depth, $r) = @_;
4256
4257        my @x = eval { $self->get_dir($left, $r) };
4258        return unless scalar @x == 3;
4259        my $dirents = $x[0];
4260        my @finalents;
4261        foreach my $de (keys %$dirents) {
4262                next if $dirents->{$de}->{kind} != $SVN::Node::dir;
4263                if ($depth > 1) {
4264                        my @args = ("$left/$de", $depth - 1, $r);
4265                        foreach my $dir ($self->get_dir_globbed(@args)) {
4266                                push @finalents, "$de/$dir";
4267                        }
4268                } else {
4269                        push @finalents, $de;
4270                }
4271        }
4272        @finalents;
4273}
4274
4275sub match_globs {
4276        my ($self, $exists, $paths, $globs, $r) = @_;
4277
4278        sub get_dir_check {
4279                my ($self, $exists, $g, $r) = @_;
4280
4281                my @dirs = $self->get_dir_globbed($g->{path}->{left},
4282                                                  $g->{path}->{depth},
4283                                                  $r);
4284
4285                foreach my $de (@dirs) {
4286                        my $p = $g->{path}->full_path($de);
4287                        next if $exists->{$p};
4288                        next if (length $g->{path}->{right} &&
4289                                 ($self->check_path($p, $r) !=
4290                                  $SVN::Node::dir));
4291                        $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
4292                                         $g->{ref}->full_path($de), 1);
4293                }
4294        }
4295        foreach my $g (@$globs) {
4296                if (my $path = $paths->{"/$g->{path}->{left}"}) {
4297                        if ($path->{action} =~ /^[AR]$/) {
4298                                get_dir_check($self, $exists, $g, $r);
4299                        }
4300                }
4301                foreach (keys %$paths) {
4302                        if (/$g->{path}->{left_regex}/ &&
4303                            !/$g->{path}->{regex}/) {
4304                                next if $paths->{$_}->{action} !~ /^[AR]$/;
4305                                get_dir_check($self, $exists, $g, $r);
4306                        }
4307                        next unless /$g->{path}->{regex}/;
4308                        my $p = $1;
4309                        my $pathname = $g->{path}->full_path($p);
4310                        next if $exists->{$pathname};
4311                        next if ($self->check_path($pathname, $r) !=
4312                                 $SVN::Node::dir);
4313                        $exists->{$pathname} = Git::SVN->init(
4314                                              $self->{url}, $pathname, undef,
4315                                              $g->{ref}->full_path($p), 1);
4316                }
4317                my $c = '';
4318                foreach (split m#/#, $g->{path}->{left}) {
4319                        $c .= "/$_";
4320                        next unless ($paths->{$c} &&
4321                                     ($paths->{$c}->{action} =~ /^[AR]$/));
4322                        get_dir_check($self, $exists, $g, $r);
4323                }
4324        }
4325        values %$exists;
4326}
4327
4328sub minimize_url {
4329        my ($self) = @_;
4330        return $self->{url} if ($self->{url} eq $self->{repos_root});
4331        my $url = $self->{repos_root};
4332        my @components = split(m!/!, $self->{svn_path});
4333        my $c = '';
4334        do {
4335                $url .= "/$c" if length $c;
4336                eval { (ref $self)->new($url)->get_latest_revnum };
4337        } while ($@ && ($c = shift @components));
4338        $url;
4339}
4340
4341sub can_do_switch {
4342        my $self = shift;
4343        unless (defined $can_do_switch) {
4344                my $pool = SVN::Pool->new;
4345                my $rep = eval {
4346                        $self->do_switch(1, '', 0, $self->{url},
4347                                         SVN::Delta::Editor->new, $pool);
4348                };
4349                if ($@) {
4350                        $can_do_switch = 0;
4351                } else {
4352                        $rep->abort_report($pool);
4353                        $can_do_switch = 1;
4354                }
4355                $pool->clear;
4356        }
4357        $can_do_switch;
4358}
4359
4360sub skip_unknown_revs {
4361        my ($err) = @_;
4362        my $errno = $err->apr_err();
4363        # Maybe the branch we're tracking didn't
4364        # exist when the repo started, so it's
4365        # not an error if it doesn't, just continue
4366        #
4367        # Wonderfully consistent library, eh?
4368        # 160013 - svn:// and file://
4369        # 175002 - http(s)://
4370        # 175007 - http(s):// (this repo required authorization, too...)
4371        #   More codes may be discovered later...
4372        if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
4373                my $err_key = $err->expanded_message;
4374                # revision numbers change every time, filter them out
4375                $err_key =~ s/\d+/\0/g;
4376                $err_key = "$errno\0$err_key";
4377                unless ($ignored_err{$err_key}) {
4378                        warn "W: Ignoring error from SVN, path probably ",
4379                             "does not exist: ($errno): ",
4380                             $err->expanded_message,"\n";
4381                        warn "W: Do not be alarmed at the above message ",
4382                             "git-svn is just searching aggressively for ",
4383                             "old history.\n",
4384                             "This may take a while on large repositories\n";
4385                        $ignored_err{$err_key} = 1;
4386                }
4387                return;
4388        }
4389        die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
4390}
4391
4392# svn_log_changed_path_t objects passed to get_log are likely to be
4393# overwritten even if only the refs are copied to an external variable,
4394# so we should dup the structures in their entirety.  Using an externally
4395# passed pool (instead of our temporary and quickly cleared pool in
4396# Git::SVN::Ra) does not help matters at all...
4397sub dup_changed_paths {
4398        my ($paths) = @_;
4399        return undef unless $paths;
4400        my %ret;
4401        foreach my $p (keys %$paths) {
4402                my $i = $paths->{$p};
4403                my %s = map { $_ => $i->$_ }
4404                              qw/copyfrom_path copyfrom_rev action/;
4405                $ret{$p} = \%s;
4406        }
4407        \%ret;
4408}
4409
4410package Git::SVN::Log;
4411use strict;
4412use warnings;
4413use POSIX qw/strftime/;
4414use constant commit_log_separator => ('-' x 72) . "\n";
4415use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
4416            %rusers $show_commit $incremental/;
4417my $l_fmt;
4418
4419sub cmt_showable {
4420        my ($c) = @_;
4421        return 1 if defined $c->{r};
4422
4423        # big commit message got truncated by the 16k pretty buffer in rev-list
4424        if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
4425                                $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
4426                @{$c->{l}} = ();
4427                my @log = command(qw/cat-file commit/, $c->{c});
4428
4429                # shift off the headers
4430                shift @log while ($log[0] ne '');
4431                shift @log;
4432
4433                # TODO: make $c->{l} not have a trailing newline in the future
4434                @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
4435
4436                (undef, $c->{r}, undef) = ::extract_metadata(
4437                                (grep(/^git-svn-id: /, @log))[-1]);
4438        }
4439        return defined $c->{r};
4440}
4441
4442sub log_use_color {
4443        return $color || Git->repository->get_colorbool('color.diff');
4444}
4445
4446sub git_svn_log_cmd {
4447        my ($r_min, $r_max, @args) = @_;
4448        my $head = 'HEAD';
4449        my (@files, @log_opts);
4450        foreach my $x (@args) {
4451                if ($x eq '--' || @files) {
4452                        push @files, $x;
4453                } else {
4454                        if (::verify_ref("$x^0")) {
4455                                $head = $x;
4456                        } else {
4457                                push @log_opts, $x;
4458                        }
4459                }
4460        }
4461
4462        my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
4463        $gs ||= Git::SVN->_new;
4464        my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
4465                   $gs->refname);
4466        push @cmd, '-r' unless $non_recursive;
4467        push @cmd, qw/--raw --name-status/ if $verbose;
4468        push @cmd, '--color' if log_use_color();
4469        push @cmd, @log_opts;
4470        if (defined $r_max && $r_max == $r_min) {
4471                push @cmd, '--max-count=1';
4472                if (my $c = $gs->rev_map_get($r_max)) {
4473                        push @cmd, $c;
4474                }
4475        } elsif (defined $r_max) {
4476                if ($r_max < $r_min) {
4477                        ($r_min, $r_max) = ($r_max, $r_min);
4478                }
4479                my (undef, $c_max) = $gs->find_rev_before($r_max, 1, $r_min);
4480                my (undef, $c_min) = $gs->find_rev_after($r_min, 1, $r_max);
4481                # If there are no commits in the range, both $c_max and $c_min
4482                # will be undefined.  If there is at least 1 commit in the
4483                # range, both will be defined.
4484                return () if !defined $c_min || !defined $c_max;
4485                if ($c_min eq $c_max) {
4486                        push @cmd, '--max-count=1', $c_min;
4487                } else {
4488                        push @cmd, '--boundary', "$c_min..$c_max";
4489                }
4490        }
4491        return (@cmd, @files);
4492}
4493
4494# adapted from pager.c
4495sub config_pager {
4496        $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
4497        if (!defined $pager) {
4498                $pager = 'less';
4499        } elsif (length $pager == 0 || $pager eq 'cat') {
4500                $pager = undef;
4501        }
4502        $ENV{GIT_PAGER_IN_USE} = defined($pager);
4503}
4504
4505sub run_pager {
4506        return unless -t *STDOUT && defined $pager;
4507        pipe my ($rfd, $wfd) or return;
4508        defined(my $pid = fork) or ::fatal "Can't fork: $!";
4509        if (!$pid) {
4510                open STDOUT, '>&', $wfd or
4511                                     ::fatal "Can't redirect to stdout: $!";
4512                return;
4513        }
4514        open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!";
4515        $ENV{LESS} ||= 'FRSX';
4516        exec $pager or ::fatal "Can't run pager: $! ($pager)";
4517}
4518
4519sub format_svn_date {
4520        return strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)", localtime(shift));
4521}
4522
4523sub parse_git_date {
4524        my ($t, $tz) = @_;
4525        # Date::Parse isn't in the standard Perl distro :(
4526        if ($tz =~ s/^\+//) {
4527                $t += tz_to_s_offset($tz);
4528        } elsif ($tz =~ s/^\-//) {
4529                $t -= tz_to_s_offset($tz);
4530        }
4531        return $t;
4532}
4533
4534sub set_local_timezone {
4535        if (defined $TZ) {
4536                $ENV{TZ} = $TZ;
4537        } else {
4538                delete $ENV{TZ};
4539        }
4540}
4541
4542sub tz_to_s_offset {
4543        my ($tz) = @_;
4544        $tz =~ s/(\d\d)$//;
4545        return ($1 * 60) + ($tz * 3600);
4546}
4547
4548sub get_author_info {
4549        my ($dest, $author, $t, $tz) = @_;
4550        $author =~ s/(?:^\s*|\s*$)//g;
4551        $dest->{a_raw} = $author;
4552        my $au;
4553        if ($::_authors) {
4554                $au = $rusers{$author} || undef;
4555        }
4556        if (!$au) {
4557                ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
4558        }
4559        $dest->{t} = $t;
4560        $dest->{tz} = $tz;
4561        $dest->{a} = $au;
4562        $dest->{t_utc} = parse_git_date($t, $tz);
4563}
4564
4565sub process_commit {
4566        my ($c, $r_min, $r_max, $defer) = @_;
4567        if (defined $r_min && defined $r_max) {
4568                if ($r_min == $c->{r} && $r_min == $r_max) {
4569                        show_commit($c);
4570                        return 0;
4571                }
4572                return 1 if $r_min == $r_max;
4573                if ($r_min < $r_max) {
4574                        # we need to reverse the print order
4575                        return 0 if (defined $limit && --$limit < 0);
4576                        push @$defer, $c;
4577                        return 1;
4578                }
4579                if ($r_min != $r_max) {
4580                        return 1 if ($r_min < $c->{r});
4581                        return 1 if ($r_max > $c->{r});
4582                }
4583        }
4584        return 0 if (defined $limit && --$limit < 0);
4585        show_commit($c);
4586        return 1;
4587}
4588
4589sub show_commit {
4590        my $c = shift;
4591        if ($oneline) {
4592                my $x = "\n";
4593                if (my $l = $c->{l}) {
4594                        while ($l->[0] =~ /^\s*$/) { shift @$l }
4595                        $x = $l->[0];
4596                }
4597                $l_fmt ||= 'A' . length($c->{r});
4598                print 'r',pack($l_fmt, $c->{r}),' | ';
4599                print "$c->{c} | " if $show_commit;
4600                print $x;
4601        } else {
4602                show_commit_normal($c);
4603        }
4604}
4605
4606sub show_commit_changed_paths {
4607        my ($c) = @_;
4608        return unless $c->{changed};
4609        print "Changed paths:\n", @{$c->{changed}};
4610}
4611
4612sub show_commit_normal {
4613        my ($c) = @_;
4614        print commit_log_separator, "r$c->{r} | ";
4615        print "$c->{c} | " if $show_commit;
4616        print "$c->{a} | ", format_svn_date($c->{t_utc}), ' | ';
4617        my $nr_line = 0;
4618
4619        if (my $l = $c->{l}) {
4620                while ($l->[$#$l] eq "\n" && $#$l > 0
4621                                          && $l->[($#$l - 1)] eq "\n") {
4622                        pop @$l;
4623                }
4624                $nr_line = scalar @$l;
4625                if (!$nr_line) {
4626                        print "1 line\n\n\n";
4627                } else {
4628                        if ($nr_line == 1) {
4629                                $nr_line = '1 line';
4630                        } else {
4631                                $nr_line .= ' lines';
4632                        }
4633                        print $nr_line, "\n";
4634                        show_commit_changed_paths($c);
4635                        print "\n";
4636                        print $_ foreach @$l;
4637                }
4638        } else {
4639                print "1 line\n";
4640                show_commit_changed_paths($c);
4641                print "\n";
4642
4643        }
4644        foreach my $x (qw/raw stat diff/) {
4645                if ($c->{$x}) {
4646                        print "\n";
4647                        print $_ foreach @{$c->{$x}}
4648                }
4649        }
4650}
4651
4652sub cmd_show_log {
4653        my (@args) = @_;
4654        my ($r_min, $r_max);
4655        my $r_last = -1; # prevent dupes
4656        set_local_timezone();
4657        if (defined $::_revision) {
4658                if ($::_revision =~ /^(\d+):(\d+)$/) {
4659                        ($r_min, $r_max) = ($1, $2);
4660                } elsif ($::_revision =~ /^\d+$/) {
4661                        $r_min = $r_max = $::_revision;
4662                } else {
4663                        ::fatal "-r$::_revision is not supported, use ",
4664                                "standard 'git log' arguments instead";
4665                }
4666        }
4667
4668        config_pager();
4669        @args = git_svn_log_cmd($r_min, $r_max, @args);
4670        if (!@args) {
4671                print commit_log_separator unless $incremental || $oneline;
4672                return;
4673        }
4674        my $log = command_output_pipe(@args);
4675        run_pager();
4676        my (@k, $c, $d, $stat);
4677        my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
4678        while (<$log>) {
4679                if (/^${esc_color}commit -?($::sha1_short)/o) {
4680                        my $cmt = $1;
4681                        if ($c && cmt_showable($c) && $c->{r} != $r_last) {
4682                                $r_last = $c->{r};
4683                                process_commit($c, $r_min, $r_max, \@k) or
4684                                                                goto out;
4685                        }
4686                        $d = undef;
4687                        $c = { c => $cmt };
4688                } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
4689                        get_author_info($c, $1, $2, $3);
4690                } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
4691                        # ignore
4692                } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
4693                        push @{$c->{raw}}, $_;
4694                } elsif (/^${esc_color}[ACRMDT]\t/) {
4695                        # we could add $SVN->{svn_path} here, but that requires
4696                        # remote access at the moment (repo_path_split)...
4697                        s#^(${esc_color})([ACRMDT])\t#$1   $2 #o;
4698                        push @{$c->{changed}}, $_;
4699                } elsif (/^${esc_color}diff /o) {
4700                        $d = 1;
4701                        push @{$c->{diff}}, $_;
4702                } elsif ($d) {
4703                        push @{$c->{diff}}, $_;
4704                } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
4705                          $esc_color*[\+\-]*$esc_color$/x) {
4706                        $stat = 1;
4707                        push @{$c->{stat}}, $_;
4708                } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
4709                        push @{$c->{stat}}, $_;
4710                        $stat = undef;
4711                } elsif (/^${esc_color}    (git-svn-id:.+)$/o) {
4712                        ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
4713                } elsif (s/^${esc_color}    //o) {
4714                        push @{$c->{l}}, $_;
4715                }
4716        }
4717        if ($c && defined $c->{r} && $c->{r} != $r_last) {
4718                $r_last = $c->{r};
4719                process_commit($c, $r_min, $r_max, \@k);
4720        }
4721        if (@k) {
4722                ($r_min, $r_max) = ($r_max, $r_min);
4723                process_commit($_, $r_min, $r_max) foreach reverse @k;
4724        }
4725out:
4726        close $log;
4727        print commit_log_separator unless $incremental || $oneline;
4728}
4729
4730sub cmd_blame {
4731        my $path = pop;
4732
4733        config_pager();
4734        run_pager();
4735
4736        my ($fh, $ctx, $rev);
4737
4738        if ($_git_format) {
4739                ($fh, $ctx) = command_output_pipe('blame', @_, $path);
4740                while (my $line = <$fh>) {
4741                        if ($line =~ /^\^?([[:xdigit:]]+)\s/) {
4742                                # Uncommitted edits show up as a rev ID of
4743                                # all zeros, which we can't look up with
4744                                # cmt_metadata
4745                                if ($1 !~ /^0+$/) {
4746                                        (undef, $rev, undef) =
4747                                                ::cmt_metadata($1);
4748                                        $rev = '0' if (!$rev);
4749                                } else {
4750                                        $rev = '0';
4751                                }
4752                                $rev = sprintf('%-10s', $rev);
4753                                $line =~ s/^\^?[[:xdigit:]]+(\s)/$rev$1/;
4754                        }
4755                        print $line;
4756                }
4757        } else {
4758                ($fh, $ctx) = command_output_pipe('blame', '-p', @_, 'HEAD',
4759                                                  '--', $path);
4760                my ($sha1);
4761                my %authors;
4762                while (my $line = <$fh>) {
4763                        if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
4764                                $sha1 = $1;
4765                                (undef, $rev, undef) = ::cmt_metadata($1);
4766                                $rev = '0' if (!$rev);
4767                        }
4768                        elsif ($line =~ /^author (.*)/) {
4769                                $authors{$rev} = $1;
4770                                $authors{$rev} =~ s/\s/_/g;
4771                        }
4772                        elsif ($line =~ /^\t(.*)$/) {
4773                                printf("%6s %10s %s\n", $rev, $authors{$rev}, $1);
4774                        }
4775                }
4776        }
4777        command_close_pipe($fh, $ctx);
4778}
4779
4780package Git::SVN::Migration;
4781# these version numbers do NOT correspond to actual version numbers
4782# of git nor git-svn.  They are just relative.
4783#
4784# v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
4785#
4786# v1 layout: .git/$id/info/url, refs/remotes/$id
4787#
4788# v2 layout: .git/svn/$id/info/url, refs/remotes/$id
4789#
4790# v3 layout: .git/svn/$id, refs/remotes/$id
4791#            - info/url may remain for backwards compatibility
4792#            - this is what we migrate up to this layout automatically,
4793#            - this will be used by git svn init on single branches
4794# v3.1 layout (auto migrated):
4795#            - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
4796#              for backwards compatibility
4797#
4798# v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
4799#            - this is only created for newly multi-init-ed
4800#              repositories.  Similar in spirit to the
4801#              --use-separate-remotes option in git-clone (now default)
4802#            - we do not automatically migrate to this (following
4803#              the example set by core git)
4804#
4805# v5 layout: .rev_db.$UUID => .rev_map.$UUID
4806#            - newer, more-efficient format that uses 24-bytes per record
4807#              with no filler space.
4808#            - use xxd -c24 < .rev_map.$UUID to view and debug
4809#            - This is a one-way migration, repositories updated to the
4810#              new format will not be able to use old git-svn without
4811#              rebuilding the .rev_db.  Rebuilding the rev_db is not
4812#              possible if noMetadata or useSvmProps are set; but should
4813#              be no problem for users that use the (sensible) defaults.
4814use strict;
4815use warnings;
4816use Carp qw/croak/;
4817use File::Path qw/mkpath/;
4818use File::Basename qw/dirname basename/;
4819use vars qw/$_minimize/;
4820
4821sub migrate_from_v0 {
4822        my $git_dir = $ENV{GIT_DIR};
4823        return undef unless -d $git_dir;
4824        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
4825        my $migrated = 0;
4826        while (<$fh>) {
4827                chomp;
4828                my ($id, $orig_ref) = ($_, $_);
4829                next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
4830                next unless -f "$git_dir/$id/info/url";
4831                my $new_ref = "refs/remotes/$id";
4832                if (::verify_ref("$new_ref^0")) {
4833                        print STDERR "W: $orig_ref is probably an old ",
4834                                     "branch used by an ancient version of ",
4835                                     "git-svn.\n",
4836                                     "However, $new_ref also exists.\n",
4837                                     "We will not be able ",
4838                                     "to use this branch until this ",
4839                                     "ambiguity is resolved.\n";
4840                        next;
4841                }
4842                print STDERR "Migrating from v0 layout...\n" if !$migrated;
4843                print STDERR "Renaming ref: $orig_ref => $new_ref\n";
4844                command_noisy('update-ref', $new_ref, $orig_ref);
4845                command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
4846                $migrated++;
4847        }
4848        command_close_pipe($fh, $ctx);
4849        print STDERR "Done migrating from v0 layout...\n" if $migrated;
4850        $migrated;
4851}
4852
4853sub migrate_from_v1 {
4854        my $git_dir = $ENV{GIT_DIR};
4855        my $migrated = 0;
4856        return $migrated unless -d $git_dir;
4857        my $svn_dir = "$git_dir/svn";
4858
4859        # just in case somebody used 'svn' as their $id at some point...
4860        return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
4861
4862        print STDERR "Migrating from a git-svn v1 layout...\n";
4863        mkpath([$svn_dir]);
4864        print STDERR "Data from a previous version of git-svn exists, but\n\t",
4865                     "$svn_dir\n\t(required for this version ",
4866                     "($::VERSION) of git-svn) does not exist.\n";
4867        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
4868        while (<$fh>) {
4869                my $x = $_;
4870                next unless $x =~ s#^refs/remotes/##;
4871                chomp $x;
4872                next unless -f "$git_dir/$x/info/url";
4873                my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
4874                next unless $u;
4875                my $dn = dirname("$git_dir/svn/$x");
4876                mkpath([$dn]) unless -d $dn;
4877                if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
4878                        mkpath(["$git_dir/svn/svn"]);
4879                        print STDERR " - $git_dir/$x/info => ",
4880                                        "$git_dir/svn/$x/info\n";
4881                        rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
4882                               croak "$!: $x";
4883                        # don't worry too much about these, they probably
4884                        # don't exist with repos this old (save for index,
4885                        # and we can easily regenerate that)
4886                        foreach my $f (qw/unhandled.log index .rev_db/) {
4887                                rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
4888                        }
4889                } else {
4890                        print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
4891                        rename "$git_dir/$x", "$git_dir/svn/$x" or
4892                               croak "$!: $x";
4893                }
4894                $migrated++;
4895        }
4896        command_close_pipe($fh, $ctx);
4897        print STDERR "Done migrating from a git-svn v1 layout\n";
4898        $migrated;
4899}
4900
4901sub read_old_urls {
4902        my ($l_map, $pfx, $path) = @_;
4903        my @dir;
4904        foreach (<$path/*>) {
4905                if (-r "$_/info/url") {
4906                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
4907                        my $ref_id = $pfx . basename $_;
4908                        my $url = ::file_to_s("$_/info/url");
4909                        $l_map->{$ref_id} = $url;
4910                } elsif (-d $_) {
4911                        push @dir, $_;
4912                }
4913        }
4914        foreach (@dir) {
4915                my $x = $_;
4916                $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
4917                read_old_urls($l_map, $x, $_);
4918        }
4919}
4920
4921sub migrate_from_v2 {
4922        my @cfg = command(qw/config -l/);
4923        return if grep /^svn-remote\..+\.url=/, @cfg;
4924        my %l_map;
4925        read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
4926        my $migrated = 0;
4927
4928        foreach my $ref_id (sort keys %l_map) {
4929                eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
4930                if ($@) {
4931                        Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
4932                }
4933                $migrated++;
4934        }
4935        $migrated;
4936}
4937
4938sub minimize_connections {
4939        my $r = Git::SVN::read_all_remotes();
4940        my $new_urls = {};
4941        my $root_repos = {};
4942        foreach my $repo_id (keys %$r) {
4943                my $url = $r->{$repo_id}->{url} or next;
4944                my $fetch = $r->{$repo_id}->{fetch} or next;
4945                my $ra = Git::SVN::Ra->new($url);
4946
4947                # skip existing cases where we already connect to the root
4948                if (($ra->{url} eq $ra->{repos_root}) ||
4949                    ($ra->{repos_root} eq $repo_id)) {
4950                        $root_repos->{$ra->{url}} = $repo_id;
4951                        next;
4952                }
4953
4954                my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
4955                my $root_path = $ra->{url};
4956                $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
4957                foreach my $path (keys %$fetch) {
4958                        my $ref_id = $fetch->{$path};
4959                        my $gs = Git::SVN->new($ref_id, $repo_id, $path);
4960
4961                        # make sure we can read when connecting to
4962                        # a higher level of a repository
4963                        my ($last_rev, undef) = $gs->last_rev_commit;
4964                        if (!defined $last_rev) {
4965                                $last_rev = eval {
4966                                        $root_ra->get_latest_revnum;
4967                                };
4968                                next if $@;
4969                        }
4970                        my $new = $root_path;
4971                        $new .= length $path ? "/$path" : '';
4972                        eval {
4973                                $root_ra->get_log([$new], $last_rev, $last_rev,
4974                                                  0, 0, 1, sub { });
4975                        };
4976                        next if $@;
4977                        $new_urls->{$ra->{repos_root}}->{$new} =
4978                                { ref_id => $ref_id,
4979                                  old_repo_id => $repo_id,
4980                                  old_path => $path };
4981                }
4982        }
4983
4984        my @emptied;
4985        foreach my $url (keys %$new_urls) {
4986                # see if we can re-use an existing [svn-remote "repo_id"]
4987                # instead of creating a(n ugly) new section:
4988                my $repo_id = $root_repos->{$url} || $url;
4989
4990                my $fetch = $new_urls->{$url};
4991                foreach my $path (keys %$fetch) {
4992                        my $x = $fetch->{$path};
4993                        Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
4994                        my $pfx = "svn-remote.$x->{old_repo_id}";
4995
4996                        my $old_fetch = quotemeta("$x->{old_path}:".
4997                                                  "refs/remotes/$x->{ref_id}");
4998                        command_noisy(qw/config --unset/,
4999                                      "$pfx.fetch", '^'. $old_fetch . '$');
5000                        delete $r->{$x->{old_repo_id}}->
5001                               {fetch}->{$x->{old_path}};
5002                        if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
5003                                command_noisy(qw/config --unset/,
5004                                              "$pfx.url");
5005                                push @emptied, $x->{old_repo_id}
5006                        }
5007                }
5008        }
5009        if (@emptied) {
5010                my $file = $ENV{GIT_CONFIG} || $ENV{GIT_CONFIG_LOCAL} ||
5011                           "$ENV{GIT_DIR}/config";
5012                print STDERR <<EOF;
5013The following [svn-remote] sections in your config file ($file) are empty
5014and can be safely removed:
5015EOF
5016                print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
5017        }
5018}
5019
5020sub migration_check {
5021        migrate_from_v0();
5022        migrate_from_v1();
5023        migrate_from_v2();
5024        minimize_connections() if $_minimize;
5025}
5026
5027package Git::IndexInfo;
5028use strict;
5029use warnings;
5030use Git qw/command_input_pipe command_close_pipe/;
5031
5032sub new {
5033        my ($class) = @_;
5034        my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
5035        bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
5036}
5037
5038sub remove {
5039        my ($self, $path) = @_;
5040        if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
5041                return ++$self->{nr};
5042        }
5043        undef;
5044}
5045
5046sub update {
5047        my ($self, $mode, $hash, $path) = @_;
5048        if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
5049                return ++$self->{nr};
5050        }
5051        undef;
5052}
5053
5054sub DESTROY {
5055        my ($self) = @_;
5056        command_close_pipe($self->{gui}, $self->{ctx});
5057}
5058
5059package Git::SVN::GlobSpec;
5060use strict;
5061use warnings;
5062
5063sub new {
5064        my ($class, $glob) = @_;
5065        my $re = $glob;
5066        $re =~ s!/+$!!g; # no need for trailing slashes
5067        $re =~ m!^([^*]*)(\*(?:/\*)*)([^*]*)$!;
5068        my $temp = $re;
5069        my ($left, $right) = ($1, $3);
5070        $re = $2;
5071        my $depth = $re =~ tr/*/*/;
5072        if ($depth != $temp =~ tr/*/*/) {
5073                die "Only one set of wildcard directories " .
5074                        "(e.g. '*' or '*/*/*') is supported: '$glob'\n";
5075        }
5076        if ($depth == 0) {
5077                die "One '*' is needed for glob: '$glob'\n";
5078        }
5079        $re =~ s!\*!\[^/\]*!g;
5080        $re = quotemeta($left) . "($re)" . quotemeta($right);
5081        if (length $left && !($left =~ s!/+$!!g)) {
5082                die "Missing trailing '/' on left side of: '$glob' ($left)\n";
5083        }
5084        if (length $right && !($right =~ s!^/+!!g)) {
5085                die "Missing leading '/' on right side of: '$glob' ($right)\n";
5086        }
5087        my $left_re = qr/^\/\Q$left\E(\/|$)/;
5088        bless { left => $left, right => $right, left_regex => $left_re,
5089                regex => qr/$re/, glob => $glob, depth => $depth }, $class;
5090}
5091
5092sub full_path {
5093        my ($self, $path) = @_;
5094        return (length $self->{left} ? "$self->{left}/" : '') .
5095               $path . (length $self->{right} ? "/$self->{right}" : '');
5096}
5097
5098__END__
5099
5100Data structures:
5101
5102
5103$remotes = { # returned by read_all_remotes()
5104        'svn' => {
5105                # svn-remote.svn.url=https://svn.musicpd.org
5106                url => 'https://svn.musicpd.org',
5107                # svn-remote.svn.fetch=mpd/trunk:trunk
5108                fetch => {
5109                        'mpd/trunk' => 'trunk',
5110                },
5111                # svn-remote.svn.tags=mpd/tags/*:tags/*
5112                tags => {
5113                        path => {
5114                                left => 'mpd/tags',
5115                                right => '',
5116                                regex => qr!mpd/tags/([^/]+)$!,
5117                                glob => 'tags/*',
5118                        },
5119                        ref => {
5120                                left => 'tags',
5121                                right => '',
5122                                regex => qr!tags/([^/]+)$!,
5123                                glob => 'tags/*',
5124                        },
5125                }
5126        }
5127};
5128
5129$log_entry hashref as returned by libsvn_log_entry()
5130{
5131        log => 'whitespace-formatted log entry
5132',                                              # trailing newline is preserved
5133        revision => '8',                        # integer
5134        date => '2004-02-24T17:01:44.108345Z',  # commit date
5135        author => 'committer name'
5136};
5137
5138
5139# this is generated by generate_diff();
5140@mods = array of diff-index line hashes, each element represents one line
5141        of diff-index output
5142
5143diff-index line ($m hash)
5144{
5145        mode_a => first column of diff-index output, no leading ':',
5146        mode_b => second column of diff-index output,
5147        sha1_b => sha1sum of the final blob,
5148        chg => change type [MCRADT],
5149        file_a => original file name of a file (iff chg is 'C' or 'R')
5150        file_b => new/current file name of a file (any chg)
5151}
5152;
5153
5154# retval of read_url_paths{,_all}();
5155$l_map = {
5156        # repository root url
5157        'https://svn.musicpd.org' => {
5158                # repository path               # GIT_SVN_ID
5159                'mpd/trunk'             =>      'trunk',
5160                'mpd/tags/0.11.5'       =>      'tags/0.11.5',
5161        },
5162}
5163
5164Notes:
5165        I don't trust the each() function on unless I created %hash myself
5166        because the internal iterator may not have started at base.