git-svn.perlon commit Merge branch 'maint-1.6.3' into maint-1.6.4 (814035c)
   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 $_authors_prog %users/;
   9$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$Git::SVN::_minimize_url = 'unset';
  23
  24$Git::SVN::Log::TZ = $ENV{TZ};
  25$ENV{TZ} = 'UTC';
  26$| = 1; # unbuffer STDOUT
  27
  28sub fatal (@) { print STDERR "@_\n"; exit 1 }
  29require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
  30require SVN::Ra;
  31require SVN::Delta;
  32if ($SVN::Core::VERSION lt '1.1.0') {
  33        fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)";
  34}
  35my $can_compress = eval { require Compress::Zlib; 1};
  36push @Git::SVN::Ra::ISA, 'SVN::Ra';
  37push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
  38push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
  39use Carp qw/croak/;
  40use Digest::MD5;
  41use IO::File qw//;
  42use File::Basename qw/dirname basename/;
  43use File::Path qw/mkpath/;
  44use File::Spec;
  45use File::Find;
  46use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev/;
  47use IPC::Open3;
  48use Git;
  49
  50BEGIN {
  51        # import functions from Git into our packages, en masse
  52        no strict 'refs';
  53        foreach (qw/command command_oneline command_noisy command_output_pipe
  54                    command_input_pipe command_close_pipe
  55                    command_bidi_pipe command_close_bidi_pipe/) {
  56                for my $package ( qw(SVN::Git::Editor SVN::Git::Fetcher
  57                        Git::SVN::Migration Git::SVN::Log Git::SVN),
  58                        __PACKAGE__) {
  59                        *{"${package}::$_"} = \&{"Git::$_"};
  60                }
  61        }
  62}
  63
  64my ($SVN);
  65
  66$sha1 = qr/[a-f\d]{40}/;
  67$sha1_short = qr/[a-f\d]{4,40}/;
  68my ($_stdin, $_help, $_edit,
  69        $_message, $_file, $_branch_dest,
  70        $_template, $_shared,
  71        $_version, $_fetch_all, $_no_rebase, $_fetch_parent,
  72        $_merge, $_strategy, $_dry_run, $_local,
  73        $_prefix, $_no_checkout, $_url, $_verbose,
  74        $_git_format, $_commit_url, $_tag);
  75$Git::SVN::_follow_parent = 1;
  76$_q ||= 0;
  77my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
  78                    'config-dir=s' => \$Git::SVN::Ra::config_dir,
  79                    'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache,
  80                    'ignore-paths=s' => \$SVN::Git::Fetcher::_ignore_regex );
  81my %fc_opts = ( 'follow-parent|follow!' => \$Git::SVN::_follow_parent,
  82                'authors-file|A=s' => \$_authors,
  83                'authors-prog=s' => \$_authors_prog,
  84                'repack:i' => \$Git::SVN::_repack,
  85                'noMetadata' => \$Git::SVN::_no_metadata,
  86                'useSvmProps' => \$Git::SVN::_use_svm_props,
  87                'useSvnsyncProps' => \$Git::SVN::_use_svnsync_props,
  88                'log-window-size=i' => \$Git::SVN::Ra::_log_window_size,
  89                'no-checkout' => \$_no_checkout,
  90                'quiet|q+' => \$_q,
  91                'repack-flags|repack-args|repack-opts=s' =>
  92                   \$Git::SVN::_repack_flags,
  93                'use-log-author' => \$Git::SVN::_use_log_author,
  94                'add-author-from' => \$Git::SVN::_add_author_from,
  95                'localtime' => \$Git::SVN::_localtime,
  96                %remote_opts );
  97
  98my ($_trunk, @_tags, @_branches, $_stdlayout);
  99my %icv;
 100my %init_opts = ( 'template=s' => \$_template, 'shared:s' => \$_shared,
 101                  'trunk|T=s' => \$_trunk, 'tags|t=s@' => \@_tags,
 102                  'branches|b=s@' => \@_branches, 'prefix=s' => \$_prefix,
 103                  'stdlayout|s' => \$_stdlayout,
 104                  'minimize-url|m!' => \$Git::SVN::_minimize_url,
 105                  'no-metadata' => sub { $icv{noMetadata} = 1 },
 106                  'use-svm-props' => sub { $icv{useSvmProps} = 1 },
 107                  'use-svnsync-props' => sub { $icv{useSvnsyncProps} = 1 },
 108                  'rewrite-root=s' => sub { $icv{rewriteRoot} = $_[1] },
 109                  %remote_opts );
 110my %cmt_opts = ( 'edit|e' => \$_edit,
 111                'rmdir' => \$SVN::Git::Editor::_rmdir,
 112                'find-copies-harder' => \$SVN::Git::Editor::_find_copies_harder,
 113                'l=i' => \$SVN::Git::Editor::_rename_limit,
 114                'copy-similarity|C=i'=> \$SVN::Git::Editor::_cp_similarity
 115);
 116
 117my %cmd = (
 118        fetch => [ \&cmd_fetch, "Download new revisions from SVN",
 119                        { 'revision|r=s' => \$_revision,
 120                          'fetch-all|all' => \$_fetch_all,
 121                          'parent|p' => \$_fetch_parent,
 122                           %fc_opts } ],
 123        clone => [ \&cmd_clone, "Initialize and fetch revisions",
 124                        { 'revision|r=s' => \$_revision,
 125                           %fc_opts, %init_opts } ],
 126        init => [ \&cmd_init, "Initialize a repo for tracking" .
 127                          " (requires URL argument)",
 128                          \%init_opts ],
 129        'multi-init' => [ \&cmd_multi_init,
 130                          "Deprecated alias for ".
 131                          "'$0 init -T<trunk> -b<branches> -t<tags>'",
 132                          \%init_opts ],
 133        dcommit => [ \&cmd_dcommit,
 134                     'Commit several diffs to merge with upstream',
 135                        { 'merge|m|M' => \$_merge,
 136                          'strategy|s=s' => \$_strategy,
 137                          'verbose|v' => \$_verbose,
 138                          'dry-run|n' => \$_dry_run,
 139                          'fetch-all|all' => \$_fetch_all,
 140                          'commit-url=s' => \$_commit_url,
 141                          'revision|r=i' => \$_revision,
 142                          'no-rebase' => \$_no_rebase,
 143                        %cmt_opts, %fc_opts } ],
 144        branch => [ \&cmd_branch,
 145                    'Create a branch in the SVN repository',
 146                    { 'message|m=s' => \$_message,
 147                      'destination|d=s' => \$_branch_dest,
 148                      'dry-run|n' => \$_dry_run,
 149                      'tag|t' => \$_tag } ],
 150        tag => [ sub { $_tag = 1; cmd_branch(@_) },
 151                 'Create a tag in the SVN repository',
 152                 { 'message|m=s' => \$_message,
 153                   'destination|d=s' => \$_branch_dest,
 154                   'dry-run|n' => \$_dry_run } ],
 155        'set-tree' => [ \&cmd_set_tree,
 156                        "Set an SVN repository to a git tree-ish",
 157                        { 'stdin' => \$_stdin, %cmt_opts, %fc_opts, } ],
 158        'create-ignore' => [ \&cmd_create_ignore,
 159                             'Create a .gitignore per svn:ignore',
 160                             { 'revision|r=i' => \$_revision
 161                             } ],
 162        'propget' => [ \&cmd_propget,
 163                       'Print the value of a property on a file or directory',
 164                       { 'revision|r=i' => \$_revision } ],
 165        'proplist' => [ \&cmd_proplist,
 166                       'List all properties of a file or directory',
 167                       { 'revision|r=i' => \$_revision } ],
 168        'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
 169                        { 'revision|r=i' => \$_revision
 170                        } ],
 171        'show-externals' => [ \&cmd_show_externals, "Show svn:externals listings",
 172                        { 'revision|r=i' => \$_revision
 173                        } ],
 174        'multi-fetch' => [ \&cmd_multi_fetch,
 175                           "Deprecated alias for $0 fetch --all",
 176                           { 'revision|r=s' => \$_revision, %fc_opts } ],
 177        'migrate' => [ sub { },
 178                       # no-op, we automatically run this anyways,
 179                       'Migrate configuration/metadata/layout from
 180                        previous versions of git-svn',
 181                       { 'minimize' => \$Git::SVN::Migration::_minimize,
 182                         %remote_opts } ],
 183        'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
 184                        { 'limit=i' => \$Git::SVN::Log::limit,
 185                          'revision|r=s' => \$_revision,
 186                          'verbose|v' => \$Git::SVN::Log::verbose,
 187                          'incremental' => \$Git::SVN::Log::incremental,
 188                          'oneline' => \$Git::SVN::Log::oneline,
 189                          'show-commit' => \$Git::SVN::Log::show_commit,
 190                          'non-recursive' => \$Git::SVN::Log::non_recursive,
 191                          'authors-file|A=s' => \$_authors,
 192                          'color' => \$Git::SVN::Log::color,
 193                          'pager=s' => \$Git::SVN::Log::pager
 194                        } ],
 195        'find-rev' => [ \&cmd_find_rev,
 196                        "Translate between SVN revision numbers and tree-ish",
 197                        {} ],
 198        'rebase' => [ \&cmd_rebase, "Fetch and rebase your working directory",
 199                        { 'merge|m|M' => \$_merge,
 200                          'verbose|v' => \$_verbose,
 201                          'strategy|s=s' => \$_strategy,
 202                          'local|l' => \$_local,
 203                          'fetch-all|all' => \$_fetch_all,
 204                          'dry-run|n' => \$_dry_run,
 205                          %fc_opts } ],
 206        'commit-diff' => [ \&cmd_commit_diff,
 207                           'Commit a diff between two trees',
 208                        { 'message|m=s' => \$_message,
 209                          'file|F=s' => \$_file,
 210                          'revision|r=s' => \$_revision,
 211                        %cmt_opts } ],
 212        'info' => [ \&cmd_info,
 213                    "Show info about the latest SVN revision
 214                     on the current branch",
 215                    { 'url' => \$_url, } ],
 216        'blame' => [ \&Git::SVN::Log::cmd_blame,
 217                    "Show what revision and author last modified each line of a file",
 218                    { 'git-format' => \$_git_format } ],
 219        'reset' => [ \&cmd_reset,
 220                     "Undo fetches back to the specified SVN revision",
 221                     { 'revision|r=s' => \$_revision,
 222                       'parent|p' => \$_fetch_parent } ],
 223        'gc' => [ \&cmd_gc,
 224                  "Compress unhandled.log files in .git/svn and remove " .
 225                  "index files in .git/svn",
 226                {} ],
 227);
 228
 229my $cmd;
 230for (my $i = 0; $i < @ARGV; $i++) {
 231        if (defined $cmd{$ARGV[$i]}) {
 232                $cmd = $ARGV[$i];
 233                splice @ARGV, $i, 1;
 234                last;
 235        } elsif ($ARGV[$i] eq 'help') {
 236                $cmd = $ARGV[$i+1];
 237                usage(0);
 238        }
 239};
 240
 241# make sure we're always running at the top-level working directory
 242unless ($cmd && $cmd =~ /(?:clone|init|multi-init)$/) {
 243        unless (-d $ENV{GIT_DIR}) {
 244                if ($git_dir_user_set) {
 245                        die "GIT_DIR=$ENV{GIT_DIR} explicitly set, ",
 246                            "but it is not a directory\n";
 247                }
 248                my $git_dir = delete $ENV{GIT_DIR};
 249                my $cdup = undef;
 250                git_cmd_try {
 251                        $cdup = command_oneline(qw/rev-parse --show-cdup/);
 252                        $git_dir = '.' unless ($cdup);
 253                        chomp $cdup if ($cdup);
 254                        $cdup = "." unless ($cdup && length $cdup);
 255                } "Already at toplevel, but $git_dir not found\n";
 256                chdir $cdup or die "Unable to chdir up to '$cdup'\n";
 257                unless (-d $git_dir) {
 258                        die "$git_dir still not found after going to ",
 259                            "'$cdup'\n";
 260                }
 261                $ENV{GIT_DIR} = $git_dir;
 262        }
 263        $_repository = Git->repository(Repository => $ENV{GIT_DIR});
 264}
 265
 266my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
 267
 268read_repo_config(\%opts);
 269if ($cmd && ($cmd eq 'log' || $cmd eq 'blame')) {
 270        Getopt::Long::Configure('pass_through');
 271}
 272my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
 273                    'minimize-connections' => \$Git::SVN::Migration::_minimize,
 274                    'id|i=s' => \$Git::SVN::default_ref_id,
 275                    'svn-remote|remote|R=s' => sub {
 276                       $Git::SVN::no_reuse_existing = 1;
 277                       $Git::SVN::default_repo_id = $_[1] });
 278exit 1 if (!$rv && $cmd && $cmd ne 'log');
 279
 280usage(0) if $_help;
 281version() if $_version;
 282usage(1) unless defined $cmd;
 283load_authors() if $_authors;
 284if (defined $_authors_prog) {
 285        $_authors_prog = "'" . File::Spec->rel2abs($_authors_prog) . "'";
 286}
 287
 288unless ($cmd =~ /^(?:clone|init|multi-init|commit-diff)$/) {
 289        Git::SVN::Migration::migration_check();
 290}
 291Git::SVN::init_vars();
 292eval {
 293        Git::SVN::verify_remotes_sanity();
 294        $cmd{$cmd}->[0]->(@ARGV);
 295};
 296fatal $@ if $@;
 297post_fetch_checkout();
 298exit 0;
 299
 300####################### primary functions ######################
 301sub usage {
 302        my $exit = shift || 0;
 303        my $fd = $exit ? \*STDERR : \*STDOUT;
 304        print $fd <<"";
 305git-svn - bidirectional operations between a single Subversion tree and git
 306Usage: git svn <command> [options] [arguments]\n
 307
 308        print $fd "Available commands:\n" unless $cmd;
 309
 310        foreach (sort keys %cmd) {
 311                next if $cmd && $cmd ne $_;
 312                next if /^multi-/; # don't show deprecated commands
 313                print $fd '  ',pack('A17',$_),$cmd{$_}->[1],"\n";
 314                foreach (sort keys %{$cmd{$_}->[2]}) {
 315                        # mixed-case options are for .git/config only
 316                        next if /[A-Z]/ && /^[a-z]+$/i;
 317                        # prints out arguments as they should be passed:
 318                        my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
 319                        print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
 320                                                        "--$_" : "-$_" }
 321                                                split /\|/,$_)," $x\n";
 322                }
 323        }
 324        print $fd <<"";
 325\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
 326arbitrary identifier if you're tracking multiple SVN branches/repositories in
 327one git repository and want to keep them separate.  See git-svn(1) for more
 328information.
 329
 330        exit $exit;
 331}
 332
 333sub version {
 334        print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
 335        exit 0;
 336}
 337
 338sub do_git_init_db {
 339        unless (-d $ENV{GIT_DIR}) {
 340                my @init_db = ('init');
 341                push @init_db, "--template=$_template" if defined $_template;
 342                if (defined $_shared) {
 343                        if ($_shared =~ /[a-z]/) {
 344                                push @init_db, "--shared=$_shared";
 345                        } else {
 346                                push @init_db, "--shared";
 347                        }
 348                }
 349                command_noisy(@init_db);
 350                $_repository = Git->repository(Repository => ".git");
 351        }
 352        command_noisy('config', 'core.autocrlf', 'false');
 353        my $set;
 354        my $pfx = "svn-remote.$Git::SVN::default_repo_id";
 355        foreach my $i (keys %icv) {
 356                die "'$set' and '$i' cannot both be set\n" if $set;
 357                next unless defined $icv{$i};
 358                command_noisy('config', "$pfx.$i", $icv{$i});
 359                $set = $i;
 360        }
 361        my $ignore_regex = \$SVN::Git::Fetcher::_ignore_regex;
 362        command_noisy('config', "$pfx.ignore-paths", $$ignore_regex)
 363                if defined $$ignore_regex;
 364}
 365
 366sub init_subdir {
 367        my $repo_path = shift or return;
 368        mkpath([$repo_path]) unless -d $repo_path;
 369        chdir $repo_path or die "Couldn't chdir to $repo_path: $!\n";
 370        $ENV{GIT_DIR} = '.git';
 371        $_repository = Git->repository(Repository => $ENV{GIT_DIR});
 372}
 373
 374sub cmd_clone {
 375        my ($url, $path) = @_;
 376        if (!defined $path &&
 377            (defined $_trunk || @_branches || @_tags ||
 378             defined $_stdlayout) &&
 379            $url !~ m#^[a-z\+]+://#) {
 380                $path = $url;
 381        }
 382        $path = basename($url) if !defined $path || !length $path;
 383        cmd_init($url, $path);
 384        Git::SVN::fetch_all($Git::SVN::default_repo_id);
 385        command_oneline('config', 'svn.authorsfile', $_authors) if $_authors;
 386}
 387
 388sub cmd_init {
 389        if (defined $_stdlayout) {
 390                $_trunk = 'trunk' if (!defined $_trunk);
 391                @_tags = 'tags' if (! @_tags);
 392                @_branches = 'branches' if (! @_branches);
 393        }
 394        if (defined $_trunk || @_branches || @_tags) {
 395                return cmd_multi_init(@_);
 396        }
 397        my $url = shift or die "SVN repository location required ",
 398                               "as a command-line argument\n";
 399        $url = canonicalize_url($url);
 400        init_subdir(@_);
 401        do_git_init_db();
 402
 403        if ($Git::SVN::_minimize_url eq 'unset') {
 404                $Git::SVN::_minimize_url = 0;
 405        }
 406
 407        Git::SVN->init($url);
 408}
 409
 410sub cmd_fetch {
 411        if (grep /^\d+=./, @_) {
 412                die "'<rev>=<commit>' fetch arguments are ",
 413                    "no longer supported.\n";
 414        }
 415        my ($remote) = @_;
 416        if (@_ > 1) {
 417                die "Usage: $0 fetch [--all] [--parent] [svn-remote]\n";
 418        }
 419        if ($_fetch_parent) {
 420                my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 421                unless ($gs) {
 422                        die "Unable to determine upstream SVN information from ",
 423                            "working tree history\n";
 424                }
 425                # just fetch, don't checkout.
 426                $_no_checkout = 'true';
 427                $_fetch_all ? $gs->fetch_all : $gs->fetch;
 428        } elsif ($_fetch_all) {
 429                cmd_multi_fetch();
 430        } else {
 431                $remote ||= $Git::SVN::default_repo_id;
 432                Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
 433        }
 434}
 435
 436sub cmd_set_tree {
 437        my (@commits) = @_;
 438        if ($_stdin || !@commits) {
 439                print "Reading from stdin...\n";
 440                @commits = ();
 441                while (<STDIN>) {
 442                        if (/\b($sha1_short)\b/o) {
 443                                unshift @commits, $1;
 444                        }
 445                }
 446        }
 447        my @revs;
 448        foreach my $c (@commits) {
 449                my @tmp = command('rev-parse',$c);
 450                if (scalar @tmp == 1) {
 451                        push @revs, $tmp[0];
 452                } elsif (scalar @tmp > 1) {
 453                        push @revs, reverse(command('rev-list',@tmp));
 454                } else {
 455                        fatal "Failed to rev-parse $c";
 456                }
 457        }
 458        my $gs = Git::SVN->new;
 459        my ($r_last, $cmt_last) = $gs->last_rev_commit;
 460        $gs->fetch;
 461        if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
 462                fatal "There are new revisions that were fetched ",
 463                      "and need to be merged (or acknowledged) ",
 464                      "before committing.\nlast rev: $r_last\n",
 465                      " current: $gs->{last_rev}";
 466        }
 467        $gs->set_tree($_) foreach @revs;
 468        print "Done committing ",scalar @revs," revisions to SVN\n";
 469        unlink $gs->{index};
 470}
 471
 472sub cmd_dcommit {
 473        my $head = shift;
 474        git_cmd_try { command_oneline(qw/diff-index --quiet HEAD/) }
 475                'Cannot dcommit with a dirty index.  Commit your changes first, '
 476                . "or stash them with `git stash'.\n";
 477        $head ||= 'HEAD';
 478
 479        my $old_head;
 480        if ($head ne 'HEAD') {
 481                $old_head = eval {
 482                        command_oneline([qw/symbolic-ref -q HEAD/])
 483                };
 484                if ($old_head) {
 485                        $old_head =~ s{^refs/heads/}{};
 486                } else {
 487                        $old_head = eval { command_oneline(qw/rev-parse HEAD/) };
 488                }
 489                command(['checkout', $head], STDERR => 0);
 490        }
 491
 492        my @refs;
 493        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD', \@refs);
 494        unless ($gs) {
 495                die "Unable to determine upstream SVN information from ",
 496                    "$head history.\nPerhaps the repository is empty.";
 497        }
 498
 499        if (defined $_commit_url) {
 500                $url = $_commit_url;
 501        } else {
 502                $url = eval { command_oneline('config', '--get',
 503                              "svn-remote.$gs->{repo_id}.commiturl") };
 504                if (!$url) {
 505                        $url = $gs->full_url
 506                }
 507        }
 508
 509        my $last_rev = $_revision if defined $_revision;
 510        if ($url) {
 511                print "Committing to $url ...\n";
 512        }
 513        my ($linear_refs, $parents) = linearize_history($gs, \@refs);
 514        if ($_no_rebase && scalar(@$linear_refs) > 1) {
 515                warn "Attempting to commit more than one change while ",
 516                     "--no-rebase is enabled.\n",
 517                     "If these changes depend on each other, re-running ",
 518                     "without --no-rebase may be required."
 519        }
 520        my $expect_url = $url;
 521        Git::SVN::remove_username($expect_url);
 522        while (1) {
 523                my $d = shift @$linear_refs or last;
 524                unless (defined $last_rev) {
 525                        (undef, $last_rev, undef) = cmt_metadata("$d~1");
 526                        unless (defined $last_rev) {
 527                                fatal "Unable to extract revision information ",
 528                                      "from commit $d~1";
 529                        }
 530                }
 531                if ($_dry_run) {
 532                        print "diff-tree $d~1 $d\n";
 533                } else {
 534                        my $cmt_rev;
 535                        my %ed_opts = ( r => $last_rev,
 536                                        log => get_commit_entry($d)->{log},
 537                                        ra => Git::SVN::Ra->new($url),
 538                                        config => SVN::Core::config_get_config(
 539                                                $Git::SVN::Ra::config_dir
 540                                        ),
 541                                        tree_a => "$d~1",
 542                                        tree_b => $d,
 543                                        editor_cb => sub {
 544                                               print "Committed r$_[0]\n";
 545                                               $cmt_rev = $_[0];
 546                                        },
 547                                        svn_path => '');
 548                        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
 549                                print "No changes\n$d~1 == $d\n";
 550                        } elsif ($parents->{$d} && @{$parents->{$d}}) {
 551                                $gs->{inject_parents_dcommit}->{$cmt_rev} =
 552                                                               $parents->{$d};
 553                        }
 554                        $_fetch_all ? $gs->fetch_all : $gs->fetch;
 555                        $last_rev = $cmt_rev;
 556                        next if $_no_rebase;
 557
 558                        # we always want to rebase against the current HEAD,
 559                        # not any head that was passed to us
 560                        my @diff = command('diff-tree', $d,
 561                                           $gs->refname, '--');
 562                        my @finish;
 563                        if (@diff) {
 564                                @finish = rebase_cmd();
 565                                print STDERR "W: $d and ", $gs->refname,
 566                                             " differ, using @finish:\n",
 567                                             join("\n", @diff), "\n";
 568                        } else {
 569                                print "No changes between current HEAD and ",
 570                                      $gs->refname,
 571                                      "\nResetting to the latest ",
 572                                      $gs->refname, "\n";
 573                                @finish = qw/reset --mixed/;
 574                        }
 575                        command_noisy(@finish, $gs->refname);
 576                        if (@diff) {
 577                                @refs = ();
 578                                my ($url_, $rev_, $uuid_, $gs_) =
 579                                              working_head_info('HEAD', \@refs);
 580                                my ($linear_refs_, $parents_) =
 581                                              linearize_history($gs_, \@refs);
 582                                if (scalar(@$linear_refs) !=
 583                                    scalar(@$linear_refs_)) {
 584                                        fatal "# of revisions changed ",
 585                                          "\nbefore:\n",
 586                                          join("\n", @$linear_refs),
 587                                          "\n\nafter:\n",
 588                                          join("\n", @$linear_refs_), "\n",
 589                                          'If you are attempting to commit ',
 590                                          "merges, try running:\n\t",
 591                                          'git rebase --interactive',
 592                                          '--preserve-merges ',
 593                                          $gs->refname,
 594                                          "\nBefore dcommitting";
 595                                }
 596                                if ($url_ ne $expect_url) {
 597                                        fatal "URL mismatch after rebase: ",
 598                                              "$url_ != $expect_url";
 599                                }
 600                                if ($uuid_ ne $uuid) {
 601                                        fatal "uuid mismatch after rebase: ",
 602                                              "$uuid_ != $uuid";
 603                                }
 604                                # remap parents
 605                                my (%p, @l, $i);
 606                                for ($i = 0; $i < scalar @$linear_refs; $i++) {
 607                                        my $new = $linear_refs_->[$i] or next;
 608                                        $p{$new} =
 609                                                $parents->{$linear_refs->[$i]};
 610                                        push @l, $new;
 611                                }
 612                                $parents = \%p;
 613                                $linear_refs = \@l;
 614                        }
 615                }
 616        }
 617
 618        if ($old_head) {
 619                my $new_head = command_oneline(qw/rev-parse HEAD/);
 620                my $new_is_symbolic = eval {
 621                        command_oneline(qw/symbolic-ref -q HEAD/);
 622                };
 623                if ($new_is_symbolic) {
 624                        print "dcommitted the branch ", $head, "\n";
 625                } else {
 626                        print "dcommitted on a detached HEAD because you gave ",
 627                              "a revision argument.\n",
 628                              "The rewritten commit is: ", $new_head, "\n";
 629                }
 630                command(['checkout', $old_head], STDERR => 0);
 631        }
 632
 633        unlink $gs->{index};
 634}
 635
 636sub cmd_branch {
 637        my ($branch_name, $head) = @_;
 638
 639        unless (defined $branch_name && length $branch_name) {
 640                die(($_tag ? "tag" : "branch") . " name required\n");
 641        }
 642        $head ||= 'HEAD';
 643
 644        my ($src, $rev, undef, $gs) = working_head_info($head);
 645
 646        my $remote = Git::SVN::read_all_remotes()->{$gs->{repo_id}};
 647        my $allglobs = $remote->{ $_tag ? 'tags' : 'branches' };
 648        my $glob;
 649        if ($#{$allglobs} == 0) {
 650                $glob = $allglobs->[0];
 651        } else {
 652                unless(defined $_branch_dest) {
 653                        die "Multiple ",
 654                            $_tag ? "tag" : "branch",
 655                            " paths defined for Subversion repository.\n",
 656                            "You must specify where you want to create the ",
 657                            $_tag ? "tag" : "branch",
 658                            " with the --destination argument.\n";
 659                }
 660                foreach my $g (@{$allglobs}) {
 661                        # SVN::Git::Editor could probably be moved to Git.pm..
 662                        my $re = SVN::Git::Editor::glob2pat($g->{path}->{left});
 663                        if ($_branch_dest =~ /$re/) {
 664                                $glob = $g;
 665                                last;
 666                        }
 667                }
 668                unless (defined $glob) {
 669                        my $dest_re = qr/\b\Q$_branch_dest\E\b/;
 670                        foreach my $g (@{$allglobs}) {
 671                                $g->{path}->{left} =~ /$dest_re/ or next;
 672                                if (defined $glob) {
 673                                        die "Ambiguous destination: ",
 674                                            $_branch_dest, "\nmatches both '",
 675                                            $glob->{path}->{left}, "' and '",
 676                                            $g->{path}->{left}, "'\n";
 677                                }
 678                                $glob = $g;
 679                        }
 680                        unless (defined $glob) {
 681                                die "Unknown ",
 682                                    $_tag ? "tag" : "branch",
 683                                    " destination $_branch_dest\n";
 684                        }
 685                }
 686        }
 687        my ($lft, $rgt) = @{ $glob->{path} }{qw/left right/};
 688        my $dst = join '/', $remote->{url}, $lft, $branch_name, ($rgt || ());
 689
 690        my $ctx = SVN::Client->new(
 691                auth    => Git::SVN::Ra::_auth_providers(),
 692                log_msg => sub {
 693                        ${ $_[0] } = defined $_message
 694                                ? $_message
 695                                : 'Create ' . ($_tag ? 'tag ' : 'branch ' )
 696                                . $branch_name;
 697                },
 698        );
 699
 700        eval {
 701                $ctx->ls($dst, 'HEAD', 0);
 702        } and die "branch ${branch_name} already exists\n";
 703
 704        print "Copying ${src} at r${rev} to ${dst}...\n";
 705        $ctx->copy($src, $rev, $dst)
 706                unless $_dry_run;
 707
 708        $gs->fetch_all;
 709}
 710
 711sub cmd_find_rev {
 712        my $revision_or_hash = shift or die "SVN or git revision required ",
 713                                            "as a command-line argument\n";
 714        my $result;
 715        if ($revision_or_hash =~ /^r\d+$/) {
 716                my $head = shift;
 717                $head ||= 'HEAD';
 718                my @refs;
 719                my (undef, undef, $uuid, $gs) = working_head_info($head, \@refs);
 720                unless ($gs) {
 721                        die "Unable to determine upstream SVN information from ",
 722                            "$head history\n";
 723                }
 724                my $desired_revision = substr($revision_or_hash, 1);
 725                $result = $gs->rev_map_get($desired_revision, $uuid);
 726        } else {
 727                my (undef, $rev, undef) = cmt_metadata($revision_or_hash);
 728                $result = $rev;
 729        }
 730        print "$result\n" if $result;
 731}
 732
 733sub cmd_rebase {
 734        command_noisy(qw/update-index --refresh/);
 735        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 736        unless ($gs) {
 737                die "Unable to determine upstream SVN information from ",
 738                    "working tree history\n";
 739        }
 740        if ($_dry_run) {
 741                print "Remote Branch: " . $gs->refname . "\n";
 742                print "SVN URL: " . $url . "\n";
 743                return;
 744        }
 745        if (command(qw/diff-index HEAD --/)) {
 746                print STDERR "Cannot rebase with uncommited changes:\n";
 747                command_noisy('status');
 748                exit 1;
 749        }
 750        unless ($_local) {
 751                # rebase will checkout for us, so no need to do it explicitly
 752                $_no_checkout = 'true';
 753                $_fetch_all ? $gs->fetch_all : $gs->fetch;
 754        }
 755        command_noisy(rebase_cmd(), $gs->refname);
 756}
 757
 758sub cmd_show_ignore {
 759        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 760        $gs ||= Git::SVN->new;
 761        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 762        $gs->prop_walk($gs->{path}, $r, sub {
 763                my ($gs, $path, $props) = @_;
 764                print STDOUT "\n# $path\n";
 765                my $s = $props->{'svn:ignore'} or return;
 766                $s =~ s/[\r\n]+/\n/g;
 767                chomp $s;
 768                $s =~ s#^#$path#gm;
 769                print STDOUT "$s\n";
 770        });
 771}
 772
 773sub cmd_show_externals {
 774        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 775        $gs ||= Git::SVN->new;
 776        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 777        $gs->prop_walk($gs->{path}, $r, sub {
 778                my ($gs, $path, $props) = @_;
 779                print STDOUT "\n# $path\n";
 780                my $s = $props->{'svn:externals'} or return;
 781                $s =~ s/[\r\n]+/\n/g;
 782                chomp $s;
 783                $s =~ s#^#$path#gm;
 784                print STDOUT "$s\n";
 785        });
 786}
 787
 788sub cmd_create_ignore {
 789        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 790        $gs ||= Git::SVN->new;
 791        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 792        $gs->prop_walk($gs->{path}, $r, sub {
 793                my ($gs, $path, $props) = @_;
 794                # $path is of the form /path/to/dir/
 795                $path = '.' . $path;
 796                # SVN can have attributes on empty directories,
 797                # which git won't track
 798                mkpath([$path]) unless -d $path;
 799                my $ignore = $path . '.gitignore';
 800                my $s = $props->{'svn:ignore'} or return;
 801                open(GITIGNORE, '>', $ignore)
 802                  or fatal("Failed to open `$ignore' for writing: $!");
 803                $s =~ s/[\r\n]+/\n/g;
 804                chomp $s;
 805                # Prefix all patterns so that the ignore doesn't apply
 806                # to sub-directories.
 807                $s =~ s#^#/#gm;
 808                print GITIGNORE "$s\n";
 809                close(GITIGNORE)
 810                  or fatal("Failed to close `$ignore': $!");
 811                command_noisy('add', '-f', $ignore);
 812        });
 813}
 814
 815sub canonicalize_path {
 816        my ($path) = @_;
 817        my $dot_slash_added = 0;
 818        if (substr($path, 0, 1) ne "/") {
 819                $path = "./" . $path;
 820                $dot_slash_added = 1;
 821        }
 822        # File::Spec->canonpath doesn't collapse x/../y into y (for a
 823        # good reason), so let's do this manually.
 824        $path =~ s#/+#/#g;
 825        $path =~ s#/\.(?:/|$)#/#g;
 826        $path =~ s#/[^/]+/\.\.##g;
 827        $path =~ s#/$##g;
 828        $path =~ s#^\./## if $dot_slash_added;
 829        $path =~ s#^/##;
 830        $path =~ s#^\.$##;
 831        return $path;
 832}
 833
 834sub canonicalize_url {
 835        my ($url) = @_;
 836        $url =~ s#^([^:]+://[^/]*/)(.*)$#$1 . canonicalize_path($2)#e;
 837        return $url;
 838}
 839
 840# get_svnprops(PATH)
 841# ------------------
 842# Helper for cmd_propget and cmd_proplist below.
 843sub get_svnprops {
 844        my $path = shift;
 845        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 846        $gs ||= Git::SVN->new;
 847
 848        # prefix THE PATH by the sub-directory from which the user
 849        # invoked us.
 850        $path = $cmd_dir_prefix . $path;
 851        fatal("No such file or directory: $path") unless -e $path;
 852        my $is_dir = -d $path ? 1 : 0;
 853        $path = $gs->{path} . '/' . $path;
 854
 855        # canonicalize the path (otherwise libsvn will abort or fail to
 856        # find the file)
 857        $path = canonicalize_path($path);
 858
 859        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 860        my $props;
 861        if ($is_dir) {
 862                (undef, undef, $props) = $gs->ra->get_dir($path, $r);
 863        }
 864        else {
 865                (undef, $props) = $gs->ra->get_file($path, $r, undef);
 866        }
 867        return $props;
 868}
 869
 870# cmd_propget (PROP, PATH)
 871# ------------------------
 872# Print the SVN property PROP for PATH.
 873sub cmd_propget {
 874        my ($prop, $path) = @_;
 875        $path = '.' if not defined $path;
 876        usage(1) if not defined $prop;
 877        my $props = get_svnprops($path);
 878        if (not defined $props->{$prop}) {
 879                fatal("`$path' does not have a `$prop' SVN property.");
 880        }
 881        print $props->{$prop} . "\n";
 882}
 883
 884# cmd_proplist (PATH)
 885# -------------------
 886# Print the list of SVN properties for PATH.
 887sub cmd_proplist {
 888        my $path = shift;
 889        $path = '.' if not defined $path;
 890        my $props = get_svnprops($path);
 891        print "Properties on '$path':\n";
 892        foreach (sort keys %{$props}) {
 893                print "  $_\n";
 894        }
 895}
 896
 897sub cmd_multi_init {
 898        my $url = shift;
 899        unless (defined $_trunk || @_branches || @_tags) {
 900                usage(1);
 901        }
 902
 903        $_prefix = '' unless defined $_prefix;
 904        if (defined $url) {
 905                $url = canonicalize_url($url);
 906                init_subdir(@_);
 907        }
 908        do_git_init_db();
 909        if (defined $_trunk) {
 910                my $trunk_ref = $_prefix . 'trunk';
 911                # try both old-style and new-style lookups:
 912                my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
 913                unless ($gs_trunk) {
 914                        my ($trunk_url, $trunk_path) =
 915                                              complete_svn_url($url, $_trunk);
 916                        $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
 917                                                   undef, $trunk_ref);
 918                }
 919        }
 920        return unless @_branches || @_tags;
 921        my $ra = $url ? Git::SVN::Ra->new($url) : undef;
 922        foreach my $path (@_branches) {
 923                complete_url_ls_init($ra, $path, '--branches/-b', $_prefix);
 924        }
 925        foreach my $path (@_tags) {
 926                complete_url_ls_init($ra, $path, '--tags/-t', $_prefix.'tags/');
 927        }
 928}
 929
 930sub cmd_multi_fetch {
 931        my $remotes = Git::SVN::read_all_remotes();
 932        foreach my $repo_id (sort keys %$remotes) {
 933                if ($remotes->{$repo_id}->{url}) {
 934                        Git::SVN::fetch_all($repo_id, $remotes);
 935                }
 936        }
 937}
 938
 939# this command is special because it requires no metadata
 940sub cmd_commit_diff {
 941        my ($ta, $tb, $url) = @_;
 942        my $usage = "Usage: $0 commit-diff -r<revision> ".
 943                    "<tree-ish> <tree-ish> [<URL>]";
 944        fatal($usage) if (!defined $ta || !defined $tb);
 945        my $svn_path = '';
 946        if (!defined $url) {
 947                my $gs = eval { Git::SVN->new };
 948                if (!$gs) {
 949                        fatal("Needed URL or usable git-svn --id in ",
 950                              "the command-line\n", $usage);
 951                }
 952                $url = $gs->{url};
 953                $svn_path = $gs->{path};
 954        }
 955        unless (defined $_revision) {
 956                fatal("-r|--revision is a required argument\n", $usage);
 957        }
 958        if (defined $_message && defined $_file) {
 959                fatal("Both --message/-m and --file/-F specified ",
 960                      "for the commit message.\n",
 961                      "I have no idea what you mean");
 962        }
 963        if (defined $_file) {
 964                $_message = file_to_s($_file);
 965        } else {
 966                $_message ||= get_commit_entry($tb)->{log};
 967        }
 968        my $ra ||= Git::SVN::Ra->new($url);
 969        my $r = $_revision;
 970        if ($r eq 'HEAD') {
 971                $r = $ra->get_latest_revnum;
 972        } elsif ($r !~ /^\d+$/) {
 973                die "revision argument: $r not understood by git-svn\n";
 974        }
 975        my %ed_opts = ( r => $r,
 976                        log => $_message,
 977                        ra => $ra,
 978                        tree_a => $ta,
 979                        tree_b => $tb,
 980                        editor_cb => sub { print "Committed r$_[0]\n" },
 981                        svn_path => $svn_path );
 982        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
 983                print "No changes\n$ta == $tb\n";
 984        }
 985}
 986
 987sub escape_uri_only {
 988        my ($uri) = @_;
 989        my @tmp;
 990        foreach (split m{/}, $uri) {
 991                s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
 992                push @tmp, $_;
 993        }
 994        join('/', @tmp);
 995}
 996
 997sub escape_url {
 998        my ($url) = @_;
 999        if ($url =~ m#^([^:]+)://([^/]*)(.*)$#) {
1000                my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
1001                $url = "$scheme://$domain$uri";
1002        }
1003        $url;
1004}
1005
1006sub cmd_info {
1007        my $path = canonicalize_path(defined($_[0]) ? $_[0] : ".");
1008        my $fullpath = canonicalize_path($cmd_dir_prefix . $path);
1009        if (exists $_[1]) {
1010                die "Too many arguments specified\n";
1011        }
1012
1013        my ($file_type, $diff_status) = find_file_type_and_diff_status($path);
1014
1015        if (!$file_type && !$diff_status) {
1016                print STDERR "svn: '$path' is not under version control\n";
1017                exit 1;
1018        }
1019
1020        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1021        unless ($gs) {
1022                die "Unable to determine upstream SVN information from ",
1023                    "working tree history\n";
1024        }
1025
1026        # canonicalize_path() will return "" to make libsvn 1.5.x happy,
1027        $path = "." if $path eq "";
1028
1029        my $full_url = $url . ($fullpath eq "" ? "" : "/$fullpath");
1030
1031        if ($_url) {
1032                print escape_url($full_url), "\n";
1033                return;
1034        }
1035
1036        my $result = "Path: $path\n";
1037        $result .= "Name: " . basename($path) . "\n" if $file_type ne "dir";
1038        $result .= "URL: " . escape_url($full_url) . "\n";
1039
1040        eval {
1041                my $repos_root = $gs->repos_root;
1042                Git::SVN::remove_username($repos_root);
1043                $result .= "Repository Root: " . escape_url($repos_root) . "\n";
1044        };
1045        if ($@) {
1046                $result .= "Repository Root: (offline)\n";
1047        }
1048        $result .= "Repository UUID: $uuid\n" unless $diff_status eq "A" &&
1049                ($SVN::Core::VERSION le '1.5.4' || $file_type ne "dir");
1050        $result .= "Revision: " . ($diff_status eq "A" ? 0 : $rev) . "\n";
1051
1052        $result .= "Node Kind: " .
1053                   ($file_type eq "dir" ? "directory" : "file") . "\n";
1054
1055        my $schedule = $diff_status eq "A"
1056                       ? "add"
1057                       : ($diff_status eq "D" ? "delete" : "normal");
1058        $result .= "Schedule: $schedule\n";
1059
1060        if ($diff_status eq "A") {
1061                print $result, "\n";
1062                return;
1063        }
1064
1065        my ($lc_author, $lc_rev, $lc_date_utc);
1066        my @args = Git::SVN::Log::git_svn_log_cmd($rev, $rev, "--", $fullpath);
1067        my $log = command_output_pipe(@args);
1068        my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
1069        while (<$log>) {
1070                if (/^${esc_color}author (.+) <[^>]+> (\d+) ([\-\+]?\d+)$/o) {
1071                        $lc_author = $1;
1072                        $lc_date_utc = Git::SVN::Log::parse_git_date($2, $3);
1073                } elsif (/^${esc_color}    (git-svn-id:.+)$/o) {
1074                        (undef, $lc_rev, undef) = ::extract_metadata($1);
1075                }
1076        }
1077        close $log;
1078
1079        Git::SVN::Log::set_local_timezone();
1080
1081        $result .= "Last Changed Author: $lc_author\n";
1082        $result .= "Last Changed Rev: $lc_rev\n";
1083        $result .= "Last Changed Date: " .
1084                   Git::SVN::Log::format_svn_date($lc_date_utc) . "\n";
1085
1086        if ($file_type ne "dir") {
1087                my $text_last_updated_date =
1088                    ($diff_status eq "D" ? $lc_date_utc : (stat $path)[9]);
1089                $result .=
1090                    "Text Last Updated: " .
1091                    Git::SVN::Log::format_svn_date($text_last_updated_date) .
1092                    "\n";
1093                my $checksum;
1094                if ($diff_status eq "D") {
1095                        my ($fh, $ctx) =
1096                            command_output_pipe(qw(cat-file blob), "HEAD:$path");
1097                        if ($file_type eq "link") {
1098                                my $file_name = <$fh>;
1099                                $checksum = md5sum("link $file_name");
1100                        } else {
1101                                $checksum = md5sum($fh);
1102                        }
1103                        command_close_pipe($fh, $ctx);
1104                } elsif ($file_type eq "link") {
1105                        my $file_name =
1106                            command(qw(cat-file blob), "HEAD:$path");
1107                        $checksum =
1108                            md5sum("link " . $file_name);
1109                } else {
1110                        open FILE, "<", $path or die $!;
1111                        $checksum = md5sum(\*FILE);
1112                        close FILE or die $!;
1113                }
1114                $result .= "Checksum: " . $checksum . "\n";
1115        }
1116
1117        print $result, "\n";
1118}
1119
1120sub cmd_reset {
1121        my $target = shift || $_revision or die "SVN revision required\n";
1122        $target = $1 if $target =~ /^r(\d+)$/;
1123        $target =~ /^\d+$/ or die "Numeric SVN revision expected\n";
1124        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1125        unless ($gs) {
1126                die "Unable to determine upstream SVN information from ".
1127                    "history\n";
1128        }
1129        my ($r, $c) = $gs->find_rev_before($target, not $_fetch_parent);
1130        $gs->rev_map_set($r, $c, 'reset', $uuid);
1131        print "r$r = $c ($gs->{ref_id})\n";
1132}
1133
1134sub cmd_gc {
1135        if (!$can_compress) {
1136                warn "Compress::Zlib could not be found; unhandled.log " .
1137                     "files will not be compressed.\n";
1138        }
1139        find({ wanted => \&gc_directory, no_chdir => 1}, "$ENV{GIT_DIR}/svn");
1140}
1141
1142########################### utility functions #########################
1143
1144sub rebase_cmd {
1145        my @cmd = qw/rebase/;
1146        push @cmd, '-v' if $_verbose;
1147        push @cmd, qw/--merge/ if $_merge;
1148        push @cmd, "--strategy=$_strategy" if $_strategy;
1149        @cmd;
1150}
1151
1152sub post_fetch_checkout {
1153        return if $_no_checkout;
1154        my $gs = $Git::SVN::_head or return;
1155        return if verify_ref('refs/heads/master^0');
1156
1157        my $valid_head = verify_ref('HEAD^0');
1158        command_noisy(qw(update-ref refs/heads/master), $gs->refname);
1159        return if ($valid_head || !verify_ref('HEAD^0'));
1160
1161        return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
1162        my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
1163        return if -f $index;
1164
1165        return if command_oneline(qw/rev-parse --is-inside-work-tree/) eq 'false';
1166        return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
1167        command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
1168        print STDERR "Checked out HEAD:\n  ",
1169                     $gs->full_url, " r", $gs->last_rev, "\n";
1170}
1171
1172sub complete_svn_url {
1173        my ($url, $path) = @_;
1174        $path =~ s#/+$##;
1175        if ($path !~ m#^[a-z\+]+://#) {
1176                if (!defined $url || $url !~ m#^[a-z\+]+://#) {
1177                        fatal("E: '$path' is not a complete URL ",
1178                              "and a separate URL is not specified");
1179                }
1180                return ($url, $path);
1181        }
1182        return ($path, '');
1183}
1184
1185sub complete_url_ls_init {
1186        my ($ra, $repo_path, $switch, $pfx) = @_;
1187        unless ($repo_path) {
1188                print STDERR "W: $switch not specified\n";
1189                return;
1190        }
1191        $repo_path =~ s#/+$##;
1192        if ($repo_path =~ m#^[a-z\+]+://#) {
1193                $ra = Git::SVN::Ra->new($repo_path);
1194                $repo_path = '';
1195        } else {
1196                $repo_path =~ s#^/+##;
1197                unless ($ra) {
1198                        fatal("E: '$repo_path' is not a complete URL ",
1199                              "and a separate URL is not specified");
1200                }
1201        }
1202        my $url = $ra->{url};
1203        my $gs = Git::SVN->init($url, undef, undef, undef, 1);
1204        my $k = "svn-remote.$gs->{repo_id}.url";
1205        my $orig_url = eval { command_oneline(qw/config --get/, $k) };
1206        if ($orig_url && ($orig_url ne $gs->{url})) {
1207                die "$k already set: $orig_url\n",
1208                    "wanted to set to: $gs->{url}\n";
1209        }
1210        command_oneline('config', $k, $gs->{url}) unless $orig_url;
1211        my $remote_path = "$gs->{path}/$repo_path";
1212        $remote_path =~ s#/+#/#g;
1213        $remote_path =~ s#^/##g;
1214        $remote_path .= "/*" if $remote_path !~ /\*/;
1215        my ($n) = ($switch =~ /^--(\w+)/);
1216        if (length $pfx && $pfx !~ m#/$#) {
1217                die "--prefix='$pfx' must have a trailing slash '/'\n";
1218        }
1219        command_noisy('config',
1220                      '--add',
1221                      "svn-remote.$gs->{repo_id}.$n",
1222                      "$remote_path:refs/remotes/$pfx*" .
1223                        ('/*' x (($remote_path =~ tr/*/*/) - 1)) );
1224}
1225
1226sub verify_ref {
1227        my ($ref) = @_;
1228        eval { command_oneline([ 'rev-parse', '--verify', $ref ],
1229                               { STDERR => 0 }); };
1230}
1231
1232sub get_tree_from_treeish {
1233        my ($treeish) = @_;
1234        # $treeish can be a symbolic ref, too:
1235        my $type = command_oneline(qw/cat-file -t/, $treeish);
1236        my $expected;
1237        while ($type eq 'tag') {
1238                ($treeish, $type) = command(qw/cat-file tag/, $treeish);
1239        }
1240        if ($type eq 'commit') {
1241                $expected = (grep /^tree /, command(qw/cat-file commit/,
1242                                                    $treeish))[0];
1243                ($expected) = ($expected =~ /^tree ($sha1)$/o);
1244                die "Unable to get tree from $treeish\n" unless $expected;
1245        } elsif ($type eq 'tree') {
1246                $expected = $treeish;
1247        } else {
1248                die "$treeish is a $type, expected tree, tag or commit\n";
1249        }
1250        return $expected;
1251}
1252
1253sub get_commit_entry {
1254        my ($treeish) = shift;
1255        my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
1256        my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
1257        my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
1258        open my $log_fh, '>', $commit_editmsg or croak $!;
1259
1260        my $type = command_oneline(qw/cat-file -t/, $treeish);
1261        if ($type eq 'commit' || $type eq 'tag') {
1262                my ($msg_fh, $ctx) = command_output_pipe('cat-file',
1263                                                         $type, $treeish);
1264                my $in_msg = 0;
1265                my $author;
1266                my $saw_from = 0;
1267                my $msgbuf = "";
1268                while (<$msg_fh>) {
1269                        if (!$in_msg) {
1270                                $in_msg = 1 if (/^\s*$/);
1271                                $author = $1 if (/^author (.*>)/);
1272                        } elsif (/^git-svn-id: /) {
1273                                # skip this for now, we regenerate the
1274                                # correct one on re-fetch anyways
1275                                # TODO: set *:merge properties or like...
1276                        } else {
1277                                if (/^From:/ || /^Signed-off-by:/) {
1278                                        $saw_from = 1;
1279                                }
1280                                $msgbuf .= $_;
1281                        }
1282                }
1283                $msgbuf =~ s/\s+$//s;
1284                if ($Git::SVN::_add_author_from && defined($author)
1285                    && !$saw_from) {
1286                        $msgbuf .= "\n\nFrom: $author";
1287                }
1288                print $log_fh $msgbuf or croak $!;
1289                command_close_pipe($msg_fh, $ctx);
1290        }
1291        close $log_fh or croak $!;
1292
1293        if ($_edit || ($type eq 'tree')) {
1294                my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
1295                # TODO: strip out spaces, comments, like git-commit.sh
1296                system($editor, $commit_editmsg);
1297        }
1298        rename $commit_editmsg, $commit_msg or croak $!;
1299        {
1300                require Encode;
1301                # SVN requires messages to be UTF-8 when entering the repo
1302                local $/;
1303                open $log_fh, '<', $commit_msg or croak $!;
1304                binmode $log_fh;
1305                chomp($log_entry{log} = <$log_fh>);
1306
1307                my $enc = Git::config('i18n.commitencoding') || 'UTF-8';
1308                my $msg = $log_entry{log};
1309
1310                eval { $msg = Encode::decode($enc, $msg, 1) };
1311                if ($@) {
1312                        die "Could not decode as $enc:\n", $msg,
1313                            "\nPerhaps you need to set i18n.commitencoding\n";
1314                }
1315
1316                eval { $msg = Encode::encode('UTF-8', $msg, 1) };
1317                die "Could not encode as UTF-8:\n$msg\n" if $@;
1318
1319                $log_entry{log} = $msg;
1320
1321                close $log_fh or croak $!;
1322        }
1323        unlink $commit_msg;
1324        \%log_entry;
1325}
1326
1327sub s_to_file {
1328        my ($str, $file, $mode) = @_;
1329        open my $fd,'>',$file or croak $!;
1330        print $fd $str,"\n" or croak $!;
1331        close $fd or croak $!;
1332        chmod ($mode &~ umask, $file) if (defined $mode);
1333}
1334
1335sub file_to_s {
1336        my $file = shift;
1337        open my $fd,'<',$file or croak "$!: file: $file\n";
1338        local $/;
1339        my $ret = <$fd>;
1340        close $fd or croak $!;
1341        $ret =~ s/\s*$//s;
1342        return $ret;
1343}
1344
1345# '<svn username> = real-name <email address>' mapping based on git-svnimport:
1346sub load_authors {
1347        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1348        my $log = $cmd eq 'log';
1349        while (<$authors>) {
1350                chomp;
1351                next unless /^(.+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
1352                my ($user, $name, $email) = ($1, $2, $3);
1353                if ($log) {
1354                        $Git::SVN::Log::rusers{"$name <$email>"} = $user;
1355                } else {
1356                        $users{$user} = [$name, $email];
1357                }
1358        }
1359        close $authors or croak $!;
1360}
1361
1362# convert GetOpt::Long specs for use by git-config
1363sub read_repo_config {
1364        return unless -d $ENV{GIT_DIR};
1365        my $opts = shift;
1366        my @config_only;
1367        foreach my $o (keys %$opts) {
1368                # if we have mixedCase and a long option-only, then
1369                # it's a config-only variable that we don't need for
1370                # the command-line.
1371                push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
1372                my $v = $opts->{$o};
1373                my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
1374                $key =~ s/-//g;
1375                my $arg = 'git config';
1376                $arg .= ' --int' if ($o =~ /[:=]i$/);
1377                $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
1378                if (ref $v eq 'ARRAY') {
1379                        chomp(my @tmp = `$arg --get-all svn.$key`);
1380                        @$v = @tmp if @tmp;
1381                } else {
1382                        chomp(my $tmp = `$arg --get svn.$key`);
1383                        if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
1384                                $$v = $tmp;
1385                        }
1386                }
1387        }
1388        delete @$opts{@config_only} if @config_only;
1389}
1390
1391sub extract_metadata {
1392        my $id = shift or return (undef, undef, undef);
1393        my ($url, $rev, $uuid) = ($id =~ /^\s*git-svn-id:\s+(.*)\@(\d+)
1394                                                        \s([a-f\d\-]+)$/ix);
1395        if (!defined $rev || !$uuid || !$url) {
1396                # some of the original repositories I made had
1397                # identifiers like this:
1398                ($rev, $uuid) = ($id =~/^\s*git-svn-id:\s(\d+)\@([a-f\d\-]+)/i);
1399        }
1400        return ($url, $rev, $uuid);
1401}
1402
1403sub cmt_metadata {
1404        return extract_metadata((grep(/^git-svn-id: /,
1405                command(qw/cat-file commit/, shift)))[-1]);
1406}
1407
1408sub cmt_sha2rev_batch {
1409        my %s2r;
1410        my ($pid, $in, $out, $ctx) = command_bidi_pipe(qw/cat-file --batch/);
1411        my $list = shift;
1412
1413        foreach my $sha (@{$list}) {
1414                my $first = 1;
1415                my $size = 0;
1416                print $out $sha, "\n";
1417
1418                while (my $line = <$in>) {
1419                        if ($first && $line =~ /^[[:xdigit:]]{40}\smissing$/) {
1420                                last;
1421                        } elsif ($first &&
1422                               $line =~ /^[[:xdigit:]]{40}\scommit\s(\d+)$/) {
1423                                $first = 0;
1424                                $size = $1;
1425                                next;
1426                        } elsif ($line =~ /^(git-svn-id: )/) {
1427                                my (undef, $rev, undef) =
1428                                                      extract_metadata($line);
1429                                $s2r{$sha} = $rev;
1430                        }
1431
1432                        $size -= length($line);
1433                        last if ($size == 0);
1434                }
1435        }
1436
1437        command_close_bidi_pipe($pid, $in, $out, $ctx);
1438
1439        return \%s2r;
1440}
1441
1442sub working_head_info {
1443        my ($head, $refs) = @_;
1444        my @args = ('log', '--no-color', '--first-parent', '--pretty=medium');
1445        my ($fh, $ctx) = command_output_pipe(@args, $head);
1446        my $hash;
1447        my %max;
1448        while (<$fh>) {
1449                if ( m{^commit ($::sha1)$} ) {
1450                        unshift @$refs, $hash if $hash and $refs;
1451                        $hash = $1;
1452                        next;
1453                }
1454                next unless s{^\s*(git-svn-id:)}{$1};
1455                my ($url, $rev, $uuid) = extract_metadata($_);
1456                if (defined $url && defined $rev) {
1457                        next if $max{$url} and $max{$url} < $rev;
1458                        if (my $gs = Git::SVN->find_by_url($url)) {
1459                                my $c = $gs->rev_map_get($rev, $uuid);
1460                                if ($c && $c eq $hash) {
1461                                        close $fh; # break the pipe
1462                                        return ($url, $rev, $uuid, $gs);
1463                                } else {
1464                                        $max{$url} ||= $gs->rev_map_max;
1465                                }
1466                        }
1467                }
1468        }
1469        command_close_pipe($fh, $ctx);
1470        (undef, undef, undef, undef);
1471}
1472
1473sub read_commit_parents {
1474        my ($parents, $c) = @_;
1475        chomp(my $p = command_oneline(qw/rev-list --parents -1/, $c));
1476        $p =~ s/^($c)\s*// or die "rev-list --parents -1 $c failed!\n";
1477        @{$parents->{$c}} = split(/ /, $p);
1478}
1479
1480sub linearize_history {
1481        my ($gs, $refs) = @_;
1482        my %parents;
1483        foreach my $c (@$refs) {
1484                read_commit_parents(\%parents, $c);
1485        }
1486
1487        my @linear_refs;
1488        my %skip = ();
1489        my $last_svn_commit = $gs->last_commit;
1490        foreach my $c (reverse @$refs) {
1491                next if $c eq $last_svn_commit;
1492                last if $skip{$c};
1493
1494                unshift @linear_refs, $c;
1495                $skip{$c} = 1;
1496
1497                # we only want the first parent to diff against for linear
1498                # history, we save the rest to inject when we finalize the
1499                # svn commit
1500                my $fp_a = verify_ref("$c~1");
1501                my $fp_b = shift @{$parents{$c}} if $parents{$c};
1502                if (!$fp_a || !$fp_b) {
1503                        die "Commit $c\n",
1504                            "has no parent commit, and therefore ",
1505                            "nothing to diff against.\n",
1506                            "You should be working from a repository ",
1507                            "originally created by git-svn\n";
1508                }
1509                if ($fp_a ne $fp_b) {
1510                        die "$c~1 = $fp_a, however parsing commit $c ",
1511                            "revealed that:\n$c~1 = $fp_b\nBUG!\n";
1512                }
1513
1514                foreach my $p (@{$parents{$c}}) {
1515                        $skip{$p} = 1;
1516                }
1517        }
1518        (\@linear_refs, \%parents);
1519}
1520
1521sub find_file_type_and_diff_status {
1522        my ($path) = @_;
1523        return ('dir', '') if $path eq '';
1524
1525        my $diff_output =
1526            command_oneline(qw(diff --cached --name-status --), $path) || "";
1527        my $diff_status = (split(' ', $diff_output))[0] || "";
1528
1529        my $ls_tree = command_oneline(qw(ls-tree HEAD), $path) || "";
1530
1531        return (undef, undef) if !$diff_status && !$ls_tree;
1532
1533        if ($diff_status eq "A") {
1534                return ("link", $diff_status) if -l $path;
1535                return ("dir", $diff_status) if -d $path;
1536                return ("file", $diff_status);
1537        }
1538
1539        my $mode = (split(' ', $ls_tree))[0] || "";
1540
1541        return ("link", $diff_status) if $mode eq "120000";
1542        return ("dir", $diff_status) if $mode eq "040000";
1543        return ("file", $diff_status);
1544}
1545
1546sub md5sum {
1547        my $arg = shift;
1548        my $ref = ref $arg;
1549        my $md5 = Digest::MD5->new();
1550        if ($ref eq 'GLOB' || $ref eq 'IO::File' || $ref eq 'File::Temp') {
1551                $md5->addfile($arg) or croak $!;
1552        } elsif ($ref eq 'SCALAR') {
1553                $md5->add($$arg) or croak $!;
1554        } elsif (!$ref) {
1555                $md5->add($arg) or croak $!;
1556        } else {
1557                ::fatal "Can't provide MD5 hash for unknown ref type: '", $ref, "'";
1558        }
1559        return $md5->hexdigest();
1560}
1561
1562sub gc_directory {
1563        if ($can_compress && -f $_ && basename($_) eq "unhandled.log") {
1564                my $out_filename = $_ . ".gz";
1565                open my $in_fh, "<", $_ or die "Unable to open $_: $!\n";
1566                binmode $in_fh;
1567                my $gz = Compress::Zlib::gzopen($out_filename, "ab") or
1568                                die "Unable to open $out_filename: $!\n";
1569
1570                my $res;
1571                while ($res = sysread($in_fh, my $str, 1024)) {
1572                        $gz->gzwrite($str) or
1573                                die "Unable to write: ".$gz->gzerror()."!\n";
1574                }
1575                unlink $_ or die "unlink $File::Find::name: $!\n";
1576        } elsif (-f $_ && basename($_) eq "index") {
1577                unlink $_ or die "unlink $_: $!\n";
1578        }
1579}
1580
1581package Git::SVN;
1582use strict;
1583use warnings;
1584use Fcntl qw/:DEFAULT :seek/;
1585use constant rev_map_fmt => 'NH40';
1586use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
1587            $_repack $_repack_flags $_use_svm_props $_head
1588            $_use_svnsync_props $no_reuse_existing $_minimize_url
1589            $_use_log_author $_add_author_from $_localtime/;
1590use Carp qw/croak/;
1591use File::Path qw/mkpath/;
1592use File::Copy qw/copy/;
1593use IPC::Open3;
1594
1595my ($_gc_nr, $_gc_period);
1596
1597# properties that we do not log:
1598my %SKIP_PROP;
1599BEGIN {
1600        %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
1601                                        svn:special svn:executable
1602                                        svn:entry:committed-rev
1603                                        svn:entry:last-author
1604                                        svn:entry:uuid
1605                                        svn:entry:committed-date/;
1606
1607        # some options are read globally, but can be overridden locally
1608        # per [svn-remote "..."] section.  Command-line options will *NOT*
1609        # override options set in an [svn-remote "..."] section
1610        no strict 'refs';
1611        for my $option (qw/follow_parent no_metadata use_svm_props
1612                           use_svnsync_props/) {
1613                my $key = $option;
1614                $key =~ tr/_//d;
1615                my $prop = "-$option";
1616                *$option = sub {
1617                        my ($self) = @_;
1618                        return $self->{$prop} if exists $self->{$prop};
1619                        my $k = "svn-remote.$self->{repo_id}.$key";
1620                        eval { command_oneline(qw/config --get/, $k) };
1621                        if ($@) {
1622                                $self->{$prop} = ${"Git::SVN::_$option"};
1623                        } else {
1624                                my $v = command_oneline(qw/config --bool/,$k);
1625                                $self->{$prop} = $v eq 'false' ? 0 : 1;
1626                        }
1627                        return $self->{$prop};
1628                }
1629        }
1630}
1631
1632
1633my (%LOCKFILES, %INDEX_FILES);
1634END {
1635        unlink keys %LOCKFILES if %LOCKFILES;
1636        unlink keys %INDEX_FILES if %INDEX_FILES;
1637}
1638
1639sub resolve_local_globs {
1640        my ($url, $fetch, $glob_spec) = @_;
1641        return unless defined $glob_spec;
1642        my $ref = $glob_spec->{ref};
1643        my $path = $glob_spec->{path};
1644        foreach (command(qw#for-each-ref --format=%(refname) refs/remotes#)) {
1645                next unless m#^refs/remotes/$ref->{regex}$#;
1646                my $p = $1;
1647                my $pathname = desanitize_refname($path->full_path($p));
1648                my $refname = desanitize_refname($ref->full_path($p));
1649                if (my $existing = $fetch->{$pathname}) {
1650                        if ($existing ne $refname) {
1651                                die "Refspec conflict:\n",
1652                                    "existing: refs/remotes/$existing\n",
1653                                    " globbed: refs/remotes/$refname\n";
1654                        }
1655                        my $u = (::cmt_metadata("refs/remotes/$refname"))[0];
1656                        $u =~ s!^\Q$url\E(/|$)!! or die
1657                          "refs/remotes/$refname: '$url' not found in '$u'\n";
1658                        if ($pathname ne $u) {
1659                                warn "W: Refspec glob conflict ",
1660                                     "(ref: refs/remotes/$refname):\n",
1661                                     "expected path: $pathname\n",
1662                                     "    real path: $u\n",
1663                                     "Continuing ahead with $u\n";
1664                                next;
1665                        }
1666                } else {
1667                        $fetch->{$pathname} = $refname;
1668                }
1669        }
1670}
1671
1672sub parse_revision_argument {
1673        my ($base, $head) = @_;
1674        if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
1675                return ($base, $head);
1676        }
1677        return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
1678        return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
1679        return ($head, $head) if ($::_revision eq 'HEAD');
1680        return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
1681        return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
1682        die "revision argument: $::_revision not understood by git-svn\n";
1683}
1684
1685sub fetch_all {
1686        my ($repo_id, $remotes) = @_;
1687        if (ref $repo_id) {
1688                my $gs = $repo_id;
1689                $repo_id = undef;
1690                $repo_id = $gs->{repo_id};
1691        }
1692        $remotes ||= read_all_remotes();
1693        my $remote = $remotes->{$repo_id} or
1694                     die "[svn-remote \"$repo_id\"] unknown\n";
1695        my $fetch = $remote->{fetch};
1696        my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
1697        my (@gs, @globs);
1698        my $ra = Git::SVN::Ra->new($url);
1699        my $uuid = $ra->get_uuid;
1700        my $head = $ra->get_latest_revnum;
1701        $ra->get_log("", $head, 0, 1, 0, 1, sub { $head = $_[1] });
1702        my $base = defined $fetch ? $head : 0;
1703
1704        # read the max revs for wildcard expansion (branches/*, tags/*)
1705        foreach my $t (qw/branches tags/) {
1706                defined $remote->{$t} or next;
1707                push @globs, @{$remote->{$t}};
1708
1709                my $max_rev = eval { tmp_config(qw/--int --get/,
1710                                         "svn-remote.$repo_id.${t}-maxRev") };
1711                if (defined $max_rev && ($max_rev < $base)) {
1712                        $base = $max_rev;
1713                } elsif (!defined $max_rev) {
1714                        $base = 0;
1715                }
1716        }
1717
1718        if ($fetch) {
1719                foreach my $p (sort keys %$fetch) {
1720                        my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
1721                        my $lr = $gs->rev_map_max;
1722                        if (defined $lr) {
1723                                $base = $lr if ($lr < $base);
1724                        }
1725                        push @gs, $gs;
1726                }
1727        }
1728
1729        ($base, $head) = parse_revision_argument($base, $head);
1730        $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
1731}
1732
1733sub read_all_remotes {
1734        my $r = {};
1735        my $use_svm_props = eval { command_oneline(qw/config --bool
1736            svn.useSvmProps/) };
1737        $use_svm_props = $use_svm_props eq 'true' if $use_svm_props;
1738        foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
1739                if (m!^(.+)\.fetch=\s*(.*)\s*:\s*(.+)\s*$!) {
1740                        my ($remote, $local_ref, $_remote_ref) = ($1, $2, $3);
1741                        die("svn-remote.$remote: remote ref '$_remote_ref' "
1742                            . "must start with 'refs/remotes/'\n")
1743                                unless $_remote_ref =~ m{^refs/remotes/(.+)};
1744                        my $remote_ref = $1;
1745                        $local_ref =~ s{^/}{};
1746                        $r->{$remote}->{fetch}->{$local_ref} = $remote_ref;
1747                        $r->{$remote}->{svm} = {} if $use_svm_props;
1748                } elsif (m!^(.+)\.usesvmprops=\s*(.*)\s*$!) {
1749                        $r->{$1}->{svm} = {};
1750                } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
1751                        $r->{$1}->{url} = $2;
1752                } elsif (m!^(.+)\.(branches|tags)=
1753                           (.*):refs/remotes/(.+)\s*$/!x) {
1754                        my ($p, $g) = ($3, $4);
1755                        my $rs = {
1756                            t => $2,
1757                            remote => $1,
1758                            path => Git::SVN::GlobSpec->new($p),
1759                            ref => Git::SVN::GlobSpec->new($g) };
1760                        if (length($rs->{ref}->{right}) != 0) {
1761                                die "The '*' glob character must be the last ",
1762                                    "character of '$g'\n";
1763                        }
1764                        push @{ $r->{$1}->{$2} }, $rs;
1765                }
1766        }
1767
1768        map {
1769                if (defined $r->{$_}->{svm}) {
1770                        my $svm;
1771                        eval {
1772                                my $section = "svn-remote.$_";
1773                                $svm = {
1774                                        source => tmp_config('--get',
1775                                            "$section.svm-source"),
1776                                        replace => tmp_config('--get',
1777                                            "$section.svm-replace"),
1778                                }
1779                        };
1780                        $r->{$_}->{svm} = $svm;
1781                }
1782        } keys %$r;
1783
1784        $r;
1785}
1786
1787sub init_vars {
1788        $_gc_nr = $_gc_period = 1000;
1789        if (defined $_repack || defined $_repack_flags) {
1790               warn "Repack options are obsolete; they have no effect.\n";
1791        }
1792}
1793
1794sub verify_remotes_sanity {
1795        return unless -d $ENV{GIT_DIR};
1796        my %seen;
1797        foreach (command(qw/config -l/)) {
1798                if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
1799                        if ($seen{$1}) {
1800                                die "Remote ref refs/remote/$1 is tracked by",
1801                                    "\n  \"$_\"\nand\n  \"$seen{$1}\"\n",
1802                                    "Please resolve this ambiguity in ",
1803                                    "your git configuration file before ",
1804                                    "continuing\n";
1805                        }
1806                        $seen{$1} = $_;
1807                }
1808        }
1809}
1810
1811sub find_existing_remote {
1812        my ($url, $remotes) = @_;
1813        return undef if $no_reuse_existing;
1814        my $existing;
1815        foreach my $repo_id (keys %$remotes) {
1816                my $u = $remotes->{$repo_id}->{url} or next;
1817                next if $u ne $url;
1818                $existing = $repo_id;
1819                last;
1820        }
1821        $existing;
1822}
1823
1824sub init_remote_config {
1825        my ($self, $url, $no_write) = @_;
1826        $url =~ s!/+$!!; # strip trailing slash
1827        my $r = read_all_remotes();
1828        my $existing = find_existing_remote($url, $r);
1829        if ($existing) {
1830                unless ($no_write) {
1831                        print STDERR "Using existing ",
1832                                     "[svn-remote \"$existing\"]\n";
1833                }
1834                $self->{repo_id} = $existing;
1835        } elsif ($_minimize_url) {
1836                my $min_url = Git::SVN::Ra->new($url)->minimize_url;
1837                $existing = find_existing_remote($min_url, $r);
1838                if ($existing) {
1839                        unless ($no_write) {
1840                                print STDERR "Using existing ",
1841                                             "[svn-remote \"$existing\"]\n";
1842                        }
1843                        $self->{repo_id} = $existing;
1844                }
1845                if ($min_url ne $url) {
1846                        unless ($no_write) {
1847                                print STDERR "Using higher level of URL: ",
1848                                             "$url => $min_url\n";
1849                        }
1850                        my $old_path = $self->{path};
1851                        $self->{path} = $url;
1852                        $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
1853                        if (length $old_path) {
1854                                $self->{path} .= "/$old_path";
1855                        }
1856                        $url = $min_url;
1857                }
1858        }
1859        my $orig_url;
1860        if (!$existing) {
1861                # verify that we aren't overwriting anything:
1862                $orig_url = eval {
1863                        command_oneline('config', '--get',
1864                                        "svn-remote.$self->{repo_id}.url")
1865                };
1866                if ($orig_url && ($orig_url ne $url)) {
1867                        die "svn-remote.$self->{repo_id}.url already set: ",
1868                            "$orig_url\nwanted to set to: $url\n";
1869                }
1870        }
1871        my ($xrepo_id, $xpath) = find_ref($self->refname);
1872        if (defined $xpath) {
1873                die "svn-remote.$xrepo_id.fetch already set to track ",
1874                    "$xpath:refs/remotes/", $self->refname, "\n";
1875        }
1876        unless ($no_write) {
1877                command_noisy('config',
1878                              "svn-remote.$self->{repo_id}.url", $url);
1879                $self->{path} =~ s{^/}{};
1880                command_noisy('config', '--add',
1881                              "svn-remote.$self->{repo_id}.fetch",
1882                              "$self->{path}:".$self->refname);
1883        }
1884        $self->{url} = $url;
1885}
1886
1887sub find_by_url { # repos_root and, path are optional
1888        my ($class, $full_url, $repos_root, $path) = @_;
1889
1890        return undef unless defined $full_url;
1891        remove_username($full_url);
1892        remove_username($repos_root) if defined $repos_root;
1893        my $remotes = read_all_remotes();
1894        if (defined $full_url && defined $repos_root && !defined $path) {
1895                $path = $full_url;
1896                $path =~ s#^\Q$repos_root\E(?:/|$)##;
1897        }
1898        foreach my $repo_id (keys %$remotes) {
1899                my $u = $remotes->{$repo_id}->{url} or next;
1900                remove_username($u);
1901                next if defined $repos_root && $repos_root ne $u;
1902
1903                my $fetch = $remotes->{$repo_id}->{fetch} || {};
1904                foreach my $t (qw/branches tags/) {
1905                        foreach my $globspec (@{$remotes->{$repo_id}->{$t}}) {
1906                                resolve_local_globs($u, $fetch, $globspec);
1907                        }
1908                }
1909                my $p = $path;
1910                my $rwr = rewrite_root({repo_id => $repo_id});
1911                my $svm = $remotes->{$repo_id}->{svm}
1912                        if defined $remotes->{$repo_id}->{svm};
1913                unless (defined $p) {
1914                        $p = $full_url;
1915                        my $z = $u;
1916                        my $prefix = '';
1917                        if ($rwr) {
1918                                $z = $rwr;
1919                                remove_username($z);
1920                        } elsif (defined $svm) {
1921                                $z = $svm->{source};
1922                                $prefix = $svm->{replace};
1923                                $prefix =~ s#^\Q$u\E(?:/|$)##;
1924                                $prefix =~ s#/$##;
1925                        }
1926                        $p =~ s#^\Q$z\E(?:/|$)#$prefix# or next;
1927                }
1928                foreach my $f (keys %$fetch) {
1929                        next if $f ne $p;
1930                        return Git::SVN->new($fetch->{$f}, $repo_id, $f);
1931                }
1932        }
1933        undef;
1934}
1935
1936sub init {
1937        my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
1938        my $self = _new($class, $repo_id, $ref_id, $path);
1939        if (defined $url) {
1940                $self->init_remote_config($url, $no_write);
1941        }
1942        $self;
1943}
1944
1945sub find_ref {
1946        my ($ref_id) = @_;
1947        foreach (command(qw/config -l/)) {
1948                next unless m!^svn-remote\.(.+)\.fetch=
1949                              \s*(.*)\s*:\s*refs/remotes/(.+)\s*$!x;
1950                my ($repo_id, $path, $ref) = ($1, $2, $3);
1951                if ($ref eq $ref_id) {
1952                        $path = '' if ($path =~ m#^\./?#);
1953                        return ($repo_id, $path);
1954                }
1955        }
1956        (undef, undef, undef);
1957}
1958
1959sub new {
1960        my ($class, $ref_id, $repo_id, $path) = @_;
1961        if (defined $ref_id && !defined $repo_id && !defined $path) {
1962                ($repo_id, $path) = find_ref($ref_id);
1963                if (!defined $repo_id) {
1964                        die "Could not find a \"svn-remote.*.fetch\" key ",
1965                            "in the repository configuration matching: ",
1966                            "refs/remotes/$ref_id\n";
1967                }
1968        }
1969        my $self = _new($class, $repo_id, $ref_id, $path);
1970        if (!defined $self->{path} || !length $self->{path}) {
1971                my $fetch = command_oneline('config', '--get',
1972                                            "svn-remote.$repo_id.fetch",
1973                                            ":refs/remotes/$ref_id\$") or
1974                     die "Failed to read \"svn-remote.$repo_id.fetch\" ",
1975                         "\":refs/remotes/$ref_id\$\" in config\n";
1976                ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
1977        }
1978        $self->{url} = command_oneline('config', '--get',
1979                                       "svn-remote.$repo_id.url") or
1980                  die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
1981        $self->rebuild;
1982        $self;
1983}
1984
1985sub refname {
1986        my ($refname) = "refs/remotes/$_[0]->{ref_id}" ;
1987
1988        # It cannot end with a slash /, we'll throw up on this because
1989        # SVN can't have directories with a slash in their name, either:
1990        if ($refname =~ m{/$}) {
1991                die "ref: '$refname' ends with a trailing slash, this is ",
1992                    "not permitted by git nor Subversion\n";
1993        }
1994
1995        # It cannot have ASCII control character space, tilde ~, caret ^,
1996        # colon :, question-mark ?, asterisk *, space, or open bracket [
1997        # anywhere.
1998        #
1999        # Additionally, % must be escaped because it is used for escaping
2000        # and we want our escaped refname to be reversible
2001        $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
2002
2003        # no slash-separated component can begin with a dot .
2004        # /.* becomes /%2E*
2005        $refname =~ s{/\.}{/%2E}g;
2006
2007        # It cannot have two consecutive dots .. anywhere
2008        # .. becomes %2E%2E
2009        $refname =~ s{\.\.}{%2E%2E}g;
2010
2011        return $refname;
2012}
2013
2014sub desanitize_refname {
2015        my ($refname) = @_;
2016        $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
2017        return $refname;
2018}
2019
2020sub svm_uuid {
2021        my ($self) = @_;
2022        return $self->{svm}->{uuid} if $self->svm;
2023        $self->ra;
2024        unless ($self->{svm}) {
2025                die "SVM UUID not cached, and reading remotely failed\n";
2026        }
2027        $self->{svm}->{uuid};
2028}
2029
2030sub svm {
2031        my ($self) = @_;
2032        return $self->{svm} if $self->{svm};
2033        my $svm;
2034        # see if we have it in our config, first:
2035        eval {
2036                my $section = "svn-remote.$self->{repo_id}";
2037                $svm = {
2038                  source => tmp_config('--get', "$section.svm-source"),
2039                  uuid => tmp_config('--get', "$section.svm-uuid"),
2040                  replace => tmp_config('--get', "$section.svm-replace"),
2041                }
2042        };
2043        if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
2044                $self->{svm} = $svm;
2045        }
2046        $self->{svm};
2047}
2048
2049sub _set_svm_vars {
2050        my ($self, $ra) = @_;
2051        return $ra if $self->svm;
2052
2053        my @err = ( "useSvmProps set, but failed to read SVM properties\n",
2054                    "(svm:source, svm:uuid) ",
2055                    "from the following URLs:\n" );
2056        sub read_svm_props {
2057                my ($self, $ra, $path, $r) = @_;
2058                my $props = ($ra->get_dir($path, $r))[2];
2059                my $src = $props->{'svm:source'};
2060                my $uuid = $props->{'svm:uuid'};
2061                return undef if (!$src || !$uuid);
2062
2063                chomp($src, $uuid);
2064
2065                $uuid =~ m{^[0-9a-f\-]{30,}$}i
2066                    or die "doesn't look right - svm:uuid is '$uuid'\n";
2067
2068                # the '!' is used to mark the repos_root!/relative/path
2069                $src =~ s{/?!/?}{/};
2070                $src =~ s{/+$}{}; # no trailing slashes please
2071                # username is of no interest
2072                $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
2073
2074                my $replace = $ra->{url};
2075                $replace .= "/$path" if length $path;
2076
2077                my $section = "svn-remote.$self->{repo_id}";
2078                tmp_config("$section.svm-source", $src);
2079                tmp_config("$section.svm-replace", $replace);
2080                tmp_config("$section.svm-uuid", $uuid);
2081                $self->{svm} = {
2082                        source => $src,
2083                        uuid => $uuid,
2084                        replace => $replace
2085                };
2086        }
2087
2088        my $r = $ra->get_latest_revnum;
2089        my $path = $self->{path};
2090        my %tried;
2091        while (length $path) {
2092                unless ($tried{"$self->{url}/$path"}) {
2093                        return $ra if $self->read_svm_props($ra, $path, $r);
2094                        $tried{"$self->{url}/$path"} = 1;
2095                }
2096                $path =~ s#/?[^/]+$##;
2097        }
2098        die "Path: '$path' should be ''\n" if $path ne '';
2099        return $ra if $self->read_svm_props($ra, $path, $r);
2100        $tried{"$self->{url}/$path"} = 1;
2101
2102        if ($ra->{repos_root} eq $self->{url}) {
2103                die @err, (map { "  $_\n" } keys %tried), "\n";
2104        }
2105
2106        # nope, make sure we're connected to the repository root:
2107        my $ok;
2108        my @tried_b;
2109        $path = $ra->{svn_path};
2110        $ra = Git::SVN::Ra->new($ra->{repos_root});
2111        while (length $path) {
2112                unless ($tried{"$ra->{url}/$path"}) {
2113                        $ok = $self->read_svm_props($ra, $path, $r);
2114                        last if $ok;
2115                        $tried{"$ra->{url}/$path"} = 1;
2116                }
2117                $path =~ s#/?[^/]+$##;
2118        }
2119        die "Path: '$path' should be ''\n" if $path ne '';
2120        $ok ||= $self->read_svm_props($ra, $path, $r);
2121        $tried{"$ra->{url}/$path"} = 1;
2122        if (!$ok) {
2123                die @err, (map { "  $_\n" } keys %tried), "\n";
2124        }
2125        Git::SVN::Ra->new($self->{url});
2126}
2127
2128sub svnsync {
2129        my ($self) = @_;
2130        return $self->{svnsync} if $self->{svnsync};
2131
2132        if ($self->no_metadata) {
2133                die "Can't have both 'noMetadata' and ",
2134                    "'useSvnsyncProps' options set!\n";
2135        }
2136        if ($self->rewrite_root) {
2137                die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
2138                    "options set!\n";
2139        }
2140
2141        my $svnsync;
2142        # see if we have it in our config, first:
2143        eval {
2144                my $section = "svn-remote.$self->{repo_id}";
2145
2146                my $url = tmp_config('--get', "$section.svnsync-url");
2147                ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2148                   die "doesn't look right - svn:sync-from-url is '$url'\n";
2149
2150                my $uuid = tmp_config('--get', "$section.svnsync-uuid");
2151                ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2152                   die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2153
2154                $svnsync = { url => $url, uuid => $uuid }
2155        };
2156        if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
2157                return $self->{svnsync} = $svnsync;
2158        }
2159
2160        my $err = "useSvnsyncProps set, but failed to read " .
2161                  "svnsync property: svn:sync-from-";
2162        my $rp = $self->ra->rev_proplist(0);
2163
2164        my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
2165        ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2166                   die "doesn't look right - svn:sync-from-url is '$url'\n";
2167
2168        my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
2169        ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2170                   die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2171
2172        my $section = "svn-remote.$self->{repo_id}";
2173        tmp_config('--add', "$section.svnsync-uuid", $uuid);
2174        tmp_config('--add', "$section.svnsync-url", $url);
2175        return $self->{svnsync} = { url => $url, uuid => $uuid };
2176}
2177
2178# this allows us to memoize our SVN::Ra UUID locally and avoid a
2179# remote lookup (useful for 'git svn log').
2180sub ra_uuid {
2181        my ($self) = @_;
2182        unless ($self->{ra_uuid}) {
2183                my $key = "svn-remote.$self->{repo_id}.uuid";
2184                my $uuid = eval { tmp_config('--get', $key) };
2185                if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/i) {
2186                        $self->{ra_uuid} = $uuid;
2187                } else {
2188                        die "ra_uuid called without URL\n" unless $self->{url};
2189                        $self->{ra_uuid} = $self->ra->get_uuid;
2190                        tmp_config('--add', $key, $self->{ra_uuid});
2191                }
2192        }
2193        $self->{ra_uuid};
2194}
2195
2196sub _set_repos_root {
2197        my ($self, $repos_root) = @_;
2198        my $k = "svn-remote.$self->{repo_id}.reposRoot";
2199        $repos_root ||= $self->ra->{repos_root};
2200        tmp_config($k, $repos_root);
2201        $repos_root;
2202}
2203
2204sub repos_root {
2205        my ($self) = @_;
2206        my $k = "svn-remote.$self->{repo_id}.reposRoot";
2207        eval { tmp_config('--get', $k) } || $self->_set_repos_root;
2208}
2209
2210sub ra {
2211        my ($self) = shift;
2212        my $ra = Git::SVN::Ra->new($self->{url});
2213        $self->_set_repos_root($ra->{repos_root});
2214        if ($self->use_svm_props && !$self->{svm}) {
2215                if ($self->no_metadata) {
2216                        die "Can't have both 'noMetadata' and ",
2217                            "'useSvmProps' options set!\n";
2218                } elsif ($self->use_svnsync_props) {
2219                        die "Can't have both 'useSvnsyncProps' and ",
2220                            "'useSvmProps' options set!\n";
2221                }
2222                $ra = $self->_set_svm_vars($ra);
2223                $self->{-want_revprops} = 1;
2224        }
2225        $ra;
2226}
2227
2228# prop_walk(PATH, REV, SUB)
2229# -------------------------
2230# Recursively traverse PATH at revision REV and invoke SUB for each
2231# directory that contains a SVN property.  SUB will be invoked as
2232# follows:  &SUB(gs, path, props);  where `gs' is this instance of
2233# Git::SVN, `path' the path to the directory where the properties
2234# `props' were found.  The `path' will be relative to point of checkout,
2235# that is, if url://repo/trunk is the current Git branch, and that
2236# directory contains a sub-directory `d', SUB will be invoked with `/d/'
2237# as `path' (note the trailing `/').
2238sub prop_walk {
2239        my ($self, $path, $rev, $sub) = @_;
2240
2241        $path =~ s#^/##;
2242        my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev);
2243        $path =~ s#^/*#/#g;
2244        my $p = $path;
2245        # Strip the irrelevant part of the path.
2246        $p =~ s#^/+\Q$self->{path}\E(/|$)#/#;
2247        # Ensure the path is terminated by a `/'.
2248        $p =~ s#/*$#/#;
2249
2250        # The properties contain all the internal SVN stuff nobody
2251        # (usually) cares about.
2252        my $interesting_props = 0;
2253        foreach (keys %{$props}) {
2254                # If it doesn't start with `svn:', it must be a
2255                # user-defined property.
2256                ++$interesting_props and next if $_ !~ /^svn:/;
2257                # FIXME: Fragile, if SVN adds new public properties,
2258                # this needs to be updated.
2259                ++$interesting_props if /^svn:(?:ignore|keywords|executable
2260                                                 |eol-style|mime-type
2261                                                 |externals|needs-lock)$/x;
2262        }
2263        &$sub($self, $p, $props) if $interesting_props;
2264
2265        foreach (sort keys %$dirent) {
2266                next if $dirent->{$_}->{kind} != $SVN::Node::dir;
2267                $self->prop_walk($self->{path} . $p . $_, $rev, $sub);
2268        }
2269}
2270
2271sub last_rev { ($_[0]->last_rev_commit)[0] }
2272sub last_commit { ($_[0]->last_rev_commit)[1] }
2273
2274# returns the newest SVN revision number and newest commit SHA1
2275sub last_rev_commit {
2276        my ($self) = @_;
2277        if (defined $self->{last_rev} && defined $self->{last_commit}) {
2278                return ($self->{last_rev}, $self->{last_commit});
2279        }
2280        my $c = ::verify_ref($self->refname.'^0');
2281        if ($c && !$self->use_svm_props && !$self->no_metadata) {
2282                my $rev = (::cmt_metadata($c))[1];
2283                if (defined $rev) {
2284                        ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
2285                        return ($rev, $c);
2286                }
2287        }
2288        my $map_path = $self->map_path;
2289        unless (-e $map_path) {
2290                ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
2291                return (undef, undef);
2292        }
2293        my ($rev, $commit) = $self->rev_map_max(1);
2294        ($self->{last_rev}, $self->{last_commit}) = ($rev, $commit);
2295        return ($rev, $commit);
2296}
2297
2298sub get_fetch_range {
2299        my ($self, $min, $max) = @_;
2300        $max ||= $self->ra->get_latest_revnum;
2301        $min ||= $self->rev_map_max;
2302        (++$min, $max);
2303}
2304
2305sub tmp_config {
2306        my (@args) = @_;
2307        my $old_def_config = "$ENV{GIT_DIR}/svn/config";
2308        my $config = "$ENV{GIT_DIR}/svn/.metadata";
2309        if (! -f $config && -f $old_def_config) {
2310                rename $old_def_config, $config or
2311                       die "Failed rename $old_def_config => $config: $!\n";
2312        }
2313        my $old_config = $ENV{GIT_CONFIG};
2314        $ENV{GIT_CONFIG} = $config;
2315        $@ = undef;
2316        my @ret = eval {
2317                unless (-f $config) {
2318                        mkfile($config);
2319                        open my $fh, '>', $config or
2320                            die "Can't open $config: $!\n";
2321                        print $fh "; This file is used internally by ",
2322                                  "git-svn\n" or die
2323                                  "Couldn't write to $config: $!\n";
2324                        print $fh "; You should not have to edit it\n" or
2325                              die "Couldn't write to $config: $!\n";
2326                        close $fh or die "Couldn't close $config: $!\n";
2327                }
2328                command('config', @args);
2329        };
2330        my $err = $@;
2331        if (defined $old_config) {
2332                $ENV{GIT_CONFIG} = $old_config;
2333        } else {
2334                delete $ENV{GIT_CONFIG};
2335        }
2336        die $err if $err;
2337        wantarray ? @ret : $ret[0];
2338}
2339
2340sub tmp_index_do {
2341        my ($self, $sub) = @_;
2342        my $old_index = $ENV{GIT_INDEX_FILE};
2343        $ENV{GIT_INDEX_FILE} = $self->{index};
2344        $@ = undef;
2345        my @ret = eval {
2346                my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
2347                mkpath([$dir]) unless -d $dir;
2348                &$sub;
2349        };
2350        my $err = $@;
2351        if (defined $old_index) {
2352                $ENV{GIT_INDEX_FILE} = $old_index;
2353        } else {
2354                delete $ENV{GIT_INDEX_FILE};
2355        }
2356        die $err if $err;
2357        wantarray ? @ret : $ret[0];
2358}
2359
2360sub assert_index_clean {
2361        my ($self, $treeish) = @_;
2362
2363        $self->tmp_index_do(sub {
2364                command_noisy('read-tree', $treeish) unless -e $self->{index};
2365                my $x = command_oneline('write-tree');
2366                my ($y) = (command(qw/cat-file commit/, $treeish) =~
2367                           /^tree ($::sha1)/mo);
2368                return if $y eq $x;
2369
2370                warn "Index mismatch: $y != $x\nrereading $treeish\n";
2371                unlink $self->{index} or die "unlink $self->{index}: $!\n";
2372                command_noisy('read-tree', $treeish);
2373                $x = command_oneline('write-tree');
2374                if ($y ne $x) {
2375                        ::fatal "trees ($treeish) $y != $x\n",
2376                                "Something is seriously wrong...";
2377                }
2378        });
2379}
2380
2381sub get_commit_parents {
2382        my ($self, $log_entry) = @_;
2383        my (%seen, @ret, @tmp);
2384        # legacy support for 'set-tree'; this is only used by set_tree_cb:
2385        if (my $ip = $self->{inject_parents}) {
2386                if (my $commit = delete $ip->{$log_entry->{revision}}) {
2387                        push @tmp, $commit;
2388                }
2389        }
2390        if (my $cur = ::verify_ref($self->refname.'^0')) {
2391                push @tmp, $cur;
2392        }
2393        if (my $ipd = $self->{inject_parents_dcommit}) {
2394                if (my $commit = delete $ipd->{$log_entry->{revision}}) {
2395                        push @tmp, @$commit;
2396                }
2397        }
2398        push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
2399        while (my $p = shift @tmp) {
2400                next if $seen{$p};
2401                $seen{$p} = 1;
2402                push @ret, $p;
2403                # MAXPARENT is defined to 16 in commit-tree.c:
2404                last if @ret >= 16;
2405        }
2406        if (@tmp) {
2407                die "r$log_entry->{revision}: No room for parents:\n\t",
2408                    join("\n\t", @tmp), "\n";
2409        }
2410        @ret;
2411}
2412
2413sub rewrite_root {
2414        my ($self) = @_;
2415        return $self->{-rewrite_root} if exists $self->{-rewrite_root};
2416        my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
2417        my $rwr = eval { command_oneline(qw/config --get/, $k) };
2418        if ($rwr) {
2419                $rwr =~ s#/+$##;
2420                if ($rwr !~ m#^[a-z\+]+://#) {
2421                        die "$rwr is not a valid URL (key: $k)\n";
2422                }
2423        }
2424        $self->{-rewrite_root} = $rwr;
2425}
2426
2427sub metadata_url {
2428        my ($self) = @_;
2429        ($self->rewrite_root || $self->{url}) .
2430           (length $self->{path} ? '/' . $self->{path} : '');
2431}
2432
2433sub full_url {
2434        my ($self) = @_;
2435        $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
2436}
2437
2438
2439sub set_commit_header_env {
2440        my ($log_entry) = @_;
2441        my %env;
2442        foreach my $ned (qw/NAME EMAIL DATE/) {
2443                foreach my $ac (qw/AUTHOR COMMITTER/) {
2444                        $env{"GIT_${ac}_${ned}"} = $ENV{"GIT_${ac}_${ned}"};
2445                }
2446        }
2447
2448        $ENV{GIT_AUTHOR_NAME} = $log_entry->{name};
2449        $ENV{GIT_AUTHOR_EMAIL} = $log_entry->{email};
2450        $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
2451
2452        $ENV{GIT_COMMITTER_NAME} = (defined $log_entry->{commit_name})
2453                                                ? $log_entry->{commit_name}
2454                                                : $log_entry->{name};
2455        $ENV{GIT_COMMITTER_EMAIL} = (defined $log_entry->{commit_email})
2456                                                ? $log_entry->{commit_email}
2457                                                : $log_entry->{email};
2458        \%env;
2459}
2460
2461sub restore_commit_header_env {
2462        my ($env) = @_;
2463        foreach my $ned (qw/NAME EMAIL DATE/) {
2464                foreach my $ac (qw/AUTHOR COMMITTER/) {
2465                        my $k = "GIT_${ac}_${ned}";
2466                        if (defined $env->{$k}) {
2467                                $ENV{$k} = $env->{$k};
2468                        } else {
2469                                delete $ENV{$k};
2470                        }
2471                }
2472        }
2473}
2474
2475sub gc {
2476        command_noisy('gc', '--auto');
2477};
2478
2479sub do_git_commit {
2480        my ($self, $log_entry) = @_;
2481        my $lr = $self->last_rev;
2482        if (defined $lr && $lr >= $log_entry->{revision}) {
2483                die "Last fetched revision of ", $self->refname,
2484                    " was r$lr, but we are about to fetch: ",
2485                    "r$log_entry->{revision}!\n";
2486        }
2487        if (my $c = $self->rev_map_get($log_entry->{revision})) {
2488                croak "$log_entry->{revision} = $c already exists! ",
2489                      "Why are we refetching it?\n";
2490        }
2491        my $old_env = set_commit_header_env($log_entry);
2492        my $tree = $log_entry->{tree};
2493        if (!defined $tree) {
2494                $tree = $self->tmp_index_do(sub {
2495                                            command_oneline('write-tree') });
2496        }
2497        die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
2498
2499        my @exec = ('git', 'commit-tree', $tree);
2500        foreach ($self->get_commit_parents($log_entry)) {
2501                push @exec, '-p', $_;
2502        }
2503        defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
2504                                                                   or croak $!;
2505        binmode $msg_fh;
2506
2507        # we always get UTF-8 from SVN, but we may want our commits in
2508        # a different encoding.
2509        if (my $enc = Git::config('i18n.commitencoding')) {
2510                require Encode;
2511                Encode::from_to($log_entry->{log}, 'UTF-8', $enc);
2512        }
2513        print $msg_fh $log_entry->{log} or croak $!;
2514        restore_commit_header_env($old_env);
2515        unless ($self->no_metadata) {
2516                print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
2517                              or croak $!;
2518        }
2519        $msg_fh->flush == 0 or croak $!;
2520        close $msg_fh or croak $!;
2521        chomp(my $commit = do { local $/; <$out_fh> });
2522        close $out_fh or croak $!;
2523        waitpid $pid, 0;
2524        croak $? if $?;
2525        if ($commit !~ /^$::sha1$/o) {
2526                die "Failed to commit, invalid sha1: $commit\n";
2527        }
2528
2529        $self->rev_map_set($log_entry->{revision}, $commit, 1);
2530
2531        $self->{last_rev} = $log_entry->{revision};
2532        $self->{last_commit} = $commit;
2533        print "r$log_entry->{revision}" unless $::_q > 1;
2534        if (defined $log_entry->{svm_revision}) {
2535                 print " (\@$log_entry->{svm_revision})" unless $::_q > 1;
2536                 $self->rev_map_set($log_entry->{svm_revision}, $commit,
2537                                   0, $self->svm_uuid);
2538        }
2539        print " = $commit ($self->{ref_id})\n" unless $::_q > 1;
2540        if (--$_gc_nr == 0) {
2541                $_gc_nr = $_gc_period;
2542                gc();
2543        }
2544        return $commit;
2545}
2546
2547sub match_paths {
2548        my ($self, $paths, $r) = @_;
2549        return 1 if $self->{path} eq '';
2550        if (my $path = $paths->{"/$self->{path}"}) {
2551                return ($path->{action} eq 'D') ? 0 : 1;
2552        }
2553        $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
2554        if (grep /$self->{path_regex}/, keys %$paths) {
2555                return 1;
2556        }
2557        my $c = '';
2558        foreach (split m#/#, $self->{path}) {
2559                $c .= "/$_";
2560                next unless ($paths->{$c} &&
2561                             ($paths->{$c}->{action} =~ /^[AR]$/));
2562                if ($self->ra->check_path($self->{path}, $r) ==
2563                    $SVN::Node::dir) {
2564                        return 1;
2565                }
2566        }
2567        return 0;
2568}
2569
2570sub find_parent_branch {
2571        my ($self, $paths, $rev) = @_;
2572        return undef unless $self->follow_parent;
2573        unless (defined $paths) {
2574                my $err_handler = $SVN::Error::handler;
2575                $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
2576                $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1,
2577                                   sub { $paths = $_[0] });
2578                $SVN::Error::handler = $err_handler;
2579        }
2580        return undef unless defined $paths;
2581
2582        # look for a parent from another branch:
2583        my @b_path_components = split m#/#, $self->{path};
2584        my @a_path_components;
2585        my $i;
2586        while (@b_path_components) {
2587                $i = $paths->{'/'.join('/', @b_path_components)};
2588                last if $i && defined $i->{copyfrom_path};
2589                unshift(@a_path_components, pop(@b_path_components));
2590        }
2591        return undef unless defined $i && defined $i->{copyfrom_path};
2592        my $branch_from = $i->{copyfrom_path};
2593        if (@a_path_components) {
2594                print STDERR "branch_from: $branch_from => ";
2595                $branch_from .= '/'.join('/', @a_path_components);
2596                print STDERR $branch_from, "\n";
2597        }
2598        my $r = $i->{copyfrom_rev};
2599        my $repos_root = $self->ra->{repos_root};
2600        my $url = $self->ra->{url};
2601        my $new_url = $url . $branch_from;
2602        print STDERR  "Found possible branch point: ",
2603                      "$new_url => ", $self->full_url, ", $r\n";
2604        $branch_from =~ s#^/##;
2605        my $gs = $self->other_gs($new_url, $url,
2606                                 $branch_from, $r, $self->{ref_id});
2607        my ($r0, $parent) = $gs->find_rev_before($r, 1);
2608        {
2609                my ($base, $head);
2610                if (!defined $r0 || !defined $parent) {
2611                        ($base, $head) = parse_revision_argument(0, $r);
2612                } else {
2613                        if ($r0 < $r) {
2614                                $gs->ra->get_log([$gs->{path}], $r0 + 1, $r, 1,
2615                                        0, 1, sub { $base = $_[1] - 1 });
2616                        }
2617                }
2618                if (defined $base && $base <= $r) {
2619                        $gs->fetch($base, $r);
2620                }
2621                ($r0, $parent) = $gs->find_rev_before($r, 1);
2622        }
2623        if (defined $r0 && defined $parent) {
2624                print STDERR "Found branch parent: ($self->{ref_id}) $parent\n";
2625                my $ed;
2626                if ($self->ra->can_do_switch) {
2627                        $self->assert_index_clean($parent);
2628                        print STDERR "Following parent with do_switch\n";
2629                        # do_switch works with svn/trunk >= r22312, but that
2630                        # is not included with SVN 1.4.3 (the latest version
2631                        # at the moment), so we can't rely on it
2632                        $self->{last_rev} = $r0;
2633                        $self->{last_commit} = $parent;
2634                        $ed = SVN::Git::Fetcher->new($self, $gs->{path});
2635                        $gs->ra->gs_do_switch($r0, $rev, $gs,
2636                                              $self->full_url, $ed)
2637                          or die "SVN connection failed somewhere...\n";
2638                } elsif ($self->ra->trees_match($new_url, $r0,
2639                                                $self->full_url, $rev)) {
2640                        print STDERR "Trees match:\n",
2641                                     "  $new_url\@$r0\n",
2642                                     "  ${\$self->full_url}\@$rev\n",
2643                                     "Following parent with no changes\n";
2644                        $self->tmp_index_do(sub {
2645                            command_noisy('read-tree', $parent);
2646                        });
2647                        $self->{last_commit} = $parent;
2648                } else {
2649                        print STDERR "Following parent with do_update\n";
2650                        $ed = SVN::Git::Fetcher->new($self);
2651                        $self->ra->gs_do_update($rev, $rev, $self, $ed)
2652                          or die "SVN connection failed somewhere...\n";
2653                }
2654                print STDERR "Successfully followed parent\n";
2655                return $self->make_log_entry($rev, [$parent], $ed);
2656        }
2657        return undef;
2658}
2659
2660sub do_fetch {
2661        my ($self, $paths, $rev) = @_;
2662        my $ed;
2663        my ($last_rev, @parents);
2664        if (my $lc = $self->last_commit) {
2665                # we can have a branch that was deleted, then re-added
2666                # under the same name but copied from another path, in
2667                # which case we'll have multiple parents (we don't
2668                # want to break the original ref, nor lose copypath info):
2669                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2670                        push @{$log_entry->{parents}}, $lc;
2671                        return $log_entry;
2672                }
2673                $ed = SVN::Git::Fetcher->new($self);
2674                $last_rev = $self->{last_rev};
2675                $ed->{c} = $lc;
2676                @parents = ($lc);
2677        } else {
2678                $last_rev = $rev;
2679                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2680                        return $log_entry;
2681                }
2682                $ed = SVN::Git::Fetcher->new($self);
2683        }
2684        unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
2685                die "SVN connection failed somewhere...\n";
2686        }
2687        $self->make_log_entry($rev, \@parents, $ed);
2688}
2689
2690sub get_untracked {
2691        my ($self, $ed) = @_;
2692        my @out;
2693        my $h = $ed->{empty};
2694        foreach (sort keys %$h) {
2695                my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
2696                push @out, "  $act: " . uri_encode($_);
2697                warn "W: $act: $_\n";
2698        }
2699        foreach my $t (qw/dir_prop file_prop/) {
2700                $h = $ed->{$t} or next;
2701                foreach my $path (sort keys %$h) {
2702                        my $ppath = $path eq '' ? '.' : $path;
2703                        foreach my $prop (sort keys %{$h->{$path}}) {
2704                                next if $SKIP_PROP{$prop};
2705                                my $v = $h->{$path}->{$prop};
2706                                my $t_ppath_prop = "$t: " .
2707                                                    uri_encode($ppath) . ' ' .
2708                                                    uri_encode($prop);
2709                                if (defined $v) {
2710                                        push @out, "  +$t_ppath_prop " .
2711                                                   uri_encode($v);
2712                                } else {
2713                                        push @out, "  -$t_ppath_prop";
2714                                }
2715                        }
2716                }
2717        }
2718        foreach my $t (qw/absent_file absent_directory/) {
2719                $h = $ed->{$t} or next;
2720                foreach my $parent (sort keys %$h) {
2721                        foreach my $path (sort @{$h->{$parent}}) {
2722                                push @out, "  $t: " .
2723                                           uri_encode("$parent/$path");
2724                                warn "W: $t: $parent/$path ",
2725                                     "Insufficient permissions?\n";
2726                        }
2727                }
2728        }
2729        \@out;
2730}
2731
2732# parse_svn_date(DATE)
2733# --------------------
2734# Given a date (in UTC) from Subversion, return a string in the format
2735# "<TZ Offset> <local date/time>" that Git will use.
2736#
2737# By default the parsed date will be in UTC; if $Git::SVN::_localtime
2738# is true we'll convert it to the local timezone instead.
2739sub parse_svn_date {
2740        my $date = shift || return '+0000 1970-01-01 00:00:00';
2741        my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2742                                            (\d\d)\:(\d\d)\:(\d\d)\.\d*Z$/x) or
2743                                         croak "Unable to parse date: $date\n";
2744        my $parsed_date;    # Set next.
2745
2746        if ($Git::SVN::_localtime) {
2747                # Translate the Subversion datetime to an epoch time.
2748                # Begin by switching ourselves to $date's timezone, UTC.
2749                my $old_env_TZ = $ENV{TZ};
2750                $ENV{TZ} = 'UTC';
2751
2752                my $epoch_in_UTC =
2753                    POSIX::strftime('%s', $S, $M, $H, $d, $m - 1, $Y - 1900);
2754
2755                # Determine our local timezone (including DST) at the
2756                # time of $epoch_in_UTC.  $Git::SVN::Log::TZ stored the
2757                # value of TZ, if any, at the time we were run.
2758                if (defined $Git::SVN::Log::TZ) {
2759                        $ENV{TZ} = $Git::SVN::Log::TZ;
2760                } else {
2761                        delete $ENV{TZ};
2762                }
2763
2764                my $our_TZ =
2765                    POSIX::strftime('%Z', $S, $M, $H, $d, $m - 1, $Y - 1900);
2766
2767                # This converts $epoch_in_UTC into our local timezone.
2768                my ($sec, $min, $hour, $mday, $mon, $year,
2769                    $wday, $yday, $isdst) = localtime($epoch_in_UTC);
2770
2771                $parsed_date = sprintf('%s %04d-%02d-%02d %02d:%02d:%02d',
2772                                       $our_TZ, $year + 1900, $mon + 1,
2773                                       $mday, $hour, $min, $sec);
2774
2775                # Reset us to the timezone in effect when we entered
2776                # this routine.
2777                if (defined $old_env_TZ) {
2778                        $ENV{TZ} = $old_env_TZ;
2779                } else {
2780                        delete $ENV{TZ};
2781                }
2782        } else {
2783                $parsed_date = "+0000 $Y-$m-$d $H:$M:$S";
2784        }
2785
2786        return $parsed_date;
2787}
2788
2789sub other_gs {
2790        my ($self, $new_url, $url,
2791            $branch_from, $r, $old_ref_id) = @_;
2792        my $gs = Git::SVN->find_by_url($new_url, $url, $branch_from);
2793        unless ($gs) {
2794                my $ref_id = $old_ref_id;
2795                $ref_id =~ s/\@\d+$//;
2796                $ref_id .= "\@$r";
2797                # just grow a tail if we're not unique enough :x
2798                $ref_id .= '-' while find_ref($ref_id);
2799                print STDERR "Initializing parent: $ref_id\n";
2800                my ($u, $p, $repo_id) = ($new_url, '', $ref_id);
2801                if ($u =~ s#^\Q$url\E(/|$)##) {
2802                        $p = $u;
2803                        $u = $url;
2804                        $repo_id = $self->{repo_id};
2805                }
2806                $gs = Git::SVN->init($u, $p, $repo_id, $ref_id, 1);
2807        }
2808        $gs
2809}
2810
2811sub call_authors_prog {
2812        my ($orig_author) = @_;
2813        $orig_author = command_oneline('rev-parse', '--sq-quote', $orig_author);
2814        my $author = `$::_authors_prog $orig_author`;
2815        if ($? != 0) {
2816                die "$::_authors_prog failed with exit code $?\n"
2817        }
2818        if ($author =~ /^\s*(.+?)\s*<(.*)>\s*$/) {
2819                my ($name, $email) = ($1, $2);
2820                $email = undef if length $2 == 0;
2821                return [$name, $email];
2822        } else {
2823                die "Author: $orig_author: $::_authors_prog returned "
2824                        . "invalid author format: $author\n";
2825        }
2826}
2827
2828sub check_author {
2829        my ($author) = @_;
2830        if (!defined $author || length $author == 0) {
2831                $author = '(no author)';
2832        }
2833        if (!defined $::users{$author}) {
2834                if (defined $::_authors_prog) {
2835                        $::users{$author} = call_authors_prog($author);
2836                } elsif (defined $::_authors) {
2837                        die "Author: $author not defined in $::_authors file\n";
2838                }
2839        }
2840        $author;
2841}
2842
2843sub make_log_entry {
2844        my ($self, $rev, $parents, $ed) = @_;
2845        my $untracked = $self->get_untracked($ed);
2846
2847        open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
2848        print $un "r$rev\n" or croak $!;
2849        print $un $_, "\n" foreach @$untracked;
2850        my %log_entry = ( parents => $parents || [], revision => $rev,
2851                          log => '');
2852
2853        my $headrev;
2854        my $logged = delete $self->{logged_rev_props};
2855        if (!$logged || $self->{-want_revprops}) {
2856                my $rp = $self->ra->rev_proplist($rev);
2857                foreach (sort keys %$rp) {
2858                        my $v = $rp->{$_};
2859                        if (/^svn:(author|date|log)$/) {
2860                                $log_entry{$1} = $v;
2861                        } elsif ($_ eq 'svm:headrev') {
2862                                $headrev = $v;
2863                        } else {
2864                                print $un "  rev_prop: ", uri_encode($_), ' ',
2865                                          uri_encode($v), "\n";
2866                        }
2867                }
2868        } else {
2869                map { $log_entry{$_} = $logged->{$_} } keys %$logged;
2870        }
2871        close $un or croak $!;
2872
2873        $log_entry{date} = parse_svn_date($log_entry{date});
2874        $log_entry{log} .= "\n";
2875        my $author = $log_entry{author} = check_author($log_entry{author});
2876        my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
2877                                                       : ($author, undef);
2878
2879        my ($commit_name, $commit_email) = ($name, $email);
2880        if ($_use_log_author) {
2881                my $name_field;
2882                if ($log_entry{log} =~ /From:\s+(.*\S)\s*\n/i) {
2883                        $name_field = $1;
2884                } elsif ($log_entry{log} =~ /Signed-off-by:\s+(.*\S)\s*\n/i) {
2885                        $name_field = $1;
2886                }
2887                if (!defined $name_field) {
2888                        if (!defined $email) {
2889                                $email = $name;
2890                        }
2891                } elsif ($name_field =~ /(.*?)\s+<(.*)>/) {
2892                        ($name, $email) = ($1, $2);
2893                } elsif ($name_field =~ /(.*)@/) {
2894                        ($name, $email) = ($1, $name_field);
2895                } else {
2896                        ($name, $email) = ($name_field, $name_field);
2897                }
2898        }
2899        if (defined $headrev && $self->use_svm_props) {
2900                if ($self->rewrite_root) {
2901                        die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
2902                            "options set!\n";
2903                }
2904                my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$}i;
2905                # we don't want "SVM: initializing mirror for junk" ...
2906                return undef if $r == 0;
2907                my $svm = $self->svm;
2908                if ($uuid ne $svm->{uuid}) {
2909                        die "UUID mismatch on SVM path:\n",
2910                            "expected: $svm->{uuid}\n",
2911                            "     got: $uuid\n";
2912                }
2913                my $full_url = $self->full_url;
2914                $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
2915                             die "Failed to replace '$svm->{replace}' with ",
2916                                 "'$svm->{source}' in $full_url\n";
2917                # throw away username for storing in records
2918                remove_username($full_url);
2919                $log_entry{metadata} = "$full_url\@$r $uuid";
2920                $log_entry{svm_revision} = $r;
2921                $email ||= "$author\@$uuid";
2922                $commit_email ||= "$author\@$uuid";
2923        } elsif ($self->use_svnsync_props) {
2924                my $full_url = $self->svnsync->{url};
2925                $full_url .= "/$self->{path}" if length $self->{path};
2926                remove_username($full_url);
2927                my $uuid = $self->svnsync->{uuid};
2928                $log_entry{metadata} = "$full_url\@$rev $uuid";
2929                $email ||= "$author\@$uuid";
2930                $commit_email ||= "$author\@$uuid";
2931        } else {
2932                my $url = $self->metadata_url;
2933                remove_username($url);
2934                $log_entry{metadata} = "$url\@$rev " .
2935                                       $self->ra->get_uuid;
2936                $email ||= "$author\@" . $self->ra->get_uuid;
2937                $commit_email ||= "$author\@" . $self->ra->get_uuid;
2938        }
2939        $log_entry{name} = $name;
2940        $log_entry{email} = $email;
2941        $log_entry{commit_name} = $commit_name;
2942        $log_entry{commit_email} = $commit_email;
2943        \%log_entry;
2944}
2945
2946sub fetch {
2947        my ($self, $min_rev, $max_rev, @parents) = @_;
2948        my ($last_rev, $last_commit) = $self->last_rev_commit;
2949        my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
2950        $self->ra->gs_fetch_loop_common($base, $head, [$self]);
2951}
2952
2953sub set_tree_cb {
2954        my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
2955        $self->{inject_parents} = { $rev => $tree };
2956        $self->fetch(undef, undef);
2957}
2958
2959sub set_tree {
2960        my ($self, $tree) = (shift, shift);
2961        my $log_entry = ::get_commit_entry($tree);
2962        unless ($self->{last_rev}) {
2963                ::fatal("Must have an existing revision to commit");
2964        }
2965        my %ed_opts = ( r => $self->{last_rev},
2966                        log => $log_entry->{log},
2967                        ra => $self->ra,
2968                        tree_a => $self->{last_commit},
2969                        tree_b => $tree,
2970                        editor_cb => sub {
2971                               $self->set_tree_cb($log_entry, $tree, @_) },
2972                        svn_path => $self->{path} );
2973        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
2974                print "No changes\nr$self->{last_rev} = $tree\n";
2975        }
2976}
2977
2978sub rebuild_from_rev_db {
2979        my ($self, $path) = @_;
2980        my $r = -1;
2981        open my $fh, '<', $path or croak "open: $!";
2982        binmode $fh or croak "binmode: $!";
2983        while (<$fh>) {
2984                length($_) == 41 or croak "inconsistent size in ($_) != 41";
2985                chomp($_);
2986                ++$r;
2987                next if $_ eq ('0' x 40);
2988                $self->rev_map_set($r, $_);
2989                print "r$r = $_\n";
2990        }
2991        close $fh or croak "close: $!";
2992        unlink $path or croak "unlink: $!";
2993}
2994
2995sub rebuild {
2996        my ($self) = @_;
2997        my $map_path = $self->map_path;
2998        my $partial = (-e $map_path && ! -z $map_path);
2999        return unless ::verify_ref($self->refname.'^0');
3000        if (!$partial && ($self->use_svm_props || $self->no_metadata)) {
3001                my $rev_db = $self->rev_db_path;
3002                $self->rebuild_from_rev_db($rev_db);
3003                if ($self->use_svm_props) {
3004                        my $svm_rev_db = $self->rev_db_path($self->svm_uuid);
3005                        $self->rebuild_from_rev_db($svm_rev_db);
3006                }
3007                $self->unlink_rev_db_symlink;
3008                return;
3009        }
3010        print "Rebuilding $map_path ...\n" if (!$partial);
3011        my ($base_rev, $head) = ($partial ? $self->rev_map_max_norebuild(1) :
3012                (undef, undef));
3013        my ($log, $ctx) =
3014            command_output_pipe(qw/rev-list --pretty=raw --no-color --reverse/,
3015                                ($head ? "$head.." : "") . $self->refname,
3016                                '--');
3017        my $metadata_url = $self->metadata_url;
3018        remove_username($metadata_url);
3019        my $svn_uuid = $self->ra_uuid;
3020        my $c;
3021        while (<$log>) {
3022                if ( m{^commit ($::sha1)$} ) {
3023                        $c = $1;
3024                        next;
3025                }
3026                next unless s{^\s*(git-svn-id:)}{$1};
3027                my ($url, $rev, $uuid) = ::extract_metadata($_);
3028                remove_username($url);
3029
3030                # ignore merges (from set-tree)
3031                next if (!defined $rev || !$uuid);
3032
3033                # if we merged or otherwise started elsewhere, this is
3034                # how we break out of it
3035                if (($uuid ne $svn_uuid) ||
3036                    ($metadata_url && $url && ($url ne $metadata_url))) {
3037                        next;
3038                }
3039                if ($partial && $head) {
3040                        print "Partial-rebuilding $map_path ...\n";
3041                        print "Currently at $base_rev = $head\n";
3042                        $head = undef;
3043                }
3044
3045                $self->rev_map_set($rev, $c);
3046                print "r$rev = $c\n";
3047        }
3048        command_close_pipe($log, $ctx);
3049        print "Done rebuilding $map_path\n" if (!$partial || !$head);
3050        my $rev_db_path = $self->rev_db_path;
3051        if (-f $self->rev_db_path) {
3052                unlink $self->rev_db_path or croak "unlink: $!";
3053        }
3054        $self->unlink_rev_db_symlink;
3055}
3056
3057# rev_map:
3058# Tie::File seems to be prone to offset errors if revisions get sparse,
3059# it's not that fast, either.  Tie::File is also not in Perl 5.6.  So
3060# one of my favorite modules is out :<  Next up would be one of the DBM
3061# modules, but I'm not sure which is most portable...
3062#
3063# This is the replacement for the rev_db format, which was too big
3064# and inefficient for large repositories with a lot of sparse history
3065# (mainly tags)
3066#
3067# The format is this:
3068#   - 24 bytes for every record,
3069#     * 4 bytes for the integer representing an SVN revision number
3070#     * 20 bytes representing the sha1 of a git commit
3071#   - No empty padding records like the old format
3072#     (except the last record, which can be overwritten)
3073#   - new records are written append-only since SVN revision numbers
3074#     increase monotonically
3075#   - lookups on SVN revision number are done via a binary search
3076#   - Piping the file to xxd -c24 is a good way of dumping it for
3077#     viewing or editing (piped back through xxd -r), should the need
3078#     ever arise.
3079#   - The last record can be padding revision with an all-zero sha1
3080#     This is used to optimize fetch performance when using multiple
3081#     "fetch" directives in .git/config
3082#
3083# These files are disposable unless noMetadata or useSvmProps is set
3084
3085sub _rev_map_set {
3086        my ($fh, $rev, $commit) = @_;
3087
3088        binmode $fh or croak "binmode: $!";
3089        my $size = (stat($fh))[7];
3090        ($size % 24) == 0 or croak "inconsistent size: $size";
3091
3092        my $wr_offset = 0;
3093        if ($size > 0) {
3094                sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3095                my $read = sysread($fh, my $buf, 24) or croak "read: $!";
3096                $read == 24 or croak "read only $read bytes (!= 24)";
3097                my ($last_rev, $last_commit) = unpack(rev_map_fmt, $buf);
3098                if ($last_commit eq ('0' x40)) {
3099                        if ($size >= 48) {
3100                                sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3101                                $read = sysread($fh, $buf, 24) or
3102                                    croak "read: $!";
3103                                $read == 24 or
3104                                    croak "read only $read bytes (!= 24)";
3105                                ($last_rev, $last_commit) =
3106                                    unpack(rev_map_fmt, $buf);
3107                                if ($last_commit eq ('0' x40)) {
3108                                        croak "inconsistent .rev_map\n";
3109                                }
3110                        }
3111                        if ($last_rev >= $rev) {
3112                                croak "last_rev is higher!: $last_rev >= $rev";
3113                        }
3114                        $wr_offset = -24;
3115                }
3116        }
3117        sysseek($fh, $wr_offset, SEEK_END) or croak "seek: $!";
3118        syswrite($fh, pack(rev_map_fmt, $rev, $commit), 24) == 24 or
3119          croak "write: $!";
3120}
3121
3122sub _rev_map_reset {
3123        my ($fh, $rev, $commit) = @_;
3124        my $c = _rev_map_get($fh, $rev);
3125        $c eq $commit or die "_rev_map_reset(@_) commit $c does not match!\n";
3126        my $offset = sysseek($fh, 0, SEEK_CUR) or croak "seek: $!";
3127        truncate $fh, $offset or croak "truncate: $!";
3128}
3129
3130sub mkfile {
3131        my ($path) = @_;
3132        unless (-e $path) {
3133                my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
3134                mkpath([$dir]) unless -d $dir;
3135                open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
3136                close $fh or die "Couldn't close (create) $path: $!\n";
3137        }
3138}
3139
3140sub rev_map_set {
3141        my ($self, $rev, $commit, $update_ref, $uuid) = @_;
3142        length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
3143        my $db = $self->map_path($uuid);
3144        my $db_lock = "$db.lock";
3145        my $sig;
3146        $update_ref ||= 0;
3147        if ($update_ref) {
3148                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3149                            $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
3150        }
3151        mkfile($db);
3152
3153        $LOCKFILES{$db_lock} = 1;
3154        my $sync;
3155        # both of these options make our .rev_db file very, very important
3156        # and we can't afford to lose it because rebuild() won't work
3157        if ($self->use_svm_props || $self->no_metadata) {
3158                $sync = 1;
3159                copy($db, $db_lock) or die "rev_map_set(@_): ",
3160                                           "Failed to copy: ",
3161                                           "$db => $db_lock ($!)\n";
3162        } else {
3163                rename $db, $db_lock or die "rev_map_set(@_): ",
3164                                            "Failed to rename: ",
3165                                            "$db => $db_lock ($!)\n";
3166        }
3167
3168        sysopen(my $fh, $db_lock, O_RDWR | O_CREAT)
3169             or croak "Couldn't open $db_lock: $!\n";
3170        $update_ref eq 'reset' ? _rev_map_reset($fh, $rev, $commit) :
3171                                 _rev_map_set($fh, $rev, $commit);
3172        if ($sync) {
3173                $fh->flush or die "Couldn't flush $db_lock: $!\n";
3174                $fh->sync or die "Couldn't sync $db_lock: $!\n";
3175        }
3176        close $fh or croak $!;
3177        if ($update_ref) {
3178                $_head = $self;
3179                my $note = "";
3180                $note = " ($update_ref)" if ($update_ref !~ /^\d*$/);
3181                command_noisy('update-ref', '-m', "r$rev$note",
3182                              $self->refname, $commit);
3183        }
3184        rename $db_lock, $db or die "rev_map_set(@_): ", "Failed to rename: ",
3185                                    "$db_lock => $db ($!)\n";
3186        delete $LOCKFILES{$db_lock};
3187        if ($update_ref) {
3188                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3189                            $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
3190                kill $sig, $$ if defined $sig;
3191        }
3192}
3193
3194# If want_commit, this will return an array of (rev, commit) where
3195# commit _must_ be a valid commit in the archive.
3196# Otherwise, it'll return the max revision (whether or not the
3197# commit is valid or just a 0x40 placeholder).
3198sub rev_map_max {
3199        my ($self, $want_commit) = @_;
3200        $self->rebuild;
3201        my ($r, $c) = $self->rev_map_max_norebuild($want_commit);
3202        $want_commit ? ($r, $c) : $r;
3203}
3204
3205sub rev_map_max_norebuild {
3206        my ($self, $want_commit) = @_;
3207        my $map_path = $self->map_path;
3208        stat $map_path or return $want_commit ? (0, undef) : 0;
3209        sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3210        binmode $fh or croak "binmode: $!";
3211        my $size = (stat($fh))[7];
3212        ($size % 24) == 0 or croak "inconsistent size: $size";
3213
3214        if ($size == 0) {
3215                close $fh or croak "close: $!";
3216                return $want_commit ? (0, undef) : 0;
3217        }
3218
3219        sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3220        sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3221        my ($r, $c) = unpack(rev_map_fmt, $buf);
3222        if ($want_commit && $c eq ('0' x40)) {
3223                if ($size < 48) {
3224                        return $want_commit ? (0, undef) : 0;
3225                }
3226                sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3227                sysread($fh, $buf, 24) == 24 or croak "read: $!";
3228                ($r, $c) = unpack(rev_map_fmt, $buf);
3229                if ($c eq ('0'x40)) {
3230                        croak "Penultimate record is all-zeroes in $map_path";
3231                }
3232        }
3233        close $fh or croak "close: $!";
3234        $want_commit ? ($r, $c) : $r;
3235}
3236
3237sub rev_map_get {
3238        my ($self, $rev, $uuid) = @_;
3239        my $map_path = $self->map_path($uuid);
3240        return undef unless -e $map_path;
3241
3242        sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3243        my $c = _rev_map_get($fh, $rev);
3244        close($fh) or croak "close: $!";
3245        $c
3246}
3247
3248sub _rev_map_get {
3249        my ($fh, $rev) = @_;
3250
3251        binmode $fh or croak "binmode: $!";
3252        my $size = (stat($fh))[7];
3253        ($size % 24) == 0 or croak "inconsistent size: $size";
3254
3255        if ($size == 0) {
3256                return undef;
3257        }
3258
3259        my ($l, $u) = (0, $size - 24);
3260        my ($r, $c, $buf);
3261
3262        while ($l <= $u) {
3263                my $i = int(($l/24 + $u/24) / 2) * 24;
3264                sysseek($fh, $i, SEEK_SET) or croak "seek: $!";
3265                sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3266                my ($r, $c) = unpack('NH40', $buf);
3267
3268                if ($r < $rev) {
3269                        $l = $i + 24;
3270                } elsif ($r > $rev) {
3271                        $u = $i - 24;
3272                } else { # $r == $rev
3273                        return $c eq ('0' x 40) ? undef : $c;
3274                }
3275        }
3276        undef;
3277}
3278
3279# Finds the first svn revision that exists on (if $eq_ok is true) or
3280# before $rev for the current branch.  It will not search any lower
3281# than $min_rev.  Returns the git commit hash and svn revision number
3282# if found, else (undef, undef).
3283sub find_rev_before {
3284        my ($self, $rev, $eq_ok, $min_rev) = @_;
3285        --$rev unless $eq_ok;
3286        $min_rev ||= 1;
3287        my $max_rev = $self->rev_map_max;
3288        $rev = $max_rev if ($rev > $max_rev);
3289        while ($rev >= $min_rev) {
3290                if (my $c = $self->rev_map_get($rev)) {
3291                        return ($rev, $c);
3292                }
3293                --$rev;
3294        }
3295        return (undef, undef);
3296}
3297
3298# Finds the first svn revision that exists on (if $eq_ok is true) or
3299# after $rev for the current branch.  It will not search any higher
3300# than $max_rev.  Returns the git commit hash and svn revision number
3301# if found, else (undef, undef).
3302sub find_rev_after {
3303        my ($self, $rev, $eq_ok, $max_rev) = @_;
3304        ++$rev unless $eq_ok;
3305        $max_rev ||= $self->rev_map_max;
3306        while ($rev <= $max_rev) {
3307                if (my $c = $self->rev_map_get($rev)) {
3308                        return ($rev, $c);
3309                }
3310                ++$rev;
3311        }
3312        return (undef, undef);
3313}
3314
3315sub _new {
3316        my ($class, $repo_id, $ref_id, $path) = @_;
3317        unless (defined $repo_id && length $repo_id) {
3318                $repo_id = $Git::SVN::default_repo_id;
3319        }
3320        unless (defined $ref_id && length $ref_id) {
3321                $_[2] = $ref_id = $Git::SVN::default_ref_id;
3322        }
3323        $_[1] = $repo_id;
3324        my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
3325        $_[3] = $path = '' unless (defined $path);
3326        mkpath(["$ENV{GIT_DIR}/svn"]);
3327        bless {
3328                ref_id => $ref_id, dir => $dir, index => "$dir/index",
3329                path => $path, config => "$ENV{GIT_DIR}/svn/config",
3330                map_root => "$dir/.rev_map", repo_id => $repo_id }, $class;
3331}
3332
3333# for read-only access of old .rev_db formats
3334sub unlink_rev_db_symlink {
3335        my ($self) = @_;
3336        my $link = $self->rev_db_path;
3337        $link =~ s/\.[\w-]+$// or croak "missing UUID at the end of $link";
3338        if (-l $link) {
3339                unlink $link or croak "unlink: $link failed!";
3340        }
3341}
3342
3343sub rev_db_path {
3344        my ($self, $uuid) = @_;
3345        my $db_path = $self->map_path($uuid);
3346        $db_path =~ s{/\.rev_map\.}{/\.rev_db\.}
3347            or croak "map_path: $db_path does not contain '/.rev_map.' !";
3348        $db_path;
3349}
3350
3351# the new replacement for .rev_db
3352sub map_path {
3353        my ($self, $uuid) = @_;
3354        $uuid ||= $self->ra_uuid;
3355        "$self->{map_root}.$uuid";
3356}
3357
3358sub uri_encode {
3359        my ($f) = @_;
3360        $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
3361        $f
3362}
3363
3364sub remove_username {
3365        $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
3366}
3367
3368package Git::SVN::Prompt;
3369use strict;
3370use warnings;
3371require SVN::Core;
3372use vars qw/$_no_auth_cache $_username/;
3373
3374sub simple {
3375        my ($cred, $realm, $default_username, $may_save, $pool) = @_;
3376        $may_save = undef if $_no_auth_cache;
3377        $default_username = $_username if defined $_username;
3378        if (defined $default_username && length $default_username) {
3379                if (defined $realm && length $realm) {
3380                        print STDERR "Authentication realm: $realm\n";
3381                        STDERR->flush;
3382                }
3383                $cred->username($default_username);
3384        } else {
3385                username($cred, $realm, $may_save, $pool);
3386        }
3387        $cred->password(_read_password("Password for '" .
3388                                       $cred->username . "': ", $realm));
3389        $cred->may_save($may_save);
3390        $SVN::_Core::SVN_NO_ERROR;
3391}
3392
3393sub ssl_server_trust {
3394        my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
3395        $may_save = undef if $_no_auth_cache;
3396        print STDERR "Error validating server certificate for '$realm':\n";
3397        {
3398                no warnings 'once';
3399                # All variables SVN::Auth::SSL::* are used only once,
3400                # so we're shutting up Perl warnings about this.
3401                if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
3402                        print STDERR " - The certificate is not issued ",
3403                            "by a trusted authority. Use the\n",
3404                            "   fingerprint to validate ",
3405                            "the certificate manually!\n";
3406                }
3407                if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
3408                        print STDERR " - The certificate hostname ",
3409                            "does not match.\n";
3410                }
3411                if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
3412                        print STDERR " - The certificate is not yet valid.\n";
3413                }
3414                if ($failures & $SVN::Auth::SSL::EXPIRED) {
3415                        print STDERR " - The certificate has expired.\n";
3416                }
3417                if ($failures & $SVN::Auth::SSL::OTHER) {
3418                        print STDERR " - The certificate has ",
3419                            "an unknown error.\n";
3420                }
3421        } # no warnings 'once'
3422        printf STDERR
3423                "Certificate information:\n".
3424                " - Hostname: %s\n".
3425                " - Valid: from %s until %s\n".
3426                " - Issuer: %s\n".
3427                " - Fingerprint: %s\n",
3428                map $cert_info->$_, qw(hostname valid_from valid_until
3429                                       issuer_dname fingerprint);
3430        my $choice;
3431prompt:
3432        print STDERR $may_save ?
3433              "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
3434              "(R)eject or accept (t)emporarily? ";
3435        STDERR->flush;
3436        $choice = lc(substr(<STDIN> || 'R', 0, 1));
3437        if ($choice =~ /^t$/i) {
3438                $cred->may_save(undef);
3439        } elsif ($choice =~ /^r$/i) {
3440                return -1;
3441        } elsif ($may_save && $choice =~ /^p$/i) {
3442                $cred->may_save($may_save);
3443        } else {
3444                goto prompt;
3445        }
3446        $cred->accepted_failures($failures);
3447        $SVN::_Core::SVN_NO_ERROR;
3448}
3449
3450sub ssl_client_cert {
3451        my ($cred, $realm, $may_save, $pool) = @_;
3452        $may_save = undef if $_no_auth_cache;
3453        print STDERR "Client certificate filename: ";
3454        STDERR->flush;
3455        chomp(my $filename = <STDIN>);
3456        $cred->cert_file($filename);
3457        $cred->may_save($may_save);
3458        $SVN::_Core::SVN_NO_ERROR;
3459}
3460
3461sub ssl_client_cert_pw {
3462        my ($cred, $realm, $may_save, $pool) = @_;
3463        $may_save = undef if $_no_auth_cache;
3464        $cred->password(_read_password("Password: ", $realm));
3465        $cred->may_save($may_save);
3466        $SVN::_Core::SVN_NO_ERROR;
3467}
3468
3469sub username {
3470        my ($cred, $realm, $may_save, $pool) = @_;
3471        $may_save = undef if $_no_auth_cache;
3472        if (defined $realm && length $realm) {
3473                print STDERR "Authentication realm: $realm\n";
3474        }
3475        my $username;
3476        if (defined $_username) {
3477                $username = $_username;
3478        } else {
3479                print STDERR "Username: ";
3480                STDERR->flush;
3481                chomp($username = <STDIN>);
3482        }
3483        $cred->username($username);
3484        $cred->may_save($may_save);
3485        $SVN::_Core::SVN_NO_ERROR;
3486}
3487
3488sub _read_password {
3489        my ($prompt, $realm) = @_;
3490        print STDERR $prompt;
3491        STDERR->flush;
3492        require Term::ReadKey;
3493        Term::ReadKey::ReadMode('noecho');
3494        my $password = '';
3495        while (defined(my $key = Term::ReadKey::ReadKey(0))) {
3496                last if $key =~ /[\012\015]/; # \n\r
3497                $password .= $key;
3498        }
3499        Term::ReadKey::ReadMode('restore');
3500        print STDERR "\n";
3501        STDERR->flush;
3502        $password;
3503}
3504
3505package SVN::Git::Fetcher;
3506use vars qw/@ISA/;
3507use strict;
3508use warnings;
3509use Carp qw/croak/;
3510use File::Temp qw/tempfile/;
3511use IO::File qw//;
3512use vars qw/$_ignore_regex/;
3513
3514# file baton members: path, mode_a, mode_b, pool, fh, blob, base
3515sub new {
3516        my ($class, $git_svn, $switch_path) = @_;
3517        my $self = SVN::Delta::Editor->new;
3518        bless $self, $class;
3519        if (exists $git_svn->{last_commit}) {
3520                $self->{c} = $git_svn->{last_commit};
3521                $self->{empty_symlinks} =
3522                                  _mark_empty_symlinks($git_svn, $switch_path);
3523        }
3524        $self->{ignore_regex} = eval { command_oneline('config', '--get',
3525                             "svn-remote.$git_svn->{repo_id}.ignore-paths") };
3526        $self->{empty} = {};
3527        $self->{dir_prop} = {};
3528        $self->{file_prop} = {};
3529        $self->{absent_dir} = {};
3530        $self->{absent_file} = {};
3531        $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
3532        $self;
3533}
3534
3535# this uses the Ra object, so it must be called before do_{switch,update},
3536# not inside them (when the Git::SVN::Fetcher object is passed) to
3537# do_{switch,update}
3538sub _mark_empty_symlinks {
3539        my ($git_svn, $switch_path) = @_;
3540        my $bool = Git::config_bool('svn.brokenSymlinkWorkaround');
3541        return {} if (!defined($bool)) || (defined($bool) && ! $bool);
3542
3543        my %ret;
3544        my ($rev, $cmt) = $git_svn->last_rev_commit;
3545        return {} unless ($rev && $cmt);
3546
3547        # allow the warning to be printed for each revision we fetch to
3548        # ensure the user sees it.  The user can also disable the workaround
3549        # on the repository even while git svn is running and the next
3550        # revision fetched will skip this expensive function.
3551        my $printed_warning;
3552        chomp(my $empty_blob = `git hash-object -t blob --stdin < /dev/null`);
3553        my ($ls, $ctx) = command_output_pipe(qw/ls-tree -r -z/, $cmt);
3554        local $/ = "\0";
3555        my $pfx = defined($switch_path) ? $switch_path : $git_svn->{path};
3556        $pfx .= '/' if length($pfx);
3557        while (<$ls>) {
3558                chomp;
3559                s/\A100644 blob $empty_blob\t//o or next;
3560                unless ($printed_warning) {
3561                        print STDERR "Scanning for empty symlinks, ",
3562                                     "this may take a while if you have ",
3563                                     "many empty files\n",
3564                                     "You may disable this with `",
3565                                     "git config svn.brokenSymlinkWorkaround ",
3566                                     "false'.\n",
3567                                     "This may be done in a different ",
3568                                     "terminal without restarting ",
3569                                     "git svn\n";
3570                        $printed_warning = 1;
3571                }
3572                my $path = $_;
3573                my (undef, $props) =
3574                               $git_svn->ra->get_file($pfx.$path, $rev, undef);
3575                if ($props->{'svn:special'}) {
3576                        $ret{$path} = 1;
3577                }
3578        }
3579        command_close_pipe($ls, $ctx);
3580        \%ret;
3581}
3582
3583# returns true if a given path is inside a ".git" directory
3584sub in_dot_git {
3585        $_[0] =~ m{(?:^|/)\.git(?:/|$)};
3586}
3587
3588# return value: 0 -- don't ignore, 1 -- ignore
3589sub is_path_ignored {
3590        my ($self, $path) = @_;
3591        return 1 if in_dot_git($path);
3592        return 1 if defined($self->{ignore_regex}) &&
3593                    $path =~ m!$self->{ignore_regex}!;
3594        return 0 unless defined($_ignore_regex);
3595        return 1 if $path =~ m!$_ignore_regex!o;
3596        return 0;
3597}
3598
3599sub set_path_strip {
3600        my ($self, $path) = @_;
3601        $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
3602}
3603
3604sub open_root {
3605        { path => '' };
3606}
3607
3608sub open_directory {
3609        my ($self, $path, $pb, $rev) = @_;
3610        { path => $path };
3611}
3612
3613sub git_path {
3614        my ($self, $path) = @_;
3615        if ($self->{path_strip}) {
3616                $path =~ s!$self->{path_strip}!! or
3617                  die "Failed to strip path '$path' ($self->{path_strip})\n";
3618        }
3619        $path;
3620}
3621
3622sub delete_entry {
3623        my ($self, $path, $rev, $pb) = @_;
3624        return undef if $self->is_path_ignored($path);
3625
3626        my $gpath = $self->git_path($path);
3627        return undef if ($gpath eq '');
3628
3629        # remove entire directories.
3630        my ($tree) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
3631                         =~ /\A040000 tree ([a-f\d]{40})\t\Q$gpath\E\0/);
3632        if ($tree) {
3633                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3634                                                     -r --name-only -z/,
3635                                                     $tree);
3636                local $/ = "\0";
3637                while (<$ls>) {
3638                        chomp;
3639                        my $rmpath = "$gpath/$_";
3640                        $self->{gii}->remove($rmpath);
3641                        print "\tD\t$rmpath\n" unless $::_q;
3642                }
3643                print "\tD\t$gpath/\n" unless $::_q;
3644                command_close_pipe($ls, $ctx);
3645                $self->{empty}->{$path} = 0
3646        } else {
3647                $self->{gii}->remove($gpath);
3648                print "\tD\t$gpath\n" unless $::_q;
3649        }
3650        undef;
3651}
3652
3653sub open_file {
3654        my ($self, $path, $pb, $rev) = @_;
3655        my ($mode, $blob);
3656
3657        goto out if $self->is_path_ignored($path);
3658
3659        my $gpath = $self->git_path($path);
3660        ($mode, $blob) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
3661                             =~ /\A(\d{6}) blob ([a-f\d]{40})\t\Q$gpath\E\0/);
3662        unless (defined $mode && defined $blob) {
3663                die "$path was not found in commit $self->{c} (r$rev)\n";
3664        }
3665        if ($mode eq '100644' && $self->{empty_symlinks}->{$path}) {
3666                $mode = '120000';
3667        }
3668out:
3669        { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
3670          pool => SVN::Pool->new, action => 'M' };
3671}
3672
3673sub add_file {
3674        my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
3675        my $mode;
3676
3677        if (!$self->is_path_ignored($path)) {
3678                my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3679                delete $self->{empty}->{$dir};
3680                $mode = '100644';
3681        }
3682        { path => $path, mode_a => $mode, mode_b => $mode,
3683          pool => SVN::Pool->new, action => 'A' };
3684}
3685
3686sub add_directory {
3687        my ($self, $path, $cp_path, $cp_rev) = @_;
3688        goto out if $self->is_path_ignored($path);
3689        my $gpath = $self->git_path($path);
3690        if ($gpath eq '') {
3691                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3692                                                     -r --name-only -z/,
3693                                                     $self->{c});
3694                local $/ = "\0";
3695                while (<$ls>) {
3696                        chomp;
3697                        $self->{gii}->remove($_);
3698                        print "\tD\t$_\n" unless $::_q;
3699                }
3700                command_close_pipe($ls, $ctx);
3701                $self->{empty}->{$path} = 0;
3702        }
3703        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3704        delete $self->{empty}->{$dir};
3705        $self->{empty}->{$path} = 1;
3706out:
3707        { path => $path };
3708}
3709
3710sub change_dir_prop {
3711        my ($self, $db, $prop, $value) = @_;
3712        return undef if $self->is_path_ignored($db->{path});
3713        $self->{dir_prop}->{$db->{path}} ||= {};
3714        $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
3715        undef;
3716}
3717
3718sub absent_directory {
3719        my ($self, $path, $pb) = @_;
3720        return undef if $self->is_path_ignored($path);
3721        $self->{absent_dir}->{$pb->{path}} ||= [];
3722        push @{$self->{absent_dir}->{$pb->{path}}}, $path;
3723        undef;
3724}
3725
3726sub absent_file {
3727        my ($self, $path, $pb) = @_;
3728        return undef if $self->is_path_ignored($path);
3729        $self->{absent_file}->{$pb->{path}} ||= [];
3730        push @{$self->{absent_file}->{$pb->{path}}}, $path;
3731        undef;
3732}
3733
3734sub change_file_prop {
3735        my ($self, $fb, $prop, $value) = @_;
3736        return undef if $self->is_path_ignored($fb->{path});
3737        if ($prop eq 'svn:executable') {
3738                if ($fb->{mode_b} != 120000) {
3739                        $fb->{mode_b} = defined $value ? 100755 : 100644;
3740                }
3741        } elsif ($prop eq 'svn:special') {
3742                $fb->{mode_b} = defined $value ? 120000 : 100644;
3743        } else {
3744                $self->{file_prop}->{$fb->{path}} ||= {};
3745                $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
3746        }
3747        undef;
3748}
3749
3750sub apply_textdelta {
3751        my ($self, $fb, $exp) = @_;
3752        return undef if $self->is_path_ignored($fb->{path});
3753        my $fh = $::_repository->temp_acquire('svn_delta');
3754        # $fh gets auto-closed() by SVN::TxDelta::apply(),
3755        # (but $base does not,) so dup() it for reading in close_file
3756        open my $dup, '<&', $fh or croak $!;
3757        my $base = $::_repository->temp_acquire('git_blob');
3758
3759        if ($fb->{blob}) {
3760                my ($base_is_link, $size);
3761
3762                if ($fb->{mode_a} eq '120000' &&
3763                    ! $self->{empty_symlinks}->{$fb->{path}}) {
3764                        print $base 'link ' or die "print $!\n";
3765                        $base_is_link = 1;
3766                }
3767        retry:
3768                $size = $::_repository->cat_blob($fb->{blob}, $base);
3769                die "Failed to read object $fb->{blob}" if ($size < 0);
3770
3771                if (defined $exp) {
3772                        seek $base, 0, 0 or croak $!;
3773                        my $got = ::md5sum($base);
3774                        if ($got ne $exp) {
3775                                my $err = "Checksum mismatch: ".
3776                                       "$fb->{path} $fb->{blob}\n" .
3777                                       "expected: $exp\n" .
3778                                       "     got: $got\n";
3779                                if ($base_is_link) {
3780                                        warn $err,
3781                                             "Retrying... (possibly ",
3782                                             "a bad symlink from SVN)\n";
3783                                        $::_repository->temp_reset($base);
3784                                        $base_is_link = 0;
3785                                        goto retry;
3786                                }
3787                                die $err;
3788                        }
3789                }
3790        }
3791        seek $base, 0, 0 or croak $!;
3792        $fb->{fh} = $fh;
3793        $fb->{base} = $base;
3794        [ SVN::TxDelta::apply($base, $dup, undef, $fb->{path}, $fb->{pool}) ];
3795}
3796
3797sub close_file {
3798        my ($self, $fb, $exp) = @_;
3799        return undef if $self->is_path_ignored($fb->{path});
3800
3801        my $hash;
3802        my $path = $self->git_path($fb->{path});
3803        if (my $fh = $fb->{fh}) {
3804                if (defined $exp) {
3805                        seek($fh, 0, 0) or croak $!;
3806                        my $got = ::md5sum($fh);
3807                        if ($got ne $exp) {
3808                                die "Checksum mismatch: $path\n",
3809                                    "expected: $exp\n    got: $got\n";
3810                        }
3811                }
3812                if ($fb->{mode_b} == 120000) {
3813                        sysseek($fh, 0, 0) or croak $!;
3814                        my $rd = sysread($fh, my $buf, 5);
3815
3816                        if (!defined $rd) {
3817                                croak "sysread: $!\n";
3818                        } elsif ($rd == 0) {
3819                                warn "$path has mode 120000",
3820                                     " but it points to nothing\n",
3821                                     "converting to an empty file with mode",
3822                                     " 100644\n";
3823                                $fb->{mode_b} = '100644';
3824                        } elsif ($buf ne 'link ') {
3825                                warn "$path has mode 120000",
3826                                     " but is not a link\n";
3827                        } else {
3828                                my $tmp_fh = $::_repository->temp_acquire(
3829                                        'svn_hash');
3830                                my $res;
3831                                while ($res = sysread($fh, my $str, 1024)) {
3832                                        my $out = syswrite($tmp_fh, $str, $res);
3833                                        defined($out) && $out == $res
3834                                                or croak("write ",
3835                                                        Git::temp_path($tmp_fh),
3836                                                        ": $!\n");
3837                                }
3838                                defined $res or croak $!;
3839
3840                                ($fh, $tmp_fh) = ($tmp_fh, $fh);
3841                                Git::temp_release($tmp_fh, 1);
3842                        }
3843                }
3844
3845                $hash = $::_repository->hash_and_insert_object(
3846                                Git::temp_path($fh));
3847                $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
3848
3849                Git::temp_release($fb->{base}, 1);
3850                Git::temp_release($fh, 1);
3851        } else {
3852                $hash = $fb->{blob} or die "no blob information\n";
3853        }
3854        $fb->{pool}->clear;
3855        $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
3856        print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
3857        undef;
3858}
3859
3860sub abort_edit {
3861        my $self = shift;
3862        $self->{nr} = $self->{gii}->{nr};
3863        delete $self->{gii};
3864        $self->SUPER::abort_edit(@_);
3865}
3866
3867sub close_edit {
3868        my $self = shift;
3869        $self->{git_commit_ok} = 1;
3870        $self->{nr} = $self->{gii}->{nr};
3871        delete $self->{gii};
3872        $self->SUPER::close_edit(@_);
3873}
3874
3875package SVN::Git::Editor;
3876use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
3877use strict;
3878use warnings;
3879use Carp qw/croak/;
3880use IO::File;
3881
3882sub new {
3883        my ($class, $opts) = @_;
3884        foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
3885                die "$_ required!\n" unless (defined $opts->{$_});
3886        }
3887
3888        my $pool = SVN::Pool->new;
3889        my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
3890        my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
3891                                     $opts->{r}, $mods);
3892
3893        # $opts->{ra} functions should not be used after this:
3894        my @ce  = $opts->{ra}->get_commit_editor($opts->{log},
3895                                                $opts->{editor_cb}, $pool);
3896        my $self = SVN::Delta::Editor->new(@ce, $pool);
3897        bless $self, $class;
3898        foreach (qw/svn_path r tree_a tree_b/) {
3899                $self->{$_} = $opts->{$_};
3900        }
3901        $self->{url} = $opts->{ra}->{url};
3902        $self->{mods} = $mods;
3903        $self->{types} = $types;
3904        $self->{pool} = $pool;
3905        $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
3906        $self->{rm} = { };
3907        $self->{path_prefix} = length $self->{svn_path} ?
3908                               "$self->{svn_path}/" : '';
3909        $self->{config} = $opts->{config};
3910        return $self;
3911}
3912
3913sub generate_diff {
3914        my ($tree_a, $tree_b) = @_;
3915        my @diff_tree = qw(diff-tree -z -r);
3916        if ($_cp_similarity) {
3917                push @diff_tree, "-C$_cp_similarity";
3918        } else {
3919                push @diff_tree, '-C';
3920        }
3921        push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
3922        push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
3923        push @diff_tree, $tree_a, $tree_b;
3924        my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
3925        local $/ = "\0";
3926        my $state = 'meta';
3927        my @mods;
3928        while (<$diff_fh>) {
3929                chomp $_; # this gets rid of the trailing "\0"
3930                if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
3931                                        ($::sha1)\s($::sha1)\s
3932                                        ([MTCRAD])\d*$/xo) {
3933                        push @mods, {   mode_a => $1, mode_b => $2,
3934                                        sha1_a => $3, sha1_b => $4,
3935                                        chg => $5 };
3936                        if ($5 =~ /^(?:C|R)$/) {
3937                                $state = 'file_a';
3938                        } else {
3939                                $state = 'file_b';
3940                        }
3941                } elsif ($state eq 'file_a') {
3942                        my $x = $mods[$#mods] or croak "Empty array\n";
3943                        if ($x->{chg} !~ /^(?:C|R)$/) {
3944                                croak "Error parsing $_, $x->{chg}\n";
3945                        }
3946                        $x->{file_a} = $_;
3947                        $state = 'file_b';
3948                } elsif ($state eq 'file_b') {
3949                        my $x = $mods[$#mods] or croak "Empty array\n";
3950                        if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
3951                                croak "Error parsing $_, $x->{chg}\n";
3952                        }
3953                        if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
3954                                croak "Error parsing $_, $x->{chg}\n";
3955                        }
3956                        $x->{file_b} = $_;
3957                        $state = 'meta';
3958                } else {
3959                        croak "Error parsing $_\n";
3960                }
3961        }
3962        command_close_pipe($diff_fh, $ctx);
3963        \@mods;
3964}
3965
3966sub check_diff_paths {
3967        my ($ra, $pfx, $rev, $mods) = @_;
3968        my %types;
3969        $pfx .= '/' if length $pfx;
3970
3971        sub type_diff_paths {
3972                my ($ra, $types, $path, $rev) = @_;
3973                my @p = split m#/+#, $path;
3974                my $c = shift @p;
3975                unless (defined $types->{$c}) {
3976                        $types->{$c} = $ra->check_path($c, $rev);
3977                }
3978                while (@p) {
3979                        $c .= '/' . shift @p;
3980                        next if defined $types->{$c};
3981                        $types->{$c} = $ra->check_path($c, $rev);
3982                }
3983        }
3984
3985        foreach my $m (@$mods) {
3986                foreach my $f (qw/file_a file_b/) {
3987                        next unless defined $m->{$f};
3988                        my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
3989                        if (length $pfx.$dir && ! defined $types{$dir}) {
3990                                type_diff_paths($ra, \%types, $pfx.$dir, $rev);
3991                        }
3992                }
3993        }
3994        \%types;
3995}
3996
3997sub split_path {
3998        return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
3999}
4000
4001sub repo_path {
4002        my ($self, $path) = @_;
4003        $self->{path_prefix}.(defined $path ? $path : '');
4004}
4005
4006sub url_path {
4007        my ($self, $path) = @_;
4008        if ($self->{url} =~ m#^https?://#) {
4009                $path =~ s!([^~a-zA-Z0-9_./-])!uc sprintf("%%%02x",ord($1))!eg;
4010        }
4011        $self->{url} . '/' . $self->repo_path($path);
4012}
4013
4014sub rmdirs {
4015        my ($self) = @_;
4016        my $rm = $self->{rm};
4017        delete $rm->{''}; # we never delete the url we're tracking
4018        return unless %$rm;
4019
4020        foreach (keys %$rm) {
4021                my @d = split m#/#, $_;
4022                my $c = shift @d;
4023                $rm->{$c} = 1;
4024                while (@d) {
4025                        $c .= '/' . shift @d;
4026                        $rm->{$c} = 1;
4027                }
4028        }
4029        delete $rm->{$self->{svn_path}};
4030        delete $rm->{''}; # we never delete the url we're tracking
4031        return unless %$rm;
4032
4033        my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
4034                                             $self->{tree_b});
4035        local $/ = "\0";
4036        while (<$fh>) {
4037                chomp;
4038                my @dn = split m#/#, $_;
4039                while (pop @dn) {
4040                        delete $rm->{join '/', @dn};
4041                }
4042                unless (%$rm) {
4043                        close $fh;
4044                        return;
4045                }
4046        }
4047        command_close_pipe($fh, $ctx);
4048
4049        my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
4050        foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
4051                $self->close_directory($bat->{$d}, $p);
4052                my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
4053                print "\tD+\t$d/\n" unless $::_q;
4054                $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
4055                delete $bat->{$d};
4056        }
4057}
4058
4059sub open_or_add_dir {
4060        my ($self, $full_path, $baton) = @_;
4061        my $t = $self->{types}->{$full_path};
4062        if (!defined $t) {
4063                die "$full_path not known in r$self->{r} or we have a bug!\n";
4064        }
4065        {
4066                no warnings 'once';
4067                # SVN::Node::none and SVN::Node::file are used only once,
4068                # so we're shutting up Perl's warnings about them.
4069                if ($t == $SVN::Node::none) {
4070                        return $self->add_directory($full_path, $baton,
4071                            undef, -1, $self->{pool});
4072                } elsif ($t == $SVN::Node::dir) {
4073                        return $self->open_directory($full_path, $baton,
4074                            $self->{r}, $self->{pool});
4075                } # no warnings 'once'
4076                print STDERR "$full_path already exists in repository at ",
4077                    "r$self->{r} and it is not a directory (",
4078                    ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
4079        } # no warnings 'once'
4080        exit 1;
4081}
4082
4083sub ensure_path {
4084        my ($self, $path) = @_;
4085        my $bat = $self->{bat};
4086        my $repo_path = $self->repo_path($path);
4087        return $bat->{''} unless (length $repo_path);
4088        my @p = split m#/+#, $repo_path;
4089        my $c = shift @p;
4090        $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
4091        while (@p) {
4092                my $c0 = $c;
4093                $c .= '/' . shift @p;
4094                $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
4095        }
4096        return $bat->{$c};
4097}
4098
4099# Subroutine to convert a globbing pattern to a regular expression.
4100# From perl cookbook.
4101sub glob2pat {
4102        my $globstr = shift;
4103        my %patmap = ('*' => '.*', '?' => '.', '[' => '[', ']' => ']');
4104        $globstr =~ s{(.)} { $patmap{$1} || "\Q$1" }ge;
4105        return '^' . $globstr . '$';
4106}
4107
4108sub check_autoprop {
4109        my ($self, $pattern, $properties, $file, $fbat) = @_;
4110        # Convert the globbing pattern to a regular expression.
4111        my $regex = glob2pat($pattern);
4112        # Check if the pattern matches the file name.
4113        if($file =~ m/($regex)/) {
4114                # Parse the list of properties to set.
4115                my @props = split(/;/, $properties);
4116                foreach my $prop (@props) {
4117                        # Parse 'name=value' syntax and set the property.
4118                        if ($prop =~ /([^=]+)=(.*)/) {
4119                                my ($n,$v) = ($1,$2);
4120                                for ($n, $v) {
4121                                        s/^\s+//; s/\s+$//;
4122                                }
4123                                $self->change_file_prop($fbat, $n, $v);
4124                        }
4125                }
4126        }
4127}
4128
4129sub apply_autoprops {
4130        my ($self, $file, $fbat) = @_;
4131        my $conf_t = ${$self->{config}}{'config'};
4132        no warnings 'once';
4133        # Check [miscellany]/enable-auto-props in svn configuration.
4134        if (SVN::_Core::svn_config_get_bool(
4135                $conf_t,
4136                $SVN::_Core::SVN_CONFIG_SECTION_MISCELLANY,
4137                $SVN::_Core::SVN_CONFIG_OPTION_ENABLE_AUTO_PROPS,
4138                0)) {
4139                # Auto-props are enabled.  Enumerate them to look for matches.
4140                my $callback = sub {
4141                        $self->check_autoprop($_[0], $_[1], $file, $fbat);
4142                };
4143                SVN::_Core::svn_config_enumerate(
4144                        $conf_t,
4145                        $SVN::_Core::SVN_CONFIG_SECTION_AUTO_PROPS,
4146                        $callback);
4147        }
4148}
4149
4150sub A {
4151        my ($self, $m) = @_;
4152        my ($dir, $file) = split_path($m->{file_b});
4153        my $pbat = $self->ensure_path($dir);
4154        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4155                                        undef, -1);
4156        print "\tA\t$m->{file_b}\n" unless $::_q;
4157        $self->apply_autoprops($file, $fbat);
4158        $self->chg_file($fbat, $m);
4159        $self->close_file($fbat,undef,$self->{pool});
4160}
4161
4162sub C {
4163        my ($self, $m) = @_;
4164        my ($dir, $file) = split_path($m->{file_b});
4165        my $pbat = $self->ensure_path($dir);
4166        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4167                                $self->url_path($m->{file_a}), $self->{r});
4168        print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
4169        $self->chg_file($fbat, $m);
4170        $self->close_file($fbat,undef,$self->{pool});
4171}
4172
4173sub delete_entry {
4174        my ($self, $path, $pbat) = @_;
4175        my $rpath = $self->repo_path($path);
4176        my ($dir, $file) = split_path($rpath);
4177        $self->{rm}->{$dir} = 1;
4178        $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
4179}
4180
4181sub R {
4182        my ($self, $m) = @_;
4183        my ($dir, $file) = split_path($m->{file_b});
4184        my $pbat = $self->ensure_path($dir);
4185        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4186                                $self->url_path($m->{file_a}), $self->{r});
4187        print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
4188        $self->apply_autoprops($file, $fbat);
4189        $self->chg_file($fbat, $m);
4190        $self->close_file($fbat,undef,$self->{pool});
4191
4192        ($dir, $file) = split_path($m->{file_a});
4193        $pbat = $self->ensure_path($dir);
4194        $self->delete_entry($m->{file_a}, $pbat);
4195}
4196
4197sub M {
4198        my ($self, $m) = @_;
4199        my ($dir, $file) = split_path($m->{file_b});
4200        my $pbat = $self->ensure_path($dir);
4201        my $fbat = $self->open_file($self->repo_path($m->{file_b}),
4202                                $pbat,$self->{r},$self->{pool});
4203        print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
4204        $self->chg_file($fbat, $m);
4205        $self->close_file($fbat,undef,$self->{pool});
4206}
4207
4208sub T { shift->M(@_) }
4209
4210sub change_file_prop {
4211        my ($self, $fbat, $pname, $pval) = @_;
4212        $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
4213}
4214
4215sub _chg_file_get_blob ($$$$) {
4216        my ($self, $fbat, $m, $which) = @_;
4217        my $fh = $::_repository->temp_acquire("git_blob_$which");
4218        if ($m->{"mode_$which"} =~ /^120/) {
4219                print $fh 'link ' or croak $!;
4220                $self->change_file_prop($fbat,'svn:special','*');
4221        } elsif ($m->{mode_a} =~ /^120/ && $m->{"mode_$which"} !~ /^120/) {
4222                $self->change_file_prop($fbat,'svn:special',undef);
4223        }
4224        my $blob = $m->{"sha1_$which"};
4225        return ($fh,) if ($blob =~ /^0{40}$/);
4226        my $size = $::_repository->cat_blob($blob, $fh);
4227        croak "Failed to read object $blob" if ($size < 0);
4228        $fh->flush == 0 or croak $!;
4229        seek $fh, 0, 0 or croak $!;
4230
4231        my $exp = ::md5sum($fh);
4232        seek $fh, 0, 0 or croak $!;
4233        return ($fh, $exp);
4234}
4235
4236sub chg_file {
4237        my ($self, $fbat, $m) = @_;
4238        if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
4239                $self->change_file_prop($fbat,'svn:executable','*');
4240        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
4241                $self->change_file_prop($fbat,'svn:executable',undef);
4242        }
4243        my ($fh_a, $exp_a) = _chg_file_get_blob $self, $fbat, $m, 'a';
4244        my ($fh_b, $exp_b) = _chg_file_get_blob $self, $fbat, $m, 'b';
4245        my $pool = SVN::Pool->new;
4246        my $atd = $self->apply_textdelta($fbat, $exp_a, $pool);
4247        if (-s $fh_a) {
4248                my $txstream = SVN::TxDelta::new ($fh_a, $fh_b, $pool);
4249                my $res = SVN::TxDelta::send_txstream($txstream, @$atd, $pool);
4250                if (defined $res) {
4251                        die "Unexpected result from send_txstream: $res\n",
4252                            "(SVN::Core::VERSION: $SVN::Core::VERSION)\n";
4253                }
4254        } else {
4255                my $got = SVN::TxDelta::send_stream($fh_b, @$atd, $pool);
4256                die "Checksum mismatch\nexpected: $exp_b\ngot: $got\n"
4257                    if ($got ne $exp_b);
4258        }
4259        Git::temp_release($fh_b, 1);
4260        Git::temp_release($fh_a, 1);
4261        $pool->clear;
4262}
4263
4264sub D {
4265        my ($self, $m) = @_;
4266        my ($dir, $file) = split_path($m->{file_b});
4267        my $pbat = $self->ensure_path($dir);
4268        print "\tD\t$m->{file_b}\n" unless $::_q;
4269        $self->delete_entry($m->{file_b}, $pbat);
4270}
4271
4272sub close_edit {
4273        my ($self) = @_;
4274        my ($p,$bat) = ($self->{pool}, $self->{bat});
4275        foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
4276                next if $_ eq '';
4277                $self->close_directory($bat->{$_}, $p);
4278        }
4279        $self->close_directory($bat->{''}, $p);
4280        $self->SUPER::close_edit($p);
4281        $p->clear;
4282}
4283
4284sub abort_edit {
4285        my ($self) = @_;
4286        $self->SUPER::abort_edit($self->{pool});
4287}
4288
4289sub DESTROY {
4290        my $self = shift;
4291        $self->SUPER::DESTROY(@_);
4292        $self->{pool}->clear;
4293}
4294
4295# this drives the editor
4296sub apply_diff {
4297        my ($self) = @_;
4298        my $mods = $self->{mods};
4299        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
4300        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
4301                my $f = $m->{chg};
4302                if (defined $o{$f}) {
4303                        $self->$f($m);
4304                } else {
4305                        fatal("Invalid change type: $f");
4306                }
4307        }
4308        $self->rmdirs if $_rmdir;
4309        if (@$mods == 0) {
4310                $self->abort_edit;
4311        } else {
4312                $self->close_edit;
4313        }
4314        return scalar @$mods;
4315}
4316
4317package Git::SVN::Ra;
4318use vars qw/@ISA $config_dir $_log_window_size/;
4319use strict;
4320use warnings;
4321my ($ra_invalid, $can_do_switch, %ignored_err, $RA);
4322
4323BEGIN {
4324        # enforce temporary pool usage for some simple functions
4325        no strict 'refs';
4326        for my $f (qw/rev_proplist get_latest_revnum get_uuid get_repos_root
4327                      get_file/) {
4328                my $SUPER = "SUPER::$f";
4329                *$f = sub {
4330                        my $self = shift;
4331                        my $pool = SVN::Pool->new;
4332                        my @ret = $self->$SUPER(@_,$pool);
4333                        $pool->clear;
4334                        wantarray ? @ret : $ret[0];
4335                };
4336        }
4337}
4338
4339sub _auth_providers () {
4340        [
4341          SVN::Client::get_simple_provider(),
4342          SVN::Client::get_ssl_server_trust_file_provider(),
4343          SVN::Client::get_simple_prompt_provider(
4344            \&Git::SVN::Prompt::simple, 2),
4345          SVN::Client::get_ssl_client_cert_file_provider(),
4346          SVN::Client::get_ssl_client_cert_prompt_provider(
4347            \&Git::SVN::Prompt::ssl_client_cert, 2),
4348          SVN::Client::get_ssl_client_cert_pw_file_provider(),
4349          SVN::Client::get_ssl_client_cert_pw_prompt_provider(
4350            \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
4351          SVN::Client::get_username_provider(),
4352          SVN::Client::get_ssl_server_trust_prompt_provider(
4353            \&Git::SVN::Prompt::ssl_server_trust),
4354          SVN::Client::get_username_prompt_provider(
4355            \&Git::SVN::Prompt::username, 2)
4356        ]
4357}
4358
4359sub escape_uri_only {
4360        my ($uri) = @_;
4361        my @tmp;
4362        foreach (split m{/}, $uri) {
4363                s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
4364                push @tmp, $_;
4365        }
4366        join('/', @tmp);
4367}
4368
4369sub escape_url {
4370        my ($url) = @_;
4371        if ($url =~ m#^(https?)://([^/]+)(.*)$#) {
4372                my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
4373                $url = "$scheme://$domain$uri";
4374        }
4375        $url;
4376}
4377
4378sub new {
4379        my ($class, $url) = @_;
4380        $url =~ s!/+$!!;
4381        return $RA if ($RA && $RA->{url} eq $url);
4382
4383        SVN::_Core::svn_config_ensure($config_dir, undef);
4384        my ($baton, $callbacks) = SVN::Core::auth_open_helper(_auth_providers);
4385        my $config = SVN::Core::config_get_config($config_dir);
4386        $RA = undef;
4387        my $dont_store_passwords = 1;
4388        my $conf_t = ${$config}{'config'};
4389        {
4390                no warnings 'once';
4391                # The usage of $SVN::_Core::SVN_CONFIG_* variables
4392                # produces warnings that variables are used only once.
4393                # I had not found the better way to shut them up, so
4394                # the warnings of type 'once' are disabled in this block.
4395                if (SVN::_Core::svn_config_get_bool($conf_t,
4396                    $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4397                    $SVN::_Core::SVN_CONFIG_OPTION_STORE_PASSWORDS,
4398                    1) == 0) {
4399                        SVN::_Core::svn_auth_set_parameter($baton,
4400                            $SVN::_Core::SVN_AUTH_PARAM_DONT_STORE_PASSWORDS,
4401                            bless (\$dont_store_passwords, "_p_void"));
4402                }
4403                if (SVN::_Core::svn_config_get_bool($conf_t,
4404                    $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4405                    $SVN::_Core::SVN_CONFIG_OPTION_STORE_AUTH_CREDS,
4406                    1) == 0) {
4407                        $Git::SVN::Prompt::_no_auth_cache = 1;
4408                }
4409        } # no warnings 'once'
4410        my $self = SVN::Ra->new(url => escape_url($url), auth => $baton,
4411                              config => $config,
4412                              pool => SVN::Pool->new,
4413                              auth_provider_callbacks => $callbacks);
4414        $self->{url} = $url;
4415        $self->{svn_path} = $url;
4416        $self->{repos_root} = $self->get_repos_root;
4417        $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
4418        $self->{cache} = { check_path => { r => 0, data => {} },
4419                           get_dir => { r => 0, data => {} } };
4420        $RA = bless $self, $class;
4421}
4422
4423sub check_path {
4424        my ($self, $path, $r) = @_;
4425        my $cache = $self->{cache}->{check_path};
4426        if ($r == $cache->{r} && exists $cache->{data}->{$path}) {
4427                return $cache->{data}->{$path};
4428        }
4429        my $pool = SVN::Pool->new;
4430        my $t = $self->SUPER::check_path($path, $r, $pool);
4431        $pool->clear;
4432        if ($r != $cache->{r}) {
4433                %{$cache->{data}} = ();
4434                $cache->{r} = $r;
4435        }
4436        $cache->{data}->{$path} = $t;
4437}
4438
4439sub get_dir {
4440        my ($self, $dir, $r) = @_;
4441        my $cache = $self->{cache}->{get_dir};
4442        if ($r == $cache->{r}) {
4443                if (my $x = $cache->{data}->{$dir}) {
4444                        return wantarray ? @$x : $x->[0];
4445                }
4446        }
4447        my $pool = SVN::Pool->new;
4448        my ($d, undef, $props) = $self->SUPER::get_dir($dir, $r, $pool);
4449        my %dirents = map { $_ => { kind => $d->{$_}->kind } } keys %$d;
4450        $pool->clear;
4451        if ($r != $cache->{r}) {
4452                %{$cache->{data}} = ();
4453                $cache->{r} = $r;
4454        }
4455        $cache->{data}->{$dir} = [ \%dirents, $r, $props ];
4456        wantarray ? (\%dirents, $r, $props) : \%dirents;
4457}
4458
4459sub DESTROY {
4460        # do not call the real DESTROY since we store ourselves in $RA
4461}
4462
4463# get_log(paths, start, end, limit,
4464#         discover_changed_paths, strict_node_history, receiver)
4465sub get_log {
4466        my ($self, @args) = @_;
4467        my $pool = SVN::Pool->new;
4468
4469        # svn_log_changed_path_t objects passed to get_log are likely to be
4470        # overwritten even if only the refs are copied to an external variable,
4471        # so we should dup the structures in their entirety.  Using an
4472        # externally passed pool (instead of our temporary and quickly cleared
4473        # pool in Git::SVN::Ra) does not help matters at all...
4474        my $receiver = pop @args;
4475        my $prefix = "/".$self->{svn_path};
4476        $prefix =~ s#/+($)##;
4477        my $prefix_regex = qr#^\Q$prefix\E#;
4478        push(@args, sub {
4479                my ($paths) = $_[0];
4480                return &$receiver(@_) unless $paths;
4481                $_[0] = ();
4482                foreach my $p (keys %$paths) {
4483                        my $i = $paths->{$p};
4484                        # Make path relative to our url, not repos_root
4485                        $p =~ s/$prefix_regex//;
4486                        my %s = map { $_ => $i->$_; }
4487                                qw/copyfrom_path copyfrom_rev action/;
4488                        if ($s{'copyfrom_path'}) {
4489                                $s{'copyfrom_path'} =~ s/$prefix_regex//;
4490                        }
4491                        $_[0]{$p} = \%s;
4492                }
4493                &$receiver(@_);
4494        });
4495
4496
4497        # the limit parameter was not supported in SVN 1.1.x, so we
4498        # drop it.  Therefore, the receiver callback passed to it
4499        # is made aware of this limitation by being wrapped if
4500        # the limit passed to is being wrapped.
4501        if ($SVN::Core::VERSION le '1.2.0') {
4502                my $limit = splice(@args, 3, 1);
4503                if ($limit > 0) {
4504                        my $receiver = pop @args;
4505                        push(@args, sub { &$receiver(@_) if (--$limit >= 0) });
4506                }
4507        }
4508        my $ret = $self->SUPER::get_log(@args, $pool);
4509        $pool->clear;
4510        $ret;
4511}
4512
4513sub trees_match {
4514        my ($self, $url1, $rev1, $url2, $rev2) = @_;
4515        my $ctx = SVN::Client->new(auth => _auth_providers);
4516        my $out = IO::File->new_tmpfile;
4517
4518        # older SVN (1.1.x) doesn't take $pool as the last parameter for
4519        # $ctx->diff(), so we'll create a default one
4520        my $pool = SVN::Pool->new_default_sub;
4521
4522        $ra_invalid = 1; # this will open a new SVN::Ra connection to $url1
4523        $ctx->diff([], $url1, $rev1, $url2, $rev2, 1, 1, 0, $out, $out);
4524        $out->flush;
4525        my $ret = (($out->stat)[7] == 0);
4526        close $out or croak $!;
4527
4528        $ret;
4529}
4530
4531sub get_commit_editor {
4532        my ($self, $log, $cb, $pool) = @_;
4533        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
4534        $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
4535}
4536
4537sub gs_do_update {
4538        my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
4539        my $new = ($rev_a == $rev_b);
4540        my $path = $gs->{path};
4541
4542        if ($new && -e $gs->{index}) {
4543                unlink $gs->{index} or die
4544                  "Couldn't unlink index: $gs->{index}: $!\n";
4545        }
4546        my $pool = SVN::Pool->new;
4547        $editor->set_path_strip($path);
4548        my (@pc) = split m#/#, $path;
4549        my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
4550                                        1, $editor, $pool);
4551        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
4552
4553        # Since we can't rely on svn_ra_reparent being available, we'll
4554        # just have to do some magic with set_path to make it so
4555        # we only want a partial path.
4556        my $sp = '';
4557        my $final = join('/', @pc);
4558        while (@pc) {
4559                $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
4560                $sp .= '/' if length $sp;
4561                $sp .= shift @pc;
4562        }
4563        die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
4564
4565        $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
4566
4567        $reporter->finish_report($pool);
4568        $pool->clear;
4569        $editor->{git_commit_ok};
4570}
4571
4572# this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
4573# svn_ra_reparent didn't work before 1.4)
4574sub gs_do_switch {
4575        my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
4576        my $path = $gs->{path};
4577        my $pool = SVN::Pool->new;
4578
4579        my $full_url = $self->{url};
4580        my $old_url = $full_url;
4581        $full_url .= '/' . $path if length $path;
4582        my ($ra, $reparented);
4583
4584        if ($old_url =~ m#^svn(\+ssh)?://# ||
4585            ($full_url =~ m#^https?://# &&
4586             escape_url($full_url) ne $full_url)) {
4587                $_[0] = undef;
4588                $self = undef;
4589                $RA = undef;
4590                $ra = Git::SVN::Ra->new($full_url);
4591                $ra_invalid = 1;
4592        } elsif ($old_url ne $full_url) {
4593                SVN::_Ra::svn_ra_reparent($self->{session}, $full_url, $pool);
4594                $self->{url} = $full_url;
4595                $reparented = 1;
4596        }
4597
4598        $ra ||= $self;
4599        $url_b = escape_url($url_b);
4600        my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
4601        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
4602        $reporter->set_path('', $rev_a, 0, @lock, $pool);
4603        $reporter->finish_report($pool);
4604
4605        if ($reparented) {
4606                SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
4607                $self->{url} = $old_url;
4608        }
4609
4610        $pool->clear;
4611        $editor->{git_commit_ok};
4612}
4613
4614sub longest_common_path {
4615        my ($gsv, $globs) = @_;
4616        my %common;
4617        my $common_max = scalar @$gsv;
4618
4619        foreach my $gs (@$gsv) {
4620                my @tmp = split m#/#, $gs->{path};
4621                my $p = '';
4622                foreach (@tmp) {
4623                        $p .= length($p) ? "/$_" : $_;
4624                        $common{$p} ||= 0;
4625                        $common{$p}++;
4626                }
4627        }
4628        $globs ||= [];
4629        $common_max += scalar @$globs;
4630        foreach my $glob (@$globs) {
4631                my @tmp = split m#/#, $glob->{path}->{left};
4632                my $p = '';
4633                foreach (@tmp) {
4634                        $p .= length($p) ? "/$_" : $_;
4635                        $common{$p} ||= 0;
4636                        $common{$p}++;
4637                }
4638        }
4639
4640        my $longest_path = '';
4641        foreach (sort {length $b <=> length $a} keys %common) {
4642                if ($common{$_} == $common_max) {
4643                        $longest_path = $_;
4644                        last;
4645                }
4646        }
4647        $longest_path;
4648}
4649
4650sub gs_fetch_loop_common {
4651        my ($self, $base, $head, $gsv, $globs) = @_;
4652        return if ($base > $head);
4653        my $inc = $_log_window_size;
4654        my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
4655        my $longest_path = longest_common_path($gsv, $globs);
4656        my $ra_url = $self->{url};
4657        my $find_trailing_edge;
4658        while (1) {
4659                my %revs;
4660                my $err;
4661                my $err_handler = $SVN::Error::handler;
4662                $SVN::Error::handler = sub {
4663                        ($err) = @_;
4664                        skip_unknown_revs($err);
4665                };
4666                sub _cb {
4667                        my ($paths, $r, $author, $date, $log) = @_;
4668                        [ $paths,
4669                          { author => $author, date => $date, log => $log } ];
4670                }
4671                $self->get_log([$longest_path], $min, $max, 0, 1, 1,
4672                               sub { $revs{$_[1]} = _cb(@_) });
4673                if ($err) {
4674                        print "Checked through r$max\r";
4675                } else {
4676                        $find_trailing_edge = 1;
4677                }
4678                if ($err and $find_trailing_edge) {
4679                        print STDERR "Path '$longest_path' ",
4680                                     "was probably deleted:\n",
4681                                     $err->expanded_message,
4682                                     "\nWill attempt to follow ",
4683                                     "revisions r$min .. r$max ",
4684                                     "committed before the deletion\n";
4685                        my $hi = $max;
4686                        while (--$hi >= $min) {
4687                                my $ok;
4688                                $self->get_log([$longest_path], $min, $hi,
4689                                               0, 1, 1, sub {
4690                                               $ok = $_[1];
4691                                               $revs{$_[1]} = _cb(@_) });
4692                                if ($ok) {
4693                                        print STDERR "r$min .. r$ok OK\n";
4694                                        last;
4695                                }
4696                        }
4697                        $find_trailing_edge = 0;
4698                }
4699                $SVN::Error::handler = $err_handler;
4700
4701                my %exists = map { $_->{path} => $_ } @$gsv;
4702                foreach my $r (sort {$a <=> $b} keys %revs) {
4703                        my ($paths, $logged) = @{$revs{$r}};
4704
4705                        foreach my $gs ($self->match_globs(\%exists, $paths,
4706                                                           $globs, $r)) {
4707                                if ($gs->rev_map_max >= $r) {
4708                                        next;
4709                                }
4710                                next unless $gs->match_paths($paths, $r);
4711                                $gs->{logged_rev_props} = $logged;
4712                                if (my $last_commit = $gs->last_commit) {
4713                                        $gs->assert_index_clean($last_commit);
4714                                }
4715                                my $log_entry = $gs->do_fetch($paths, $r);
4716                                if ($log_entry) {
4717                                        $gs->do_git_commit($log_entry);
4718                                }
4719                                $INDEX_FILES{$gs->{index}} = 1;
4720                        }
4721                        foreach my $g (@$globs) {
4722                                my $k = "svn-remote.$g->{remote}." .
4723                                        "$g->{t}-maxRev";
4724                                Git::SVN::tmp_config($k, $r);
4725                        }
4726                        if ($ra_invalid) {
4727                                $_[0] = undef;
4728                                $self = undef;
4729                                $RA = undef;
4730                                $self = Git::SVN::Ra->new($ra_url);
4731                                $ra_invalid = undef;
4732                        }
4733                }
4734                # pre-fill the .rev_db since it'll eventually get filled in
4735                # with '0' x40 if something new gets committed
4736                foreach my $gs (@$gsv) {
4737                        next if $gs->rev_map_max >= $max;
4738                        next if defined $gs->rev_map_get($max);
4739                        $gs->rev_map_set($max, 0 x40);
4740                }
4741                foreach my $g (@$globs) {
4742                        my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
4743                        Git::SVN::tmp_config($k, $max);
4744                }
4745                last if $max >= $head;
4746                $min = $max + 1;
4747                $max += $inc;
4748                $max = $head if ($max > $head);
4749        }
4750        Git::SVN::gc();
4751}
4752
4753sub get_dir_globbed {
4754        my ($self, $left, $depth, $r) = @_;
4755
4756        my @x = eval { $self->get_dir($left, $r) };
4757        return unless scalar @x == 3;
4758        my $dirents = $x[0];
4759        my @finalents;
4760        foreach my $de (keys %$dirents) {
4761                next if $dirents->{$de}->{kind} != $SVN::Node::dir;
4762                if ($depth > 1) {
4763                        my @args = ("$left/$de", $depth - 1, $r);
4764                        foreach my $dir ($self->get_dir_globbed(@args)) {
4765                                push @finalents, "$de/$dir";
4766                        }
4767                } else {
4768                        push @finalents, $de;
4769                }
4770        }
4771        @finalents;
4772}
4773
4774sub match_globs {
4775        my ($self, $exists, $paths, $globs, $r) = @_;
4776
4777        sub get_dir_check {
4778                my ($self, $exists, $g, $r) = @_;
4779
4780                my @dirs = $self->get_dir_globbed($g->{path}->{left},
4781                                                  $g->{path}->{depth},
4782                                                  $r);
4783
4784                foreach my $de (@dirs) {
4785                        my $p = $g->{path}->full_path($de);
4786                        next if $exists->{$p};
4787                        next if (length $g->{path}->{right} &&
4788                                 ($self->check_path($p, $r) !=
4789                                  $SVN::Node::dir));
4790                        $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
4791                                         $g->{ref}->full_path($de), 1);
4792                }
4793        }
4794        foreach my $g (@$globs) {
4795                if (my $path = $paths->{"/$g->{path}->{left}"}) {
4796                        if ($path->{action} =~ /^[AR]$/) {
4797                                get_dir_check($self, $exists, $g, $r);
4798                        }
4799                }
4800                foreach (keys %$paths) {
4801                        if (/$g->{path}->{left_regex}/ &&
4802                            !/$g->{path}->{regex}/) {
4803                                next if $paths->{$_}->{action} !~ /^[AR]$/;
4804                                get_dir_check($self, $exists, $g, $r);
4805                        }
4806                        next unless /$g->{path}->{regex}/;
4807                        my $p = $1;
4808                        my $pathname = $g->{path}->full_path($p);
4809                        next if $exists->{$pathname};
4810                        next if ($self->check_path($pathname, $r) !=
4811                                 $SVN::Node::dir);
4812                        $exists->{$pathname} = Git::SVN->init(
4813                                              $self->{url}, $pathname, undef,
4814                                              $g->{ref}->full_path($p), 1);
4815                }
4816                my $c = '';
4817                foreach (split m#/#, $g->{path}->{left}) {
4818                        $c .= "/$_";
4819                        next unless ($paths->{$c} &&
4820                                     ($paths->{$c}->{action} =~ /^[AR]$/));
4821                        get_dir_check($self, $exists, $g, $r);
4822                }
4823        }
4824        values %$exists;
4825}
4826
4827sub minimize_url {
4828        my ($self) = @_;
4829        return $self->{url} if ($self->{url} eq $self->{repos_root});
4830        my $url = $self->{repos_root};
4831        my @components = split(m!/!, $self->{svn_path});
4832        my $c = '';
4833        do {
4834                $url .= "/$c" if length $c;
4835                eval {
4836                        my $ra = (ref $self)->new($url);
4837                        my $latest = $ra->get_latest_revnum;
4838                        $ra->get_log("", $latest, 0, 1, 0, 1, sub {});
4839                };
4840        } while ($@ && ($c = shift @components));
4841        $url;
4842}
4843
4844sub can_do_switch {
4845        my $self = shift;
4846        unless (defined $can_do_switch) {
4847                my $pool = SVN::Pool->new;
4848                my $rep = eval {
4849                        $self->do_switch(1, '', 0, $self->{url},
4850                                         SVN::Delta::Editor->new, $pool);
4851                };
4852                if ($@) {
4853                        $can_do_switch = 0;
4854                } else {
4855                        $rep->abort_report($pool);
4856                        $can_do_switch = 1;
4857                }
4858                $pool->clear;
4859        }
4860        $can_do_switch;
4861}
4862
4863sub skip_unknown_revs {
4864        my ($err) = @_;
4865        my $errno = $err->apr_err();
4866        # Maybe the branch we're tracking didn't
4867        # exist when the repo started, so it's
4868        # not an error if it doesn't, just continue
4869        #
4870        # Wonderfully consistent library, eh?
4871        # 160013 - svn:// and file://
4872        # 175002 - http(s)://
4873        # 175007 - http(s):// (this repo required authorization, too...)
4874        #   More codes may be discovered later...
4875        if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
4876                my $err_key = $err->expanded_message;
4877                # revision numbers change every time, filter them out
4878                $err_key =~ s/\d+/\0/g;
4879                $err_key = "$errno\0$err_key";
4880                unless ($ignored_err{$err_key}) {
4881                        warn "W: Ignoring error from SVN, path probably ",
4882                             "does not exist: ($errno): ",
4883                             $err->expanded_message,"\n";
4884                        warn "W: Do not be alarmed at the above message ",
4885                             "git-svn is just searching aggressively for ",
4886                             "old history.\n",
4887                             "This may take a while on large repositories\n";
4888                        $ignored_err{$err_key} = 1;
4889                }
4890                return;
4891        }
4892        die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
4893}
4894
4895package Git::SVN::Log;
4896use strict;
4897use warnings;
4898use POSIX qw/strftime/;
4899use Time::Local;
4900use constant commit_log_separator => ('-' x 72) . "\n";
4901use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
4902            %rusers $show_commit $incremental/;
4903my $l_fmt;
4904
4905sub cmt_showable {
4906        my ($c) = @_;
4907        return 1 if defined $c->{r};
4908
4909        # big commit message got truncated by the 16k pretty buffer in rev-list
4910        if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
4911                                $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
4912                @{$c->{l}} = ();
4913                my @log = command(qw/cat-file commit/, $c->{c});
4914
4915                # shift off the headers
4916                shift @log while ($log[0] ne '');
4917                shift @log;
4918
4919                # TODO: make $c->{l} not have a trailing newline in the future
4920                @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
4921
4922                (undef, $c->{r}, undef) = ::extract_metadata(
4923                                (grep(/^git-svn-id: /, @log))[-1]);
4924        }
4925        return defined $c->{r};
4926}
4927
4928sub log_use_color {
4929        return $color || Git->repository->get_colorbool('color.diff');
4930}
4931
4932sub git_svn_log_cmd {
4933        my ($r_min, $r_max, @args) = @_;
4934        my $head = 'HEAD';
4935        my (@files, @log_opts);
4936        foreach my $x (@args) {
4937                if ($x eq '--' || @files) {
4938                        push @files, $x;
4939                } else {
4940                        if (::verify_ref("$x^0")) {
4941                                $head = $x;
4942                        } else {
4943                                push @log_opts, $x;
4944                        }
4945                }
4946        }
4947
4948        my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
4949        $gs ||= Git::SVN->_new;
4950        my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
4951                   $gs->refname);
4952        push @cmd, '-r' unless $non_recursive;
4953        push @cmd, qw/--raw --name-status/ if $verbose;
4954        push @cmd, '--color' if log_use_color();
4955        push @cmd, @log_opts;
4956        if (defined $r_max && $r_max == $r_min) {
4957                push @cmd, '--max-count=1';
4958                if (my $c = $gs->rev_map_get($r_max)) {
4959                        push @cmd, $c;
4960                }
4961        } elsif (defined $r_max) {
4962                if ($r_max < $r_min) {
4963                        ($r_min, $r_max) = ($r_max, $r_min);
4964                }
4965                my (undef, $c_max) = $gs->find_rev_before($r_max, 1, $r_min);
4966                my (undef, $c_min) = $gs->find_rev_after($r_min, 1, $r_max);
4967                # If there are no commits in the range, both $c_max and $c_min
4968                # will be undefined.  If there is at least 1 commit in the
4969                # range, both will be defined.
4970                return () if !defined $c_min || !defined $c_max;
4971                if ($c_min eq $c_max) {
4972                        push @cmd, '--max-count=1', $c_min;
4973                } else {
4974                        push @cmd, '--boundary', "$c_min..$c_max";
4975                }
4976        }
4977        return (@cmd, @files);
4978}
4979
4980# adapted from pager.c
4981sub config_pager {
4982        $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
4983        if (!defined $pager) {
4984                $pager = 'less';
4985        } elsif (length $pager == 0 || $pager eq 'cat') {
4986                $pager = undef;
4987        }
4988        $ENV{GIT_PAGER_IN_USE} = defined($pager);
4989}
4990
4991sub run_pager {
4992        return unless -t *STDOUT && defined $pager;
4993        pipe my ($rfd, $wfd) or return;
4994        defined(my $pid = fork) or ::fatal "Can't fork: $!";
4995        if (!$pid) {
4996                open STDOUT, '>&', $wfd or
4997                                     ::fatal "Can't redirect to stdout: $!";
4998                return;
4999        }
5000        open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!";
5001        $ENV{LESS} ||= 'FRSX';
5002        exec $pager or ::fatal "Can't run pager: $! ($pager)";
5003}
5004
5005sub format_svn_date {
5006        # some systmes don't handle or mishandle %z, so be creative.
5007        my $t = shift || time;
5008        my $gm = timelocal(gmtime($t));
5009        my $sign = qw( + + - )[ $t <=> $gm ];
5010        my $gmoff = sprintf("%s%02d%02d", $sign, (gmtime(abs($t - $gm)))[2,1]);
5011        return strftime("%Y-%m-%d %H:%M:%S $gmoff (%a, %d %b %Y)", localtime($t));
5012}
5013
5014sub parse_git_date {
5015        my ($t, $tz) = @_;
5016        # Date::Parse isn't in the standard Perl distro :(
5017        if ($tz =~ s/^\+//) {
5018                $t += tz_to_s_offset($tz);
5019        } elsif ($tz =~ s/^\-//) {
5020                $t -= tz_to_s_offset($tz);
5021        }
5022        return $t;
5023}
5024
5025sub set_local_timezone {
5026        if (defined $TZ) {
5027                $ENV{TZ} = $TZ;
5028        } else {
5029                delete $ENV{TZ};
5030        }
5031}
5032
5033sub tz_to_s_offset {
5034        my ($tz) = @_;
5035        $tz =~ s/(\d\d)$//;
5036        return ($1 * 60) + ($tz * 3600);
5037}
5038
5039sub get_author_info {
5040        my ($dest, $author, $t, $tz) = @_;
5041        $author =~ s/(?:^\s*|\s*$)//g;
5042        $dest->{a_raw} = $author;
5043        my $au;
5044        if ($::_authors) {
5045                $au = $rusers{$author} || undef;
5046        }
5047        if (!$au) {
5048                ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
5049        }
5050        $dest->{t} = $t;
5051        $dest->{tz} = $tz;
5052        $dest->{a} = $au;
5053        $dest->{t_utc} = parse_git_date($t, $tz);
5054}
5055
5056sub process_commit {
5057        my ($c, $r_min, $r_max, $defer) = @_;
5058        if (defined $r_min && defined $r_max) {
5059                if ($r_min == $c->{r} && $r_min == $r_max) {
5060                        show_commit($c);
5061                        return 0;
5062                }
5063                return 1 if $r_min == $r_max;
5064                if ($r_min < $r_max) {
5065                        # we need to reverse the print order
5066                        return 0 if (defined $limit && --$limit < 0);
5067                        push @$defer, $c;
5068                        return 1;
5069                }
5070                if ($r_min != $r_max) {
5071                        return 1 if ($r_min < $c->{r});
5072                        return 1 if ($r_max > $c->{r});
5073                }
5074        }
5075        return 0 if (defined $limit && --$limit < 0);
5076        show_commit($c);
5077        return 1;
5078}
5079
5080sub show_commit {
5081        my $c = shift;
5082        if ($oneline) {
5083                my $x = "\n";
5084                if (my $l = $c->{l}) {
5085                        while ($l->[0] =~ /^\s*$/) { shift @$l }
5086                        $x = $l->[0];
5087                }
5088                $l_fmt ||= 'A' . length($c->{r});
5089                print 'r',pack($l_fmt, $c->{r}),' | ';
5090                print "$c->{c} | " if $show_commit;
5091                print $x;
5092        } else {
5093                show_commit_normal($c);
5094        }
5095}
5096
5097sub show_commit_changed_paths {
5098        my ($c) = @_;
5099        return unless $c->{changed};
5100        print "Changed paths:\n", @{$c->{changed}};
5101}
5102
5103sub show_commit_normal {
5104        my ($c) = @_;
5105        print commit_log_separator, "r$c->{r} | ";
5106        print "$c->{c} | " if $show_commit;
5107        print "$c->{a} | ", format_svn_date($c->{t_utc}), ' | ';
5108        my $nr_line = 0;
5109
5110        if (my $l = $c->{l}) {
5111                while ($l->[$#$l] eq "\n" && $#$l > 0
5112                                          && $l->[($#$l - 1)] eq "\n") {
5113                        pop @$l;
5114                }
5115                $nr_line = scalar @$l;
5116                if (!$nr_line) {
5117                        print "1 line\n\n\n";
5118                } else {
5119                        if ($nr_line == 1) {
5120                                $nr_line = '1 line';
5121                        } else {
5122                                $nr_line .= ' lines';
5123                        }
5124                        print $nr_line, "\n";
5125                        show_commit_changed_paths($c);
5126                        print "\n";
5127                        print $_ foreach @$l;
5128                }
5129        } else {
5130                print "1 line\n";
5131                show_commit_changed_paths($c);
5132                print "\n";
5133
5134        }
5135        foreach my $x (qw/raw stat diff/) {
5136                if ($c->{$x}) {
5137                        print "\n";
5138                        print $_ foreach @{$c->{$x}}
5139                }
5140        }
5141}
5142
5143sub cmd_show_log {
5144        my (@args) = @_;
5145        my ($r_min, $r_max);
5146        my $r_last = -1; # prevent dupes
5147        set_local_timezone();
5148        if (defined $::_revision) {
5149                if ($::_revision =~ /^(\d+):(\d+)$/) {
5150                        ($r_min, $r_max) = ($1, $2);
5151                } elsif ($::_revision =~ /^\d+$/) {
5152                        $r_min = $r_max = $::_revision;
5153                } else {
5154                        ::fatal "-r$::_revision is not supported, use ",
5155                                "standard 'git log' arguments instead";
5156                }
5157        }
5158
5159        config_pager();
5160        @args = git_svn_log_cmd($r_min, $r_max, @args);
5161        if (!@args) {
5162                print commit_log_separator unless $incremental || $oneline;
5163                return;
5164        }
5165        my $log = command_output_pipe(@args);
5166        run_pager();
5167        my (@k, $c, $d, $stat);
5168        my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
5169        while (<$log>) {
5170                if (/^${esc_color}commit -?($::sha1_short)/o) {
5171                        my $cmt = $1;
5172                        if ($c && cmt_showable($c) && $c->{r} != $r_last) {
5173                                $r_last = $c->{r};
5174                                process_commit($c, $r_min, $r_max, \@k) or
5175                                                                goto out;
5176                        }
5177                        $d = undef;
5178                        $c = { c => $cmt };
5179                } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
5180                        get_author_info($c, $1, $2, $3);
5181                } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
5182                        # ignore
5183                } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
5184                        push @{$c->{raw}}, $_;
5185                } elsif (/^${esc_color}[ACRMDT]\t/) {
5186                        # we could add $SVN->{svn_path} here, but that requires
5187                        # remote access at the moment (repo_path_split)...
5188                        s#^(${esc_color})([ACRMDT])\t#$1   $2 #o;
5189                        push @{$c->{changed}}, $_;
5190                } elsif (/^${esc_color}diff /o) {
5191                        $d = 1;
5192                        push @{$c->{diff}}, $_;
5193                } elsif ($d) {
5194                        push @{$c->{diff}}, $_;
5195                } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
5196                          $esc_color*[\+\-]*$esc_color$/x) {
5197                        $stat = 1;
5198                        push @{$c->{stat}}, $_;
5199                } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
5200                        push @{$c->{stat}}, $_;
5201                        $stat = undef;
5202                } elsif (/^${esc_color}    (git-svn-id:.+)$/o) {
5203                        ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
5204                } elsif (s/^${esc_color}    //o) {
5205                        push @{$c->{l}}, $_;
5206                }
5207        }
5208        if ($c && defined $c->{r} && $c->{r} != $r_last) {
5209                $r_last = $c->{r};
5210                process_commit($c, $r_min, $r_max, \@k);
5211        }
5212        if (@k) {
5213                ($r_min, $r_max) = ($r_max, $r_min);
5214                process_commit($_, $r_min, $r_max) foreach reverse @k;
5215        }
5216out:
5217        close $log;
5218        print commit_log_separator unless $incremental || $oneline;
5219}
5220
5221sub cmd_blame {
5222        my $path = pop;
5223
5224        config_pager();
5225        run_pager();
5226
5227        my ($fh, $ctx, $rev);
5228
5229        if ($_git_format) {
5230                ($fh, $ctx) = command_output_pipe('blame', @_, $path);
5231                while (my $line = <$fh>) {
5232                        if ($line =~ /^\^?([[:xdigit:]]+)\s/) {
5233                                # Uncommitted edits show up as a rev ID of
5234                                # all zeros, which we can't look up with
5235                                # cmt_metadata
5236                                if ($1 !~ /^0+$/) {
5237                                        (undef, $rev, undef) =
5238                                                ::cmt_metadata($1);
5239                                        $rev = '0' if (!$rev);
5240                                } else {
5241                                        $rev = '0';
5242                                }
5243                                $rev = sprintf('%-10s', $rev);
5244                                $line =~ s/^\^?[[:xdigit:]]+(\s)/$rev$1/;
5245                        }
5246                        print $line;
5247                }
5248        } else {
5249                ($fh, $ctx) = command_output_pipe('blame', '-p', @_, 'HEAD',
5250                                                  '--', $path);
5251                my ($sha1);
5252                my %authors;
5253                my @buffer;
5254                my %dsha; #distinct sha keys
5255
5256                while (my $line = <$fh>) {
5257                        push @buffer, $line;
5258                        if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5259                                $dsha{$1} = 1;
5260                        }
5261                }
5262
5263                my $s2r = ::cmt_sha2rev_batch([keys %dsha]);
5264
5265                foreach my $line (@buffer) {
5266                        if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5267                                $rev = $s2r->{$1};
5268                                $rev = '0' if (!$rev)
5269                        }
5270                        elsif ($line =~ /^author (.*)/) {
5271                                $authors{$rev} = $1;
5272                                $authors{$rev} =~ s/\s/_/g;
5273                        }
5274                        elsif ($line =~ /^\t(.*)$/) {
5275                                printf("%6s %10s %s\n", $rev, $authors{$rev}, $1);
5276                        }
5277                }
5278        }
5279        command_close_pipe($fh, $ctx);
5280}
5281
5282package Git::SVN::Migration;
5283# these version numbers do NOT correspond to actual version numbers
5284# of git nor git-svn.  They are just relative.
5285#
5286# v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
5287#
5288# v1 layout: .git/$id/info/url, refs/remotes/$id
5289#
5290# v2 layout: .git/svn/$id/info/url, refs/remotes/$id
5291#
5292# v3 layout: .git/svn/$id, refs/remotes/$id
5293#            - info/url may remain for backwards compatibility
5294#            - this is what we migrate up to this layout automatically,
5295#            - this will be used by git svn init on single branches
5296# v3.1 layout (auto migrated):
5297#            - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
5298#              for backwards compatibility
5299#
5300# v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
5301#            - this is only created for newly multi-init-ed
5302#              repositories.  Similar in spirit to the
5303#              --use-separate-remotes option in git-clone (now default)
5304#            - we do not automatically migrate to this (following
5305#              the example set by core git)
5306#
5307# v5 layout: .rev_db.$UUID => .rev_map.$UUID
5308#            - newer, more-efficient format that uses 24-bytes per record
5309#              with no filler space.
5310#            - use xxd -c24 < .rev_map.$UUID to view and debug
5311#            - This is a one-way migration, repositories updated to the
5312#              new format will not be able to use old git-svn without
5313#              rebuilding the .rev_db.  Rebuilding the rev_db is not
5314#              possible if noMetadata or useSvmProps are set; but should
5315#              be no problem for users that use the (sensible) defaults.
5316use strict;
5317use warnings;
5318use Carp qw/croak/;
5319use File::Path qw/mkpath/;
5320use File::Basename qw/dirname basename/;
5321use vars qw/$_minimize/;
5322
5323sub migrate_from_v0 {
5324        my $git_dir = $ENV{GIT_DIR};
5325        return undef unless -d $git_dir;
5326        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5327        my $migrated = 0;
5328        while (<$fh>) {
5329                chomp;
5330                my ($id, $orig_ref) = ($_, $_);
5331                next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
5332                next unless -f "$git_dir/$id/info/url";
5333                my $new_ref = "refs/remotes/$id";
5334                if (::verify_ref("$new_ref^0")) {
5335                        print STDERR "W: $orig_ref is probably an old ",
5336                                     "branch used by an ancient version of ",
5337                                     "git-svn.\n",
5338                                     "However, $new_ref also exists.\n",
5339                                     "We will not be able ",
5340                                     "to use this branch until this ",
5341                                     "ambiguity is resolved.\n";
5342                        next;
5343                }
5344                print STDERR "Migrating from v0 layout...\n" if !$migrated;
5345                print STDERR "Renaming ref: $orig_ref => $new_ref\n";
5346                command_noisy('update-ref', $new_ref, $orig_ref);
5347                command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
5348                $migrated++;
5349        }
5350        command_close_pipe($fh, $ctx);
5351        print STDERR "Done migrating from v0 layout...\n" if $migrated;
5352        $migrated;
5353}
5354
5355sub migrate_from_v1 {
5356        my $git_dir = $ENV{GIT_DIR};
5357        my $migrated = 0;
5358        return $migrated unless -d $git_dir;
5359        my $svn_dir = "$git_dir/svn";
5360
5361        # just in case somebody used 'svn' as their $id at some point...
5362        return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
5363
5364        print STDERR "Migrating from a git-svn v1 layout...\n";
5365        mkpath([$svn_dir]);
5366        print STDERR "Data from a previous version of git-svn exists, but\n\t",
5367                     "$svn_dir\n\t(required for this version ",
5368                     "($::VERSION) of git-svn) does not exist.\n";
5369        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5370        while (<$fh>) {
5371                my $x = $_;
5372                next unless $x =~ s#^refs/remotes/##;
5373                chomp $x;
5374                next unless -f "$git_dir/$x/info/url";
5375                my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
5376                next unless $u;
5377                my $dn = dirname("$git_dir/svn/$x");
5378                mkpath([$dn]) unless -d $dn;
5379                if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
5380                        mkpath(["$git_dir/svn/svn"]);
5381                        print STDERR " - $git_dir/$x/info => ",
5382                                        "$git_dir/svn/$x/info\n";
5383                        rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
5384                               croak "$!: $x";
5385                        # don't worry too much about these, they probably
5386                        # don't exist with repos this old (save for index,
5387                        # and we can easily regenerate that)
5388                        foreach my $f (qw/unhandled.log index .rev_db/) {
5389                                rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
5390                        }
5391                } else {
5392                        print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
5393                        rename "$git_dir/$x", "$git_dir/svn/$x" or
5394                               croak "$!: $x";
5395                }
5396                $migrated++;
5397        }
5398        command_close_pipe($fh, $ctx);
5399        print STDERR "Done migrating from a git-svn v1 layout\n";
5400        $migrated;
5401}
5402
5403sub read_old_urls {
5404        my ($l_map, $pfx, $path) = @_;
5405        my @dir;
5406        foreach (<$path/*>) {
5407                if (-r "$_/info/url") {
5408                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
5409                        my $ref_id = $pfx . basename $_;
5410                        my $url = ::file_to_s("$_/info/url");
5411                        $l_map->{$ref_id} = $url;
5412                } elsif (-d $_) {
5413                        push @dir, $_;
5414                }
5415        }
5416        foreach (@dir) {
5417                my $x = $_;
5418                $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
5419                read_old_urls($l_map, $x, $_);
5420        }
5421}
5422
5423sub migrate_from_v2 {
5424        my @cfg = command(qw/config -l/);
5425        return if grep /^svn-remote\..+\.url=/, @cfg;
5426        my %l_map;
5427        read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
5428        my $migrated = 0;
5429
5430        foreach my $ref_id (sort keys %l_map) {
5431                eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
5432                if ($@) {
5433                        Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
5434                }
5435                $migrated++;
5436        }
5437        $migrated;
5438}
5439
5440sub minimize_connections {
5441        my $r = Git::SVN::read_all_remotes();
5442        my $new_urls = {};
5443        my $root_repos = {};
5444        foreach my $repo_id (keys %$r) {
5445                my $url = $r->{$repo_id}->{url} or next;
5446                my $fetch = $r->{$repo_id}->{fetch} or next;
5447                my $ra = Git::SVN::Ra->new($url);
5448
5449                # skip existing cases where we already connect to the root
5450                if (($ra->{url} eq $ra->{repos_root}) ||
5451                    ($ra->{repos_root} eq $repo_id)) {
5452                        $root_repos->{$ra->{url}} = $repo_id;
5453                        next;
5454                }
5455
5456                my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
5457                my $root_path = $ra->{url};
5458                $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
5459                foreach my $path (keys %$fetch) {
5460                        my $ref_id = $fetch->{$path};
5461                        my $gs = Git::SVN->new($ref_id, $repo_id, $path);
5462
5463                        # make sure we can read when connecting to
5464                        # a higher level of a repository
5465                        my ($last_rev, undef) = $gs->last_rev_commit;
5466                        if (!defined $last_rev) {
5467                                $last_rev = eval {
5468                                        $root_ra->get_latest_revnum;
5469                                };
5470                                next if $@;
5471                        }
5472                        my $new = $root_path;
5473                        $new .= length $path ? "/$path" : '';
5474                        eval {
5475                                $root_ra->get_log([$new], $last_rev, $last_rev,
5476                                                  0, 0, 1, sub { });
5477                        };
5478                        next if $@;
5479                        $new_urls->{$ra->{repos_root}}->{$new} =
5480                                { ref_id => $ref_id,
5481                                  old_repo_id => $repo_id,
5482                                  old_path => $path };
5483                }
5484        }
5485
5486        my @emptied;
5487        foreach my $url (keys %$new_urls) {
5488                # see if we can re-use an existing [svn-remote "repo_id"]
5489                # instead of creating a(n ugly) new section:
5490                my $repo_id = $root_repos->{$url} || $url;
5491
5492                my $fetch = $new_urls->{$url};
5493                foreach my $path (keys %$fetch) {
5494                        my $x = $fetch->{$path};
5495                        Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
5496                        my $pfx = "svn-remote.$x->{old_repo_id}";
5497
5498                        my $old_fetch = quotemeta("$x->{old_path}:".
5499                                                  "refs/remotes/$x->{ref_id}");
5500                        command_noisy(qw/config --unset/,
5501                                      "$pfx.fetch", '^'. $old_fetch . '$');
5502                        delete $r->{$x->{old_repo_id}}->
5503                               {fetch}->{$x->{old_path}};
5504                        if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
5505                                command_noisy(qw/config --unset/,
5506                                              "$pfx.url");
5507                                push @emptied, $x->{old_repo_id}
5508                        }
5509                }
5510        }
5511        if (@emptied) {
5512                my $file = $ENV{GIT_CONFIG} || "$ENV{GIT_DIR}/config";
5513                print STDERR <<EOF;
5514The following [svn-remote] sections in your config file ($file) are empty
5515and can be safely removed:
5516EOF
5517                print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
5518        }
5519}
5520
5521sub migration_check {
5522        migrate_from_v0();
5523        migrate_from_v1();
5524        migrate_from_v2();
5525        minimize_connections() if $_minimize;
5526}
5527
5528package Git::IndexInfo;
5529use strict;
5530use warnings;
5531use Git qw/command_input_pipe command_close_pipe/;
5532
5533sub new {
5534        my ($class) = @_;
5535        my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
5536        bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
5537}
5538
5539sub remove {
5540        my ($self, $path) = @_;
5541        if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
5542                return ++$self->{nr};
5543        }
5544        undef;
5545}
5546
5547sub update {
5548        my ($self, $mode, $hash, $path) = @_;
5549        if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
5550                return ++$self->{nr};
5551        }
5552        undef;
5553}
5554
5555sub DESTROY {
5556        my ($self) = @_;
5557        command_close_pipe($self->{gui}, $self->{ctx});
5558}
5559
5560package Git::SVN::GlobSpec;
5561use strict;
5562use warnings;
5563
5564sub new {
5565        my ($class, $glob) = @_;
5566        my $re = $glob;
5567        $re =~ s!/+$!!g; # no need for trailing slashes
5568        $re =~ m!^([^*]*)(\*(?:/\*)*)([^*]*)$!;
5569        my $temp = $re;
5570        my ($left, $right) = ($1, $3);
5571        $re = $2;
5572        my $depth = $re =~ tr/*/*/;
5573        if ($depth != $temp =~ tr/*/*/) {
5574                die "Only one set of wildcard directories " .
5575                        "(e.g. '*' or '*/*/*') is supported: '$glob'\n";
5576        }
5577        if ($depth == 0) {
5578                die "One '*' is needed for glob: '$glob'\n";
5579        }
5580        $re =~ s!\*!\[^/\]*!g;
5581        $re = quotemeta($left) . "($re)" . quotemeta($right);
5582        if (length $left && !($left =~ s!/+$!!g)) {
5583                die "Missing trailing '/' on left side of: '$glob' ($left)\n";
5584        }
5585        if (length $right && !($right =~ s!^/+!!g)) {
5586                die "Missing leading '/' on right side of: '$glob' ($right)\n";
5587        }
5588        my $left_re = qr/^\/\Q$left\E(\/|$)/;
5589        bless { left => $left, right => $right, left_regex => $left_re,
5590                regex => qr/$re/, glob => $glob, depth => $depth }, $class;
5591}
5592
5593sub full_path {
5594        my ($self, $path) = @_;
5595        return (length $self->{left} ? "$self->{left}/" : '') .
5596               $path . (length $self->{right} ? "/$self->{right}" : '');
5597}
5598
5599__END__
5600
5601Data structures:
5602
5603
5604$remotes = { # returned by read_all_remotes()
5605        'svn' => {
5606                # svn-remote.svn.url=https://svn.musicpd.org
5607                url => 'https://svn.musicpd.org',
5608                # svn-remote.svn.fetch=mpd/trunk:trunk
5609                fetch => {
5610                        'mpd/trunk' => 'trunk',
5611                },
5612                # svn-remote.svn.tags=mpd/tags/*:tags/*
5613                tags => {
5614                        path => {
5615                                left => 'mpd/tags',
5616                                right => '',
5617                                regex => qr!mpd/tags/([^/]+)$!,
5618                                glob => 'tags/*',
5619                        },
5620                        ref => {
5621                                left => 'tags',
5622                                right => '',
5623                                regex => qr!tags/([^/]+)$!,
5624                                glob => 'tags/*',
5625                        },
5626                }
5627        }
5628};
5629
5630$log_entry hashref as returned by libsvn_log_entry()
5631{
5632        log => 'whitespace-formatted log entry
5633',                                              # trailing newline is preserved
5634        revision => '8',                        # integer
5635        date => '2004-02-24T17:01:44.108345Z',  # commit date
5636        author => 'committer name'
5637};
5638
5639
5640# this is generated by generate_diff();
5641@mods = array of diff-index line hashes, each element represents one line
5642        of diff-index output
5643
5644diff-index line ($m hash)
5645{
5646        mode_a => first column of diff-index output, no leading ':',
5647        mode_b => second column of diff-index output,
5648        sha1_b => sha1sum of the final blob,
5649        chg => change type [MCRADT],
5650        file_a => original file name of a file (iff chg is 'C' or 'R')
5651        file_b => new/current file name of a file (any chg)
5652}
5653;
5654
5655# retval of read_url_paths{,_all}();
5656$l_map = {
5657        # repository root url
5658        'https://svn.musicpd.org' => {
5659                # repository path               # GIT_SVN_ID
5660                'mpd/trunk'             =>      'trunk',
5661                'mpd/tags/0.11.5'       =>      'tags/0.11.5',
5662        },
5663}
5664
5665Notes:
5666        I don't trust the each() function on unless I created %hash myself
5667        because the internal iterator may not have started at base.