git-svn.perlon commit Merge branch 'maint-1.6.3' into maint (749086f)
   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        my $author = `$::_authors_prog $orig_author`;
2814        if ($? != 0) {
2815                die "$::_authors_prog failed with exit code $?\n"
2816        }
2817        if ($author =~ /^\s*(.+?)\s*<(.*)>\s*$/) {
2818                my ($name, $email) = ($1, $2);
2819                $email = undef if length $2 == 0;
2820                return [$name, $email];
2821        } else {
2822                die "Author: $orig_author: $::_authors_prog returned "
2823                        . "invalid author format: $author\n";
2824        }
2825}
2826
2827sub check_author {
2828        my ($author) = @_;
2829        if (!defined $author || length $author == 0) {
2830                $author = '(no author)';
2831        }
2832        if (!defined $::users{$author}) {
2833                if (defined $::_authors_prog) {
2834                        $::users{$author} = call_authors_prog($author);
2835                } elsif (defined $::_authors) {
2836                        die "Author: $author not defined in $::_authors file\n";
2837                }
2838        }
2839        $author;
2840}
2841
2842sub make_log_entry {
2843        my ($self, $rev, $parents, $ed) = @_;
2844        my $untracked = $self->get_untracked($ed);
2845
2846        open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
2847        print $un "r$rev\n" or croak $!;
2848        print $un $_, "\n" foreach @$untracked;
2849        my %log_entry = ( parents => $parents || [], revision => $rev,
2850                          log => '');
2851
2852        my $headrev;
2853        my $logged = delete $self->{logged_rev_props};
2854        if (!$logged || $self->{-want_revprops}) {
2855                my $rp = $self->ra->rev_proplist($rev);
2856                foreach (sort keys %$rp) {
2857                        my $v = $rp->{$_};
2858                        if (/^svn:(author|date|log)$/) {
2859                                $log_entry{$1} = $v;
2860                        } elsif ($_ eq 'svm:headrev') {
2861                                $headrev = $v;
2862                        } else {
2863                                print $un "  rev_prop: ", uri_encode($_), ' ',
2864                                          uri_encode($v), "\n";
2865                        }
2866                }
2867        } else {
2868                map { $log_entry{$_} = $logged->{$_} } keys %$logged;
2869        }
2870        close $un or croak $!;
2871
2872        $log_entry{date} = parse_svn_date($log_entry{date});
2873        $log_entry{log} .= "\n";
2874        my $author = $log_entry{author} = check_author($log_entry{author});
2875        my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
2876                                                       : ($author, undef);
2877
2878        my ($commit_name, $commit_email) = ($name, $email);
2879        if ($_use_log_author) {
2880                my $name_field;
2881                if ($log_entry{log} =~ /From:\s+(.*\S)\s*\n/i) {
2882                        $name_field = $1;
2883                } elsif ($log_entry{log} =~ /Signed-off-by:\s+(.*\S)\s*\n/i) {
2884                        $name_field = $1;
2885                }
2886                if (!defined $name_field) {
2887                        if (!defined $email) {
2888                                $email = $name;
2889                        }
2890                } elsif ($name_field =~ /(.*?)\s+<(.*)>/) {
2891                        ($name, $email) = ($1, $2);
2892                } elsif ($name_field =~ /(.*)@/) {
2893                        ($name, $email) = ($1, $name_field);
2894                } else {
2895                        ($name, $email) = ($name_field, $name_field);
2896                }
2897        }
2898        if (defined $headrev && $self->use_svm_props) {
2899                if ($self->rewrite_root) {
2900                        die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
2901                            "options set!\n";
2902                }
2903                my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$}i;
2904                # we don't want "SVM: initializing mirror for junk" ...
2905                return undef if $r == 0;
2906                my $svm = $self->svm;
2907                if ($uuid ne $svm->{uuid}) {
2908                        die "UUID mismatch on SVM path:\n",
2909                            "expected: $svm->{uuid}\n",
2910                            "     got: $uuid\n";
2911                }
2912                my $full_url = $self->full_url;
2913                $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
2914                             die "Failed to replace '$svm->{replace}' with ",
2915                                 "'$svm->{source}' in $full_url\n";
2916                # throw away username for storing in records
2917                remove_username($full_url);
2918                $log_entry{metadata} = "$full_url\@$r $uuid";
2919                $log_entry{svm_revision} = $r;
2920                $email ||= "$author\@$uuid";
2921                $commit_email ||= "$author\@$uuid";
2922        } elsif ($self->use_svnsync_props) {
2923                my $full_url = $self->svnsync->{url};
2924                $full_url .= "/$self->{path}" if length $self->{path};
2925                remove_username($full_url);
2926                my $uuid = $self->svnsync->{uuid};
2927                $log_entry{metadata} = "$full_url\@$rev $uuid";
2928                $email ||= "$author\@$uuid";
2929                $commit_email ||= "$author\@$uuid";
2930        } else {
2931                my $url = $self->metadata_url;
2932                remove_username($url);
2933                $log_entry{metadata} = "$url\@$rev " .
2934                                       $self->ra->get_uuid;
2935                $email ||= "$author\@" . $self->ra->get_uuid;
2936                $commit_email ||= "$author\@" . $self->ra->get_uuid;
2937        }
2938        $log_entry{name} = $name;
2939        $log_entry{email} = $email;
2940        $log_entry{commit_name} = $commit_name;
2941        $log_entry{commit_email} = $commit_email;
2942        \%log_entry;
2943}
2944
2945sub fetch {
2946        my ($self, $min_rev, $max_rev, @parents) = @_;
2947        my ($last_rev, $last_commit) = $self->last_rev_commit;
2948        my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
2949        $self->ra->gs_fetch_loop_common($base, $head, [$self]);
2950}
2951
2952sub set_tree_cb {
2953        my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
2954        $self->{inject_parents} = { $rev => $tree };
2955        $self->fetch(undef, undef);
2956}
2957
2958sub set_tree {
2959        my ($self, $tree) = (shift, shift);
2960        my $log_entry = ::get_commit_entry($tree);
2961        unless ($self->{last_rev}) {
2962                ::fatal("Must have an existing revision to commit");
2963        }
2964        my %ed_opts = ( r => $self->{last_rev},
2965                        log => $log_entry->{log},
2966                        ra => $self->ra,
2967                        tree_a => $self->{last_commit},
2968                        tree_b => $tree,
2969                        editor_cb => sub {
2970                               $self->set_tree_cb($log_entry, $tree, @_) },
2971                        svn_path => $self->{path} );
2972        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
2973                print "No changes\nr$self->{last_rev} = $tree\n";
2974        }
2975}
2976
2977sub rebuild_from_rev_db {
2978        my ($self, $path) = @_;
2979        my $r = -1;
2980        open my $fh, '<', $path or croak "open: $!";
2981        binmode $fh or croak "binmode: $!";
2982        while (<$fh>) {
2983                length($_) == 41 or croak "inconsistent size in ($_) != 41";
2984                chomp($_);
2985                ++$r;
2986                next if $_ eq ('0' x 40);
2987                $self->rev_map_set($r, $_);
2988                print "r$r = $_\n";
2989        }
2990        close $fh or croak "close: $!";
2991        unlink $path or croak "unlink: $!";
2992}
2993
2994sub rebuild {
2995        my ($self) = @_;
2996        my $map_path = $self->map_path;
2997        my $partial = (-e $map_path && ! -z $map_path);
2998        return unless ::verify_ref($self->refname.'^0');
2999        if (!$partial && ($self->use_svm_props || $self->no_metadata)) {
3000                my $rev_db = $self->rev_db_path;
3001                $self->rebuild_from_rev_db($rev_db);
3002                if ($self->use_svm_props) {
3003                        my $svm_rev_db = $self->rev_db_path($self->svm_uuid);
3004                        $self->rebuild_from_rev_db($svm_rev_db);
3005                }
3006                $self->unlink_rev_db_symlink;
3007                return;
3008        }
3009        print "Rebuilding $map_path ...\n" if (!$partial);
3010        my ($base_rev, $head) = ($partial ? $self->rev_map_max_norebuild(1) :
3011                (undef, undef));
3012        my ($log, $ctx) =
3013            command_output_pipe(qw/rev-list --pretty=raw --no-color --reverse/,
3014                                ($head ? "$head.." : "") . $self->refname,
3015                                '--');
3016        my $metadata_url = $self->metadata_url;
3017        remove_username($metadata_url);
3018        my $svn_uuid = $self->ra_uuid;
3019        my $c;
3020        while (<$log>) {
3021                if ( m{^commit ($::sha1)$} ) {
3022                        $c = $1;
3023                        next;
3024                }
3025                next unless s{^\s*(git-svn-id:)}{$1};
3026                my ($url, $rev, $uuid) = ::extract_metadata($_);
3027                remove_username($url);
3028
3029                # ignore merges (from set-tree)
3030                next if (!defined $rev || !$uuid);
3031
3032                # if we merged or otherwise started elsewhere, this is
3033                # how we break out of it
3034                if (($uuid ne $svn_uuid) ||
3035                    ($metadata_url && $url && ($url ne $metadata_url))) {
3036                        next;
3037                }
3038                if ($partial && $head) {
3039                        print "Partial-rebuilding $map_path ...\n";
3040                        print "Currently at $base_rev = $head\n";
3041                        $head = undef;
3042                }
3043
3044                $self->rev_map_set($rev, $c);
3045                print "r$rev = $c\n";
3046        }
3047        command_close_pipe($log, $ctx);
3048        print "Done rebuilding $map_path\n" if (!$partial || !$head);
3049        my $rev_db_path = $self->rev_db_path;
3050        if (-f $self->rev_db_path) {
3051                unlink $self->rev_db_path or croak "unlink: $!";
3052        }
3053        $self->unlink_rev_db_symlink;
3054}
3055
3056# rev_map:
3057# Tie::File seems to be prone to offset errors if revisions get sparse,
3058# it's not that fast, either.  Tie::File is also not in Perl 5.6.  So
3059# one of my favorite modules is out :<  Next up would be one of the DBM
3060# modules, but I'm not sure which is most portable...
3061#
3062# This is the replacement for the rev_db format, which was too big
3063# and inefficient for large repositories with a lot of sparse history
3064# (mainly tags)
3065#
3066# The format is this:
3067#   - 24 bytes for every record,
3068#     * 4 bytes for the integer representing an SVN revision number
3069#     * 20 bytes representing the sha1 of a git commit
3070#   - No empty padding records like the old format
3071#     (except the last record, which can be overwritten)
3072#   - new records are written append-only since SVN revision numbers
3073#     increase monotonically
3074#   - lookups on SVN revision number are done via a binary search
3075#   - Piping the file to xxd -c24 is a good way of dumping it for
3076#     viewing or editing (piped back through xxd -r), should the need
3077#     ever arise.
3078#   - The last record can be padding revision with an all-zero sha1
3079#     This is used to optimize fetch performance when using multiple
3080#     "fetch" directives in .git/config
3081#
3082# These files are disposable unless noMetadata or useSvmProps is set
3083
3084sub _rev_map_set {
3085        my ($fh, $rev, $commit) = @_;
3086
3087        binmode $fh or croak "binmode: $!";
3088        my $size = (stat($fh))[7];
3089        ($size % 24) == 0 or croak "inconsistent size: $size";
3090
3091        my $wr_offset = 0;
3092        if ($size > 0) {
3093                sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3094                my $read = sysread($fh, my $buf, 24) or croak "read: $!";
3095                $read == 24 or croak "read only $read bytes (!= 24)";
3096                my ($last_rev, $last_commit) = unpack(rev_map_fmt, $buf);
3097                if ($last_commit eq ('0' x40)) {
3098                        if ($size >= 48) {
3099                                sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3100                                $read = sysread($fh, $buf, 24) or
3101                                    croak "read: $!";
3102                                $read == 24 or
3103                                    croak "read only $read bytes (!= 24)";
3104                                ($last_rev, $last_commit) =
3105                                    unpack(rev_map_fmt, $buf);
3106                                if ($last_commit eq ('0' x40)) {
3107                                        croak "inconsistent .rev_map\n";
3108                                }
3109                        }
3110                        if ($last_rev >= $rev) {
3111                                croak "last_rev is higher!: $last_rev >= $rev";
3112                        }
3113                        $wr_offset = -24;
3114                }
3115        }
3116        sysseek($fh, $wr_offset, SEEK_END) or croak "seek: $!";
3117        syswrite($fh, pack(rev_map_fmt, $rev, $commit), 24) == 24 or
3118          croak "write: $!";
3119}
3120
3121sub _rev_map_reset {
3122        my ($fh, $rev, $commit) = @_;
3123        my $c = _rev_map_get($fh, $rev);
3124        $c eq $commit or die "_rev_map_reset(@_) commit $c does not match!\n";
3125        my $offset = sysseek($fh, 0, SEEK_CUR) or croak "seek: $!";
3126        truncate $fh, $offset or croak "truncate: $!";
3127}
3128
3129sub mkfile {
3130        my ($path) = @_;
3131        unless (-e $path) {
3132                my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
3133                mkpath([$dir]) unless -d $dir;
3134                open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
3135                close $fh or die "Couldn't close (create) $path: $!\n";
3136        }
3137}
3138
3139sub rev_map_set {
3140        my ($self, $rev, $commit, $update_ref, $uuid) = @_;
3141        length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
3142        my $db = $self->map_path($uuid);
3143        my $db_lock = "$db.lock";
3144        my $sig;
3145        $update_ref ||= 0;
3146        if ($update_ref) {
3147                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3148                            $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
3149        }
3150        mkfile($db);
3151
3152        $LOCKFILES{$db_lock} = 1;
3153        my $sync;
3154        # both of these options make our .rev_db file very, very important
3155        # and we can't afford to lose it because rebuild() won't work
3156        if ($self->use_svm_props || $self->no_metadata) {
3157                $sync = 1;
3158                copy($db, $db_lock) or die "rev_map_set(@_): ",
3159                                           "Failed to copy: ",
3160                                           "$db => $db_lock ($!)\n";
3161        } else {
3162                rename $db, $db_lock or die "rev_map_set(@_): ",
3163                                            "Failed to rename: ",
3164                                            "$db => $db_lock ($!)\n";
3165        }
3166
3167        sysopen(my $fh, $db_lock, O_RDWR | O_CREAT)
3168             or croak "Couldn't open $db_lock: $!\n";
3169        $update_ref eq 'reset' ? _rev_map_reset($fh, $rev, $commit) :
3170                                 _rev_map_set($fh, $rev, $commit);
3171        if ($sync) {
3172                $fh->flush or die "Couldn't flush $db_lock: $!\n";
3173                $fh->sync or die "Couldn't sync $db_lock: $!\n";
3174        }
3175        close $fh or croak $!;
3176        if ($update_ref) {
3177                $_head = $self;
3178                my $note = "";
3179                $note = " ($update_ref)" if ($update_ref !~ /^\d*$/);
3180                command_noisy('update-ref', '-m', "r$rev$note",
3181                              $self->refname, $commit);
3182        }
3183        rename $db_lock, $db or die "rev_map_set(@_): ", "Failed to rename: ",
3184                                    "$db_lock => $db ($!)\n";
3185        delete $LOCKFILES{$db_lock};
3186        if ($update_ref) {
3187                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3188                            $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
3189                kill $sig, $$ if defined $sig;
3190        }
3191}
3192
3193# If want_commit, this will return an array of (rev, commit) where
3194# commit _must_ be a valid commit in the archive.
3195# Otherwise, it'll return the max revision (whether or not the
3196# commit is valid or just a 0x40 placeholder).
3197sub rev_map_max {
3198        my ($self, $want_commit) = @_;
3199        $self->rebuild;
3200        my ($r, $c) = $self->rev_map_max_norebuild($want_commit);
3201        $want_commit ? ($r, $c) : $r;
3202}
3203
3204sub rev_map_max_norebuild {
3205        my ($self, $want_commit) = @_;
3206        my $map_path = $self->map_path;
3207        stat $map_path or return $want_commit ? (0, undef) : 0;
3208        sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3209        binmode $fh or croak "binmode: $!";
3210        my $size = (stat($fh))[7];
3211        ($size % 24) == 0 or croak "inconsistent size: $size";
3212
3213        if ($size == 0) {
3214                close $fh or croak "close: $!";
3215                return $want_commit ? (0, undef) : 0;
3216        }
3217
3218        sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3219        sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3220        my ($r, $c) = unpack(rev_map_fmt, $buf);
3221        if ($want_commit && $c eq ('0' x40)) {
3222                if ($size < 48) {
3223                        return $want_commit ? (0, undef) : 0;
3224                }
3225                sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3226                sysread($fh, $buf, 24) == 24 or croak "read: $!";
3227                ($r, $c) = unpack(rev_map_fmt, $buf);
3228                if ($c eq ('0'x40)) {
3229                        croak "Penultimate record is all-zeroes in $map_path";
3230                }
3231        }
3232        close $fh or croak "close: $!";
3233        $want_commit ? ($r, $c) : $r;
3234}
3235
3236sub rev_map_get {
3237        my ($self, $rev, $uuid) = @_;
3238        my $map_path = $self->map_path($uuid);
3239        return undef unless -e $map_path;
3240
3241        sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3242        my $c = _rev_map_get($fh, $rev);
3243        close($fh) or croak "close: $!";
3244        $c
3245}
3246
3247sub _rev_map_get {
3248        my ($fh, $rev) = @_;
3249
3250        binmode $fh or croak "binmode: $!";
3251        my $size = (stat($fh))[7];
3252        ($size % 24) == 0 or croak "inconsistent size: $size";
3253
3254        if ($size == 0) {
3255                return undef;
3256        }
3257
3258        my ($l, $u) = (0, $size - 24);
3259        my ($r, $c, $buf);
3260
3261        while ($l <= $u) {
3262                my $i = int(($l/24 + $u/24) / 2) * 24;
3263                sysseek($fh, $i, SEEK_SET) or croak "seek: $!";
3264                sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3265                my ($r, $c) = unpack('NH40', $buf);
3266
3267                if ($r < $rev) {
3268                        $l = $i + 24;
3269                } elsif ($r > $rev) {
3270                        $u = $i - 24;
3271                } else { # $r == $rev
3272                        return $c eq ('0' x 40) ? undef : $c;
3273                }
3274        }
3275        undef;
3276}
3277
3278# Finds the first svn revision that exists on (if $eq_ok is true) or
3279# before $rev for the current branch.  It will not search any lower
3280# than $min_rev.  Returns the git commit hash and svn revision number
3281# if found, else (undef, undef).
3282sub find_rev_before {
3283        my ($self, $rev, $eq_ok, $min_rev) = @_;
3284        --$rev unless $eq_ok;
3285        $min_rev ||= 1;
3286        my $max_rev = $self->rev_map_max;
3287        $rev = $max_rev if ($rev > $max_rev);
3288        while ($rev >= $min_rev) {
3289                if (my $c = $self->rev_map_get($rev)) {
3290                        return ($rev, $c);
3291                }
3292                --$rev;
3293        }
3294        return (undef, undef);
3295}
3296
3297# Finds the first svn revision that exists on (if $eq_ok is true) or
3298# after $rev for the current branch.  It will not search any higher
3299# than $max_rev.  Returns the git commit hash and svn revision number
3300# if found, else (undef, undef).
3301sub find_rev_after {
3302        my ($self, $rev, $eq_ok, $max_rev) = @_;
3303        ++$rev unless $eq_ok;
3304        $max_rev ||= $self->rev_map_max;
3305        while ($rev <= $max_rev) {
3306                if (my $c = $self->rev_map_get($rev)) {
3307                        return ($rev, $c);
3308                }
3309                ++$rev;
3310        }
3311        return (undef, undef);
3312}
3313
3314sub _new {
3315        my ($class, $repo_id, $ref_id, $path) = @_;
3316        unless (defined $repo_id && length $repo_id) {
3317                $repo_id = $Git::SVN::default_repo_id;
3318        }
3319        unless (defined $ref_id && length $ref_id) {
3320                $_[2] = $ref_id = $Git::SVN::default_ref_id;
3321        }
3322        $_[1] = $repo_id;
3323        my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
3324        $_[3] = $path = '' unless (defined $path);
3325        mkpath(["$ENV{GIT_DIR}/svn"]);
3326        bless {
3327                ref_id => $ref_id, dir => $dir, index => "$dir/index",
3328                path => $path, config => "$ENV{GIT_DIR}/svn/config",
3329                map_root => "$dir/.rev_map", repo_id => $repo_id }, $class;
3330}
3331
3332# for read-only access of old .rev_db formats
3333sub unlink_rev_db_symlink {
3334        my ($self) = @_;
3335        my $link = $self->rev_db_path;
3336        $link =~ s/\.[\w-]+$// or croak "missing UUID at the end of $link";
3337        if (-l $link) {
3338                unlink $link or croak "unlink: $link failed!";
3339        }
3340}
3341
3342sub rev_db_path {
3343        my ($self, $uuid) = @_;
3344        my $db_path = $self->map_path($uuid);
3345        $db_path =~ s{/\.rev_map\.}{/\.rev_db\.}
3346            or croak "map_path: $db_path does not contain '/.rev_map.' !";
3347        $db_path;
3348}
3349
3350# the new replacement for .rev_db
3351sub map_path {
3352        my ($self, $uuid) = @_;
3353        $uuid ||= $self->ra_uuid;
3354        "$self->{map_root}.$uuid";
3355}
3356
3357sub uri_encode {
3358        my ($f) = @_;
3359        $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
3360        $f
3361}
3362
3363sub remove_username {
3364        $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
3365}
3366
3367package Git::SVN::Prompt;
3368use strict;
3369use warnings;
3370require SVN::Core;
3371use vars qw/$_no_auth_cache $_username/;
3372
3373sub simple {
3374        my ($cred, $realm, $default_username, $may_save, $pool) = @_;
3375        $may_save = undef if $_no_auth_cache;
3376        $default_username = $_username if defined $_username;
3377        if (defined $default_username && length $default_username) {
3378                if (defined $realm && length $realm) {
3379                        print STDERR "Authentication realm: $realm\n";
3380                        STDERR->flush;
3381                }
3382                $cred->username($default_username);
3383        } else {
3384                username($cred, $realm, $may_save, $pool);
3385        }
3386        $cred->password(_read_password("Password for '" .
3387                                       $cred->username . "': ", $realm));
3388        $cred->may_save($may_save);
3389        $SVN::_Core::SVN_NO_ERROR;
3390}
3391
3392sub ssl_server_trust {
3393        my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
3394        $may_save = undef if $_no_auth_cache;
3395        print STDERR "Error validating server certificate for '$realm':\n";
3396        {
3397                no warnings 'once';
3398                # All variables SVN::Auth::SSL::* are used only once,
3399                # so we're shutting up Perl warnings about this.
3400                if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
3401                        print STDERR " - The certificate is not issued ",
3402                            "by a trusted authority. Use the\n",
3403                            "   fingerprint to validate ",
3404                            "the certificate manually!\n";
3405                }
3406                if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
3407                        print STDERR " - The certificate hostname ",
3408                            "does not match.\n";
3409                }
3410                if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
3411                        print STDERR " - The certificate is not yet valid.\n";
3412                }
3413                if ($failures & $SVN::Auth::SSL::EXPIRED) {
3414                        print STDERR " - The certificate has expired.\n";
3415                }
3416                if ($failures & $SVN::Auth::SSL::OTHER) {
3417                        print STDERR " - The certificate has ",
3418                            "an unknown error.\n";
3419                }
3420        } # no warnings 'once'
3421        printf STDERR
3422                "Certificate information:\n".
3423                " - Hostname: %s\n".
3424                " - Valid: from %s until %s\n".
3425                " - Issuer: %s\n".
3426                " - Fingerprint: %s\n",
3427                map $cert_info->$_, qw(hostname valid_from valid_until
3428                                       issuer_dname fingerprint);
3429        my $choice;
3430prompt:
3431        print STDERR $may_save ?
3432              "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
3433              "(R)eject or accept (t)emporarily? ";
3434        STDERR->flush;
3435        $choice = lc(substr(<STDIN> || 'R', 0, 1));
3436        if ($choice =~ /^t$/i) {
3437                $cred->may_save(undef);
3438        } elsif ($choice =~ /^r$/i) {
3439                return -1;
3440        } elsif ($may_save && $choice =~ /^p$/i) {
3441                $cred->may_save($may_save);
3442        } else {
3443                goto prompt;
3444        }
3445        $cred->accepted_failures($failures);
3446        $SVN::_Core::SVN_NO_ERROR;
3447}
3448
3449sub ssl_client_cert {
3450        my ($cred, $realm, $may_save, $pool) = @_;
3451        $may_save = undef if $_no_auth_cache;
3452        print STDERR "Client certificate filename: ";
3453        STDERR->flush;
3454        chomp(my $filename = <STDIN>);
3455        $cred->cert_file($filename);
3456        $cred->may_save($may_save);
3457        $SVN::_Core::SVN_NO_ERROR;
3458}
3459
3460sub ssl_client_cert_pw {
3461        my ($cred, $realm, $may_save, $pool) = @_;
3462        $may_save = undef if $_no_auth_cache;
3463        $cred->password(_read_password("Password: ", $realm));
3464        $cred->may_save($may_save);
3465        $SVN::_Core::SVN_NO_ERROR;
3466}
3467
3468sub username {
3469        my ($cred, $realm, $may_save, $pool) = @_;
3470        $may_save = undef if $_no_auth_cache;
3471        if (defined $realm && length $realm) {
3472                print STDERR "Authentication realm: $realm\n";
3473        }
3474        my $username;
3475        if (defined $_username) {
3476                $username = $_username;
3477        } else {
3478                print STDERR "Username: ";
3479                STDERR->flush;
3480                chomp($username = <STDIN>);
3481        }
3482        $cred->username($username);
3483        $cred->may_save($may_save);
3484        $SVN::_Core::SVN_NO_ERROR;
3485}
3486
3487sub _read_password {
3488        my ($prompt, $realm) = @_;
3489        print STDERR $prompt;
3490        STDERR->flush;
3491        require Term::ReadKey;
3492        Term::ReadKey::ReadMode('noecho');
3493        my $password = '';
3494        while (defined(my $key = Term::ReadKey::ReadKey(0))) {
3495                last if $key =~ /[\012\015]/; # \n\r
3496                $password .= $key;
3497        }
3498        Term::ReadKey::ReadMode('restore');
3499        print STDERR "\n";
3500        STDERR->flush;
3501        $password;
3502}
3503
3504package SVN::Git::Fetcher;
3505use vars qw/@ISA/;
3506use strict;
3507use warnings;
3508use Carp qw/croak/;
3509use File::Temp qw/tempfile/;
3510use IO::File qw//;
3511use vars qw/$_ignore_regex/;
3512
3513# file baton members: path, mode_a, mode_b, pool, fh, blob, base
3514sub new {
3515        my ($class, $git_svn, $switch_path) = @_;
3516        my $self = SVN::Delta::Editor->new;
3517        bless $self, $class;
3518        if (exists $git_svn->{last_commit}) {
3519                $self->{c} = $git_svn->{last_commit};
3520                $self->{empty_symlinks} =
3521                                  _mark_empty_symlinks($git_svn, $switch_path);
3522        }
3523        $self->{ignore_regex} = eval { command_oneline('config', '--get',
3524                             "svn-remote.$git_svn->{repo_id}.ignore-paths") };
3525        $self->{empty} = {};
3526        $self->{dir_prop} = {};
3527        $self->{file_prop} = {};
3528        $self->{absent_dir} = {};
3529        $self->{absent_file} = {};
3530        $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
3531        $self;
3532}
3533
3534# this uses the Ra object, so it must be called before do_{switch,update},
3535# not inside them (when the Git::SVN::Fetcher object is passed) to
3536# do_{switch,update}
3537sub _mark_empty_symlinks {
3538        my ($git_svn, $switch_path) = @_;
3539        my $bool = Git::config_bool('svn.brokenSymlinkWorkaround');
3540        return {} if (!defined($bool)) || (defined($bool) && ! $bool);
3541
3542        my %ret;
3543        my ($rev, $cmt) = $git_svn->last_rev_commit;
3544        return {} unless ($rev && $cmt);
3545
3546        # allow the warning to be printed for each revision we fetch to
3547        # ensure the user sees it.  The user can also disable the workaround
3548        # on the repository even while git svn is running and the next
3549        # revision fetched will skip this expensive function.
3550        my $printed_warning;
3551        chomp(my $empty_blob = `git hash-object -t blob --stdin < /dev/null`);
3552        my ($ls, $ctx) = command_output_pipe(qw/ls-tree -r -z/, $cmt);
3553        local $/ = "\0";
3554        my $pfx = defined($switch_path) ? $switch_path : $git_svn->{path};
3555        $pfx .= '/' if length($pfx);
3556        while (<$ls>) {
3557                chomp;
3558                s/\A100644 blob $empty_blob\t//o or next;
3559                unless ($printed_warning) {
3560                        print STDERR "Scanning for empty symlinks, ",
3561                                     "this may take a while if you have ",
3562                                     "many empty files\n",
3563                                     "You may disable this with `",
3564                                     "git config svn.brokenSymlinkWorkaround ",
3565                                     "false'.\n",
3566                                     "This may be done in a different ",
3567                                     "terminal without restarting ",
3568                                     "git svn\n";
3569                        $printed_warning = 1;
3570                }
3571                my $path = $_;
3572                my (undef, $props) =
3573                               $git_svn->ra->get_file($pfx.$path, $rev, undef);
3574                if ($props->{'svn:special'}) {
3575                        $ret{$path} = 1;
3576                }
3577        }
3578        command_close_pipe($ls, $ctx);
3579        \%ret;
3580}
3581
3582# returns true if a given path is inside a ".git" directory
3583sub in_dot_git {
3584        $_[0] =~ m{(?:^|/)\.git(?:/|$)};
3585}
3586
3587# return value: 0 -- don't ignore, 1 -- ignore
3588sub is_path_ignored {
3589        my ($self, $path) = @_;
3590        return 1 if in_dot_git($path);
3591        return 1 if defined($self->{ignore_regex}) &&
3592                    $path =~ m!$self->{ignore_regex}!;
3593        return 0 unless defined($_ignore_regex);
3594        return 1 if $path =~ m!$_ignore_regex!o;
3595        return 0;
3596}
3597
3598sub set_path_strip {
3599        my ($self, $path) = @_;
3600        $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
3601}
3602
3603sub open_root {
3604        { path => '' };
3605}
3606
3607sub open_directory {
3608        my ($self, $path, $pb, $rev) = @_;
3609        { path => $path };
3610}
3611
3612sub git_path {
3613        my ($self, $path) = @_;
3614        if ($self->{path_strip}) {
3615                $path =~ s!$self->{path_strip}!! or
3616                  die "Failed to strip path '$path' ($self->{path_strip})\n";
3617        }
3618        $path;
3619}
3620
3621sub delete_entry {
3622        my ($self, $path, $rev, $pb) = @_;
3623        return undef if $self->is_path_ignored($path);
3624
3625        my $gpath = $self->git_path($path);
3626        return undef if ($gpath eq '');
3627
3628        # remove entire directories.
3629        my ($tree) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
3630                         =~ /\A040000 tree ([a-f\d]{40})\t\Q$gpath\E\0/);
3631        if ($tree) {
3632                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3633                                                     -r --name-only -z/,
3634                                                     $tree);
3635                local $/ = "\0";
3636                while (<$ls>) {
3637                        chomp;
3638                        my $rmpath = "$gpath/$_";
3639                        $self->{gii}->remove($rmpath);
3640                        print "\tD\t$rmpath\n" unless $::_q;
3641                }
3642                print "\tD\t$gpath/\n" unless $::_q;
3643                command_close_pipe($ls, $ctx);
3644                $self->{empty}->{$path} = 0
3645        } else {
3646                $self->{gii}->remove($gpath);
3647                print "\tD\t$gpath\n" unless $::_q;
3648        }
3649        undef;
3650}
3651
3652sub open_file {
3653        my ($self, $path, $pb, $rev) = @_;
3654        my ($mode, $blob);
3655
3656        goto out if $self->is_path_ignored($path);
3657
3658        my $gpath = $self->git_path($path);
3659        ($mode, $blob) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
3660                             =~ /\A(\d{6}) blob ([a-f\d]{40})\t\Q$gpath\E\0/);
3661        unless (defined $mode && defined $blob) {
3662                die "$path was not found in commit $self->{c} (r$rev)\n";
3663        }
3664        if ($mode eq '100644' && $self->{empty_symlinks}->{$path}) {
3665                $mode = '120000';
3666        }
3667out:
3668        { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
3669          pool => SVN::Pool->new, action => 'M' };
3670}
3671
3672sub add_file {
3673        my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
3674        my $mode;
3675
3676        if (!$self->is_path_ignored($path)) {
3677                my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3678                delete $self->{empty}->{$dir};
3679                $mode = '100644';
3680        }
3681        { path => $path, mode_a => $mode, mode_b => $mode,
3682          pool => SVN::Pool->new, action => 'A' };
3683}
3684
3685sub add_directory {
3686        my ($self, $path, $cp_path, $cp_rev) = @_;
3687        goto out if $self->is_path_ignored($path);
3688        my $gpath = $self->git_path($path);
3689        if ($gpath eq '') {
3690                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3691                                                     -r --name-only -z/,
3692                                                     $self->{c});
3693                local $/ = "\0";
3694                while (<$ls>) {
3695                        chomp;
3696                        $self->{gii}->remove($_);
3697                        print "\tD\t$_\n" unless $::_q;
3698                }
3699                command_close_pipe($ls, $ctx);
3700                $self->{empty}->{$path} = 0;
3701        }
3702        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3703        delete $self->{empty}->{$dir};
3704        $self->{empty}->{$path} = 1;
3705out:
3706        { path => $path };
3707}
3708
3709sub change_dir_prop {
3710        my ($self, $db, $prop, $value) = @_;
3711        return undef if $self->is_path_ignored($db->{path});
3712        $self->{dir_prop}->{$db->{path}} ||= {};
3713        $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
3714        undef;
3715}
3716
3717sub absent_directory {
3718        my ($self, $path, $pb) = @_;
3719        return undef if $self->is_path_ignored($path);
3720        $self->{absent_dir}->{$pb->{path}} ||= [];
3721        push @{$self->{absent_dir}->{$pb->{path}}}, $path;
3722        undef;
3723}
3724
3725sub absent_file {
3726        my ($self, $path, $pb) = @_;
3727        return undef if $self->is_path_ignored($path);
3728        $self->{absent_file}->{$pb->{path}} ||= [];
3729        push @{$self->{absent_file}->{$pb->{path}}}, $path;
3730        undef;
3731}
3732
3733sub change_file_prop {
3734        my ($self, $fb, $prop, $value) = @_;
3735        return undef if $self->is_path_ignored($fb->{path});
3736        if ($prop eq 'svn:executable') {
3737                if ($fb->{mode_b} != 120000) {
3738                        $fb->{mode_b} = defined $value ? 100755 : 100644;
3739                }
3740        } elsif ($prop eq 'svn:special') {
3741                $fb->{mode_b} = defined $value ? 120000 : 100644;
3742        } else {
3743                $self->{file_prop}->{$fb->{path}} ||= {};
3744                $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
3745        }
3746        undef;
3747}
3748
3749sub apply_textdelta {
3750        my ($self, $fb, $exp) = @_;
3751        return undef if $self->is_path_ignored($fb->{path});
3752        my $fh = $::_repository->temp_acquire('svn_delta');
3753        # $fh gets auto-closed() by SVN::TxDelta::apply(),
3754        # (but $base does not,) so dup() it for reading in close_file
3755        open my $dup, '<&', $fh or croak $!;
3756        my $base = $::_repository->temp_acquire('git_blob');
3757
3758        if ($fb->{blob}) {
3759                my ($base_is_link, $size);
3760
3761                if ($fb->{mode_a} eq '120000' &&
3762                    ! $self->{empty_symlinks}->{$fb->{path}}) {
3763                        print $base 'link ' or die "print $!\n";
3764                        $base_is_link = 1;
3765                }
3766        retry:
3767                $size = $::_repository->cat_blob($fb->{blob}, $base);
3768                die "Failed to read object $fb->{blob}" if ($size < 0);
3769
3770                if (defined $exp) {
3771                        seek $base, 0, 0 or croak $!;
3772                        my $got = ::md5sum($base);
3773                        if ($got ne $exp) {
3774                                my $err = "Checksum mismatch: ".
3775                                       "$fb->{path} $fb->{blob}\n" .
3776                                       "expected: $exp\n" .
3777                                       "     got: $got\n";
3778                                if ($base_is_link) {
3779                                        warn $err,
3780                                             "Retrying... (possibly ",
3781                                             "a bad symlink from SVN)\n";
3782                                        $::_repository->temp_reset($base);
3783                                        $base_is_link = 0;
3784                                        goto retry;
3785                                }
3786                                die $err;
3787                        }
3788                }
3789        }
3790        seek $base, 0, 0 or croak $!;
3791        $fb->{fh} = $fh;
3792        $fb->{base} = $base;
3793        [ SVN::TxDelta::apply($base, $dup, undef, $fb->{path}, $fb->{pool}) ];
3794}
3795
3796sub close_file {
3797        my ($self, $fb, $exp) = @_;
3798        return undef if $self->is_path_ignored($fb->{path});
3799
3800        my $hash;
3801        my $path = $self->git_path($fb->{path});
3802        if (my $fh = $fb->{fh}) {
3803                if (defined $exp) {
3804                        seek($fh, 0, 0) or croak $!;
3805                        my $got = ::md5sum($fh);
3806                        if ($got ne $exp) {
3807                                die "Checksum mismatch: $path\n",
3808                                    "expected: $exp\n    got: $got\n";
3809                        }
3810                }
3811                if ($fb->{mode_b} == 120000) {
3812                        sysseek($fh, 0, 0) or croak $!;
3813                        my $rd = sysread($fh, my $buf, 5);
3814
3815                        if (!defined $rd) {
3816                                croak "sysread: $!\n";
3817                        } elsif ($rd == 0) {
3818                                warn "$path has mode 120000",
3819                                     " but it points to nothing\n",
3820                                     "converting to an empty file with mode",
3821                                     " 100644\n";
3822                                $fb->{mode_b} = '100644';
3823                        } elsif ($buf ne 'link ') {
3824                                warn "$path has mode 120000",
3825                                     " but is not a link\n";
3826                        } else {
3827                                my $tmp_fh = $::_repository->temp_acquire(
3828                                        'svn_hash');
3829                                my $res;
3830                                while ($res = sysread($fh, my $str, 1024)) {
3831                                        my $out = syswrite($tmp_fh, $str, $res);
3832                                        defined($out) && $out == $res
3833                                                or croak("write ",
3834                                                        Git::temp_path($tmp_fh),
3835                                                        ": $!\n");
3836                                }
3837                                defined $res or croak $!;
3838
3839                                ($fh, $tmp_fh) = ($tmp_fh, $fh);
3840                                Git::temp_release($tmp_fh, 1);
3841                        }
3842                }
3843
3844                $hash = $::_repository->hash_and_insert_object(
3845                                Git::temp_path($fh));
3846                $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
3847
3848                Git::temp_release($fb->{base}, 1);
3849                Git::temp_release($fh, 1);
3850        } else {
3851                $hash = $fb->{blob} or die "no blob information\n";
3852        }
3853        $fb->{pool}->clear;
3854        $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
3855        print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
3856        undef;
3857}
3858
3859sub abort_edit {
3860        my $self = shift;
3861        $self->{nr} = $self->{gii}->{nr};
3862        delete $self->{gii};
3863        $self->SUPER::abort_edit(@_);
3864}
3865
3866sub close_edit {
3867        my $self = shift;
3868        $self->{git_commit_ok} = 1;
3869        $self->{nr} = $self->{gii}->{nr};
3870        delete $self->{gii};
3871        $self->SUPER::close_edit(@_);
3872}
3873
3874package SVN::Git::Editor;
3875use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
3876use strict;
3877use warnings;
3878use Carp qw/croak/;
3879use IO::File;
3880
3881sub new {
3882        my ($class, $opts) = @_;
3883        foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
3884                die "$_ required!\n" unless (defined $opts->{$_});
3885        }
3886
3887        my $pool = SVN::Pool->new;
3888        my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
3889        my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
3890                                     $opts->{r}, $mods);
3891
3892        # $opts->{ra} functions should not be used after this:
3893        my @ce  = $opts->{ra}->get_commit_editor($opts->{log},
3894                                                $opts->{editor_cb}, $pool);
3895        my $self = SVN::Delta::Editor->new(@ce, $pool);
3896        bless $self, $class;
3897        foreach (qw/svn_path r tree_a tree_b/) {
3898                $self->{$_} = $opts->{$_};
3899        }
3900        $self->{url} = $opts->{ra}->{url};
3901        $self->{mods} = $mods;
3902        $self->{types} = $types;
3903        $self->{pool} = $pool;
3904        $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
3905        $self->{rm} = { };
3906        $self->{path_prefix} = length $self->{svn_path} ?
3907                               "$self->{svn_path}/" : '';
3908        $self->{config} = $opts->{config};
3909        return $self;
3910}
3911
3912sub generate_diff {
3913        my ($tree_a, $tree_b) = @_;
3914        my @diff_tree = qw(diff-tree -z -r);
3915        if ($_cp_similarity) {
3916                push @diff_tree, "-C$_cp_similarity";
3917        } else {
3918                push @diff_tree, '-C';
3919        }
3920        push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
3921        push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
3922        push @diff_tree, $tree_a, $tree_b;
3923        my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
3924        local $/ = "\0";
3925        my $state = 'meta';
3926        my @mods;
3927        while (<$diff_fh>) {
3928                chomp $_; # this gets rid of the trailing "\0"
3929                if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
3930                                        ($::sha1)\s($::sha1)\s
3931                                        ([MTCRAD])\d*$/xo) {
3932                        push @mods, {   mode_a => $1, mode_b => $2,
3933                                        sha1_a => $3, sha1_b => $4,
3934                                        chg => $5 };
3935                        if ($5 =~ /^(?:C|R)$/) {
3936                                $state = 'file_a';
3937                        } else {
3938                                $state = 'file_b';
3939                        }
3940                } elsif ($state eq 'file_a') {
3941                        my $x = $mods[$#mods] or croak "Empty array\n";
3942                        if ($x->{chg} !~ /^(?:C|R)$/) {
3943                                croak "Error parsing $_, $x->{chg}\n";
3944                        }
3945                        $x->{file_a} = $_;
3946                        $state = 'file_b';
3947                } elsif ($state eq 'file_b') {
3948                        my $x = $mods[$#mods] or croak "Empty array\n";
3949                        if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
3950                                croak "Error parsing $_, $x->{chg}\n";
3951                        }
3952                        if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
3953                                croak "Error parsing $_, $x->{chg}\n";
3954                        }
3955                        $x->{file_b} = $_;
3956                        $state = 'meta';
3957                } else {
3958                        croak "Error parsing $_\n";
3959                }
3960        }
3961        command_close_pipe($diff_fh, $ctx);
3962        \@mods;
3963}
3964
3965sub check_diff_paths {
3966        my ($ra, $pfx, $rev, $mods) = @_;
3967        my %types;
3968        $pfx .= '/' if length $pfx;
3969
3970        sub type_diff_paths {
3971                my ($ra, $types, $path, $rev) = @_;
3972                my @p = split m#/+#, $path;
3973                my $c = shift @p;
3974                unless (defined $types->{$c}) {
3975                        $types->{$c} = $ra->check_path($c, $rev);
3976                }
3977                while (@p) {
3978                        $c .= '/' . shift @p;
3979                        next if defined $types->{$c};
3980                        $types->{$c} = $ra->check_path($c, $rev);
3981                }
3982        }
3983
3984        foreach my $m (@$mods) {
3985                foreach my $f (qw/file_a file_b/) {
3986                        next unless defined $m->{$f};
3987                        my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
3988                        if (length $pfx.$dir && ! defined $types{$dir}) {
3989                                type_diff_paths($ra, \%types, $pfx.$dir, $rev);
3990                        }
3991                }
3992        }
3993        \%types;
3994}
3995
3996sub split_path {
3997        return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
3998}
3999
4000sub repo_path {
4001        my ($self, $path) = @_;
4002        $self->{path_prefix}.(defined $path ? $path : '');
4003}
4004
4005sub url_path {
4006        my ($self, $path) = @_;
4007        if ($self->{url} =~ m#^https?://#) {
4008                $path =~ s!([^~a-zA-Z0-9_./-])!uc sprintf("%%%02x",ord($1))!eg;
4009        }
4010        $self->{url} . '/' . $self->repo_path($path);
4011}
4012
4013sub rmdirs {
4014        my ($self) = @_;
4015        my $rm = $self->{rm};
4016        delete $rm->{''}; # we never delete the url we're tracking
4017        return unless %$rm;
4018
4019        foreach (keys %$rm) {
4020                my @d = split m#/#, $_;
4021                my $c = shift @d;
4022                $rm->{$c} = 1;
4023                while (@d) {
4024                        $c .= '/' . shift @d;
4025                        $rm->{$c} = 1;
4026                }
4027        }
4028        delete $rm->{$self->{svn_path}};
4029        delete $rm->{''}; # we never delete the url we're tracking
4030        return unless %$rm;
4031
4032        my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
4033                                             $self->{tree_b});
4034        local $/ = "\0";
4035        while (<$fh>) {
4036                chomp;
4037                my @dn = split m#/#, $_;
4038                while (pop @dn) {
4039                        delete $rm->{join '/', @dn};
4040                }
4041                unless (%$rm) {
4042                        close $fh;
4043                        return;
4044                }
4045        }
4046        command_close_pipe($fh, $ctx);
4047
4048        my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
4049        foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
4050                $self->close_directory($bat->{$d}, $p);
4051                my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
4052                print "\tD+\t$d/\n" unless $::_q;
4053                $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
4054                delete $bat->{$d};
4055        }
4056}
4057
4058sub open_or_add_dir {
4059        my ($self, $full_path, $baton) = @_;
4060        my $t = $self->{types}->{$full_path};
4061        if (!defined $t) {
4062                die "$full_path not known in r$self->{r} or we have a bug!\n";
4063        }
4064        {
4065                no warnings 'once';
4066                # SVN::Node::none and SVN::Node::file are used only once,
4067                # so we're shutting up Perl's warnings about them.
4068                if ($t == $SVN::Node::none) {
4069                        return $self->add_directory($full_path, $baton,
4070                            undef, -1, $self->{pool});
4071                } elsif ($t == $SVN::Node::dir) {
4072                        return $self->open_directory($full_path, $baton,
4073                            $self->{r}, $self->{pool});
4074                } # no warnings 'once'
4075                print STDERR "$full_path already exists in repository at ",
4076                    "r$self->{r} and it is not a directory (",
4077                    ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
4078        } # no warnings 'once'
4079        exit 1;
4080}
4081
4082sub ensure_path {
4083        my ($self, $path) = @_;
4084        my $bat = $self->{bat};
4085        my $repo_path = $self->repo_path($path);
4086        return $bat->{''} unless (length $repo_path);
4087        my @p = split m#/+#, $repo_path;
4088        my $c = shift @p;
4089        $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
4090        while (@p) {
4091                my $c0 = $c;
4092                $c .= '/' . shift @p;
4093                $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
4094        }
4095        return $bat->{$c};
4096}
4097
4098# Subroutine to convert a globbing pattern to a regular expression.
4099# From perl cookbook.
4100sub glob2pat {
4101        my $globstr = shift;
4102        my %patmap = ('*' => '.*', '?' => '.', '[' => '[', ']' => ']');
4103        $globstr =~ s{(.)} { $patmap{$1} || "\Q$1" }ge;
4104        return '^' . $globstr . '$';
4105}
4106
4107sub check_autoprop {
4108        my ($self, $pattern, $properties, $file, $fbat) = @_;
4109        # Convert the globbing pattern to a regular expression.
4110        my $regex = glob2pat($pattern);
4111        # Check if the pattern matches the file name.
4112        if($file =~ m/($regex)/) {
4113                # Parse the list of properties to set.
4114                my @props = split(/;/, $properties);
4115                foreach my $prop (@props) {
4116                        # Parse 'name=value' syntax and set the property.
4117                        if ($prop =~ /([^=]+)=(.*)/) {
4118                                my ($n,$v) = ($1,$2);
4119                                for ($n, $v) {
4120                                        s/^\s+//; s/\s+$//;
4121                                }
4122                                $self->change_file_prop($fbat, $n, $v);
4123                        }
4124                }
4125        }
4126}
4127
4128sub apply_autoprops {
4129        my ($self, $file, $fbat) = @_;
4130        my $conf_t = ${$self->{config}}{'config'};
4131        no warnings 'once';
4132        # Check [miscellany]/enable-auto-props in svn configuration.
4133        if (SVN::_Core::svn_config_get_bool(
4134                $conf_t,
4135                $SVN::_Core::SVN_CONFIG_SECTION_MISCELLANY,
4136                $SVN::_Core::SVN_CONFIG_OPTION_ENABLE_AUTO_PROPS,
4137                0)) {
4138                # Auto-props are enabled.  Enumerate them to look for matches.
4139                my $callback = sub {
4140                        $self->check_autoprop($_[0], $_[1], $file, $fbat);
4141                };
4142                SVN::_Core::svn_config_enumerate(
4143                        $conf_t,
4144                        $SVN::_Core::SVN_CONFIG_SECTION_AUTO_PROPS,
4145                        $callback);
4146        }
4147}
4148
4149sub A {
4150        my ($self, $m) = @_;
4151        my ($dir, $file) = split_path($m->{file_b});
4152        my $pbat = $self->ensure_path($dir);
4153        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4154                                        undef, -1);
4155        print "\tA\t$m->{file_b}\n" unless $::_q;
4156        $self->apply_autoprops($file, $fbat);
4157        $self->chg_file($fbat, $m);
4158        $self->close_file($fbat,undef,$self->{pool});
4159}
4160
4161sub C {
4162        my ($self, $m) = @_;
4163        my ($dir, $file) = split_path($m->{file_b});
4164        my $pbat = $self->ensure_path($dir);
4165        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4166                                $self->url_path($m->{file_a}), $self->{r});
4167        print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
4168        $self->chg_file($fbat, $m);
4169        $self->close_file($fbat,undef,$self->{pool});
4170}
4171
4172sub delete_entry {
4173        my ($self, $path, $pbat) = @_;
4174        my $rpath = $self->repo_path($path);
4175        my ($dir, $file) = split_path($rpath);
4176        $self->{rm}->{$dir} = 1;
4177        $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
4178}
4179
4180sub R {
4181        my ($self, $m) = @_;
4182        my ($dir, $file) = split_path($m->{file_b});
4183        my $pbat = $self->ensure_path($dir);
4184        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4185                                $self->url_path($m->{file_a}), $self->{r});
4186        print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
4187        $self->apply_autoprops($file, $fbat);
4188        $self->chg_file($fbat, $m);
4189        $self->close_file($fbat,undef,$self->{pool});
4190
4191        ($dir, $file) = split_path($m->{file_a});
4192        $pbat = $self->ensure_path($dir);
4193        $self->delete_entry($m->{file_a}, $pbat);
4194}
4195
4196sub M {
4197        my ($self, $m) = @_;
4198        my ($dir, $file) = split_path($m->{file_b});
4199        my $pbat = $self->ensure_path($dir);
4200        my $fbat = $self->open_file($self->repo_path($m->{file_b}),
4201                                $pbat,$self->{r},$self->{pool});
4202        print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
4203        $self->chg_file($fbat, $m);
4204        $self->close_file($fbat,undef,$self->{pool});
4205}
4206
4207sub T { shift->M(@_) }
4208
4209sub change_file_prop {
4210        my ($self, $fbat, $pname, $pval) = @_;
4211        $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
4212}
4213
4214sub _chg_file_get_blob ($$$$) {
4215        my ($self, $fbat, $m, $which) = @_;
4216        my $fh = $::_repository->temp_acquire("git_blob_$which");
4217        if ($m->{"mode_$which"} =~ /^120/) {
4218                print $fh 'link ' or croak $!;
4219                $self->change_file_prop($fbat,'svn:special','*');
4220        } elsif ($m->{mode_a} =~ /^120/ && $m->{"mode_$which"} !~ /^120/) {
4221                $self->change_file_prop($fbat,'svn:special',undef);
4222        }
4223        my $blob = $m->{"sha1_$which"};
4224        return ($fh,) if ($blob =~ /^0{40}$/);
4225        my $size = $::_repository->cat_blob($blob, $fh);
4226        croak "Failed to read object $blob" if ($size < 0);
4227        $fh->flush == 0 or croak $!;
4228        seek $fh, 0, 0 or croak $!;
4229
4230        my $exp = ::md5sum($fh);
4231        seek $fh, 0, 0 or croak $!;
4232        return ($fh, $exp);
4233}
4234
4235sub chg_file {
4236        my ($self, $fbat, $m) = @_;
4237        if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
4238                $self->change_file_prop($fbat,'svn:executable','*');
4239        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
4240                $self->change_file_prop($fbat,'svn:executable',undef);
4241        }
4242        my ($fh_a, $exp_a) = _chg_file_get_blob $self, $fbat, $m, 'a';
4243        my ($fh_b, $exp_b) = _chg_file_get_blob $self, $fbat, $m, 'b';
4244        my $pool = SVN::Pool->new;
4245        my $atd = $self->apply_textdelta($fbat, $exp_a, $pool);
4246        if (-s $fh_a) {
4247                my $txstream = SVN::TxDelta::new ($fh_a, $fh_b, $pool);
4248                my $res = SVN::TxDelta::send_txstream($txstream, @$atd, $pool);
4249                if (defined $res) {
4250                        die "Unexpected result from send_txstream: $res\n",
4251                            "(SVN::Core::VERSION: $SVN::Core::VERSION)\n";
4252                }
4253        } else {
4254                my $got = SVN::TxDelta::send_stream($fh_b, @$atd, $pool);
4255                die "Checksum mismatch\nexpected: $exp_b\ngot: $got\n"
4256                    if ($got ne $exp_b);
4257        }
4258        Git::temp_release($fh_b, 1);
4259        Git::temp_release($fh_a, 1);
4260        $pool->clear;
4261}
4262
4263sub D {
4264        my ($self, $m) = @_;
4265        my ($dir, $file) = split_path($m->{file_b});
4266        my $pbat = $self->ensure_path($dir);
4267        print "\tD\t$m->{file_b}\n" unless $::_q;
4268        $self->delete_entry($m->{file_b}, $pbat);
4269}
4270
4271sub close_edit {
4272        my ($self) = @_;
4273        my ($p,$bat) = ($self->{pool}, $self->{bat});
4274        foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
4275                next if $_ eq '';
4276                $self->close_directory($bat->{$_}, $p);
4277        }
4278        $self->close_directory($bat->{''}, $p);
4279        $self->SUPER::close_edit($p);
4280        $p->clear;
4281}
4282
4283sub abort_edit {
4284        my ($self) = @_;
4285        $self->SUPER::abort_edit($self->{pool});
4286}
4287
4288sub DESTROY {
4289        my $self = shift;
4290        $self->SUPER::DESTROY(@_);
4291        $self->{pool}->clear;
4292}
4293
4294# this drives the editor
4295sub apply_diff {
4296        my ($self) = @_;
4297        my $mods = $self->{mods};
4298        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
4299        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
4300                my $f = $m->{chg};
4301                if (defined $o{$f}) {
4302                        $self->$f($m);
4303                } else {
4304                        fatal("Invalid change type: $f");
4305                }
4306        }
4307        $self->rmdirs if $_rmdir;
4308        if (@$mods == 0) {
4309                $self->abort_edit;
4310        } else {
4311                $self->close_edit;
4312        }
4313        return scalar @$mods;
4314}
4315
4316package Git::SVN::Ra;
4317use vars qw/@ISA $config_dir $_log_window_size/;
4318use strict;
4319use warnings;
4320my ($ra_invalid, $can_do_switch, %ignored_err, $RA);
4321
4322BEGIN {
4323        # enforce temporary pool usage for some simple functions
4324        no strict 'refs';
4325        for my $f (qw/rev_proplist get_latest_revnum get_uuid get_repos_root
4326                      get_file/) {
4327                my $SUPER = "SUPER::$f";
4328                *$f = sub {
4329                        my $self = shift;
4330                        my $pool = SVN::Pool->new;
4331                        my @ret = $self->$SUPER(@_,$pool);
4332                        $pool->clear;
4333                        wantarray ? @ret : $ret[0];
4334                };
4335        }
4336}
4337
4338sub _auth_providers () {
4339        [
4340          SVN::Client::get_simple_provider(),
4341          SVN::Client::get_ssl_server_trust_file_provider(),
4342          SVN::Client::get_simple_prompt_provider(
4343            \&Git::SVN::Prompt::simple, 2),
4344          SVN::Client::get_ssl_client_cert_file_provider(),
4345          SVN::Client::get_ssl_client_cert_prompt_provider(
4346            \&Git::SVN::Prompt::ssl_client_cert, 2),
4347          SVN::Client::get_ssl_client_cert_pw_file_provider(),
4348          SVN::Client::get_ssl_client_cert_pw_prompt_provider(
4349            \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
4350          SVN::Client::get_username_provider(),
4351          SVN::Client::get_ssl_server_trust_prompt_provider(
4352            \&Git::SVN::Prompt::ssl_server_trust),
4353          SVN::Client::get_username_prompt_provider(
4354            \&Git::SVN::Prompt::username, 2)
4355        ]
4356}
4357
4358sub escape_uri_only {
4359        my ($uri) = @_;
4360        my @tmp;
4361        foreach (split m{/}, $uri) {
4362                s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
4363                push @tmp, $_;
4364        }
4365        join('/', @tmp);
4366}
4367
4368sub escape_url {
4369        my ($url) = @_;
4370        if ($url =~ m#^(https?)://([^/]+)(.*)$#) {
4371                my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
4372                $url = "$scheme://$domain$uri";
4373        }
4374        $url;
4375}
4376
4377sub new {
4378        my ($class, $url) = @_;
4379        $url =~ s!/+$!!;
4380        return $RA if ($RA && $RA->{url} eq $url);
4381
4382        SVN::_Core::svn_config_ensure($config_dir, undef);
4383        my ($baton, $callbacks) = SVN::Core::auth_open_helper(_auth_providers);
4384        my $config = SVN::Core::config_get_config($config_dir);
4385        $RA = undef;
4386        my $dont_store_passwords = 1;
4387        my $conf_t = ${$config}{'config'};
4388        {
4389                no warnings 'once';
4390                # The usage of $SVN::_Core::SVN_CONFIG_* variables
4391                # produces warnings that variables are used only once.
4392                # I had not found the better way to shut them up, so
4393                # the warnings of type 'once' are disabled in this block.
4394                if (SVN::_Core::svn_config_get_bool($conf_t,
4395                    $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4396                    $SVN::_Core::SVN_CONFIG_OPTION_STORE_PASSWORDS,
4397                    1) == 0) {
4398                        SVN::_Core::svn_auth_set_parameter($baton,
4399                            $SVN::_Core::SVN_AUTH_PARAM_DONT_STORE_PASSWORDS,
4400                            bless (\$dont_store_passwords, "_p_void"));
4401                }
4402                if (SVN::_Core::svn_config_get_bool($conf_t,
4403                    $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4404                    $SVN::_Core::SVN_CONFIG_OPTION_STORE_AUTH_CREDS,
4405                    1) == 0) {
4406                        $Git::SVN::Prompt::_no_auth_cache = 1;
4407                }
4408        } # no warnings 'once'
4409        my $self = SVN::Ra->new(url => escape_url($url), auth => $baton,
4410                              config => $config,
4411                              pool => SVN::Pool->new,
4412                              auth_provider_callbacks => $callbacks);
4413        $self->{url} = $url;
4414        $self->{svn_path} = $url;
4415        $self->{repos_root} = $self->get_repos_root;
4416        $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
4417        $self->{cache} = { check_path => { r => 0, data => {} },
4418                           get_dir => { r => 0, data => {} } };
4419        $RA = bless $self, $class;
4420}
4421
4422sub check_path {
4423        my ($self, $path, $r) = @_;
4424        my $cache = $self->{cache}->{check_path};
4425        if ($r == $cache->{r} && exists $cache->{data}->{$path}) {
4426                return $cache->{data}->{$path};
4427        }
4428        my $pool = SVN::Pool->new;
4429        my $t = $self->SUPER::check_path($path, $r, $pool);
4430        $pool->clear;
4431        if ($r != $cache->{r}) {
4432                %{$cache->{data}} = ();
4433                $cache->{r} = $r;
4434        }
4435        $cache->{data}->{$path} = $t;
4436}
4437
4438sub get_dir {
4439        my ($self, $dir, $r) = @_;
4440        my $cache = $self->{cache}->{get_dir};
4441        if ($r == $cache->{r}) {
4442                if (my $x = $cache->{data}->{$dir}) {
4443                        return wantarray ? @$x : $x->[0];
4444                }
4445        }
4446        my $pool = SVN::Pool->new;
4447        my ($d, undef, $props) = $self->SUPER::get_dir($dir, $r, $pool);
4448        my %dirents = map { $_ => { kind => $d->{$_}->kind } } keys %$d;
4449        $pool->clear;
4450        if ($r != $cache->{r}) {
4451                %{$cache->{data}} = ();
4452                $cache->{r} = $r;
4453        }
4454        $cache->{data}->{$dir} = [ \%dirents, $r, $props ];
4455        wantarray ? (\%dirents, $r, $props) : \%dirents;
4456}
4457
4458sub DESTROY {
4459        # do not call the real DESTROY since we store ourselves in $RA
4460}
4461
4462# get_log(paths, start, end, limit,
4463#         discover_changed_paths, strict_node_history, receiver)
4464sub get_log {
4465        my ($self, @args) = @_;
4466        my $pool = SVN::Pool->new;
4467
4468        # svn_log_changed_path_t objects passed to get_log are likely to be
4469        # overwritten even if only the refs are copied to an external variable,
4470        # so we should dup the structures in their entirety.  Using an
4471        # externally passed pool (instead of our temporary and quickly cleared
4472        # pool in Git::SVN::Ra) does not help matters at all...
4473        my $receiver = pop @args;
4474        my $prefix = "/".$self->{svn_path};
4475        $prefix =~ s#/+($)##;
4476        my $prefix_regex = qr#^\Q$prefix\E#;
4477        push(@args, sub {
4478                my ($paths) = $_[0];
4479                return &$receiver(@_) unless $paths;
4480                $_[0] = ();
4481                foreach my $p (keys %$paths) {
4482                        my $i = $paths->{$p};
4483                        # Make path relative to our url, not repos_root
4484                        $p =~ s/$prefix_regex//;
4485                        my %s = map { $_ => $i->$_; }
4486                                qw/copyfrom_path copyfrom_rev action/;
4487                        if ($s{'copyfrom_path'}) {
4488                                $s{'copyfrom_path'} =~ s/$prefix_regex//;
4489                        }
4490                        $_[0]{$p} = \%s;
4491                }
4492                &$receiver(@_);
4493        });
4494
4495
4496        # the limit parameter was not supported in SVN 1.1.x, so we
4497        # drop it.  Therefore, the receiver callback passed to it
4498        # is made aware of this limitation by being wrapped if
4499        # the limit passed to is being wrapped.
4500        if ($SVN::Core::VERSION le '1.2.0') {
4501                my $limit = splice(@args, 3, 1);
4502                if ($limit > 0) {
4503                        my $receiver = pop @args;
4504                        push(@args, sub { &$receiver(@_) if (--$limit >= 0) });
4505                }
4506        }
4507        my $ret = $self->SUPER::get_log(@args, $pool);
4508        $pool->clear;
4509        $ret;
4510}
4511
4512sub trees_match {
4513        my ($self, $url1, $rev1, $url2, $rev2) = @_;
4514        my $ctx = SVN::Client->new(auth => _auth_providers);
4515        my $out = IO::File->new_tmpfile;
4516
4517        # older SVN (1.1.x) doesn't take $pool as the last parameter for
4518        # $ctx->diff(), so we'll create a default one
4519        my $pool = SVN::Pool->new_default_sub;
4520
4521        $ra_invalid = 1; # this will open a new SVN::Ra connection to $url1
4522        $ctx->diff([], $url1, $rev1, $url2, $rev2, 1, 1, 0, $out, $out);
4523        $out->flush;
4524        my $ret = (($out->stat)[7] == 0);
4525        close $out or croak $!;
4526
4527        $ret;
4528}
4529
4530sub get_commit_editor {
4531        my ($self, $log, $cb, $pool) = @_;
4532        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
4533        $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
4534}
4535
4536sub gs_do_update {
4537        my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
4538        my $new = ($rev_a == $rev_b);
4539        my $path = $gs->{path};
4540
4541        if ($new && -e $gs->{index}) {
4542                unlink $gs->{index} or die
4543                  "Couldn't unlink index: $gs->{index}: $!\n";
4544        }
4545        my $pool = SVN::Pool->new;
4546        $editor->set_path_strip($path);
4547        my (@pc) = split m#/#, $path;
4548        my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
4549                                        1, $editor, $pool);
4550        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
4551
4552        # Since we can't rely on svn_ra_reparent being available, we'll
4553        # just have to do some magic with set_path to make it so
4554        # we only want a partial path.
4555        my $sp = '';
4556        my $final = join('/', @pc);
4557        while (@pc) {
4558                $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
4559                $sp .= '/' if length $sp;
4560                $sp .= shift @pc;
4561        }
4562        die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
4563
4564        $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
4565
4566        $reporter->finish_report($pool);
4567        $pool->clear;
4568        $editor->{git_commit_ok};
4569}
4570
4571# this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
4572# svn_ra_reparent didn't work before 1.4)
4573sub gs_do_switch {
4574        my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
4575        my $path = $gs->{path};
4576        my $pool = SVN::Pool->new;
4577
4578        my $full_url = $self->{url};
4579        my $old_url = $full_url;
4580        $full_url .= '/' . $path if length $path;
4581        my ($ra, $reparented);
4582
4583        if ($old_url =~ m#^svn(\+ssh)?://# ||
4584            ($full_url =~ m#^https?://# &&
4585             escape_url($full_url) ne $full_url)) {
4586                $_[0] = undef;
4587                $self = undef;
4588                $RA = undef;
4589                $ra = Git::SVN::Ra->new($full_url);
4590                $ra_invalid = 1;
4591        } elsif ($old_url ne $full_url) {
4592                SVN::_Ra::svn_ra_reparent($self->{session}, $full_url, $pool);
4593                $self->{url} = $full_url;
4594                $reparented = 1;
4595        }
4596
4597        $ra ||= $self;
4598        $url_b = escape_url($url_b);
4599        my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
4600        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
4601        $reporter->set_path('', $rev_a, 0, @lock, $pool);
4602        $reporter->finish_report($pool);
4603
4604        if ($reparented) {
4605                SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
4606                $self->{url} = $old_url;
4607        }
4608
4609        $pool->clear;
4610        $editor->{git_commit_ok};
4611}
4612
4613sub longest_common_path {
4614        my ($gsv, $globs) = @_;
4615        my %common;
4616        my $common_max = scalar @$gsv;
4617
4618        foreach my $gs (@$gsv) {
4619                my @tmp = split m#/#, $gs->{path};
4620                my $p = '';
4621                foreach (@tmp) {
4622                        $p .= length($p) ? "/$_" : $_;
4623                        $common{$p} ||= 0;
4624                        $common{$p}++;
4625                }
4626        }
4627        $globs ||= [];
4628        $common_max += scalar @$globs;
4629        foreach my $glob (@$globs) {
4630                my @tmp = split m#/#, $glob->{path}->{left};
4631                my $p = '';
4632                foreach (@tmp) {
4633                        $p .= length($p) ? "/$_" : $_;
4634                        $common{$p} ||= 0;
4635                        $common{$p}++;
4636                }
4637        }
4638
4639        my $longest_path = '';
4640        foreach (sort {length $b <=> length $a} keys %common) {
4641                if ($common{$_} == $common_max) {
4642                        $longest_path = $_;
4643                        last;
4644                }
4645        }
4646        $longest_path;
4647}
4648
4649sub gs_fetch_loop_common {
4650        my ($self, $base, $head, $gsv, $globs) = @_;
4651        return if ($base > $head);
4652        my $inc = $_log_window_size;
4653        my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
4654        my $longest_path = longest_common_path($gsv, $globs);
4655        my $ra_url = $self->{url};
4656        my $find_trailing_edge;
4657        while (1) {
4658                my %revs;
4659                my $err;
4660                my $err_handler = $SVN::Error::handler;
4661                $SVN::Error::handler = sub {
4662                        ($err) = @_;
4663                        skip_unknown_revs($err);
4664                };
4665                sub _cb {
4666                        my ($paths, $r, $author, $date, $log) = @_;
4667                        [ $paths,
4668                          { author => $author, date => $date, log => $log } ];
4669                }
4670                $self->get_log([$longest_path], $min, $max, 0, 1, 1,
4671                               sub { $revs{$_[1]} = _cb(@_) });
4672                if ($err) {
4673                        print "Checked through r$max\r";
4674                } else {
4675                        $find_trailing_edge = 1;
4676                }
4677                if ($err and $find_trailing_edge) {
4678                        print STDERR "Path '$longest_path' ",
4679                                     "was probably deleted:\n",
4680                                     $err->expanded_message,
4681                                     "\nWill attempt to follow ",
4682                                     "revisions r$min .. r$max ",
4683                                     "committed before the deletion\n";
4684                        my $hi = $max;
4685                        while (--$hi >= $min) {
4686                                my $ok;
4687                                $self->get_log([$longest_path], $min, $hi,
4688                                               0, 1, 1, sub {
4689                                               $ok = $_[1];
4690                                               $revs{$_[1]} = _cb(@_) });
4691                                if ($ok) {
4692                                        print STDERR "r$min .. r$ok OK\n";
4693                                        last;
4694                                }
4695                        }
4696                        $find_trailing_edge = 0;
4697                }
4698                $SVN::Error::handler = $err_handler;
4699
4700                my %exists = map { $_->{path} => $_ } @$gsv;
4701                foreach my $r (sort {$a <=> $b} keys %revs) {
4702                        my ($paths, $logged) = @{$revs{$r}};
4703
4704                        foreach my $gs ($self->match_globs(\%exists, $paths,
4705                                                           $globs, $r)) {
4706                                if ($gs->rev_map_max >= $r) {
4707                                        next;
4708                                }
4709                                next unless $gs->match_paths($paths, $r);
4710                                $gs->{logged_rev_props} = $logged;
4711                                if (my $last_commit = $gs->last_commit) {
4712                                        $gs->assert_index_clean($last_commit);
4713                                }
4714                                my $log_entry = $gs->do_fetch($paths, $r);
4715                                if ($log_entry) {
4716                                        $gs->do_git_commit($log_entry);
4717                                }
4718                                $INDEX_FILES{$gs->{index}} = 1;
4719                        }
4720                        foreach my $g (@$globs) {
4721                                my $k = "svn-remote.$g->{remote}." .
4722                                        "$g->{t}-maxRev";
4723                                Git::SVN::tmp_config($k, $r);
4724                        }
4725                        if ($ra_invalid) {
4726                                $_[0] = undef;
4727                                $self = undef;
4728                                $RA = undef;
4729                                $self = Git::SVN::Ra->new($ra_url);
4730                                $ra_invalid = undef;
4731                        }
4732                }
4733                # pre-fill the .rev_db since it'll eventually get filled in
4734                # with '0' x40 if something new gets committed
4735                foreach my $gs (@$gsv) {
4736                        next if $gs->rev_map_max >= $max;
4737                        next if defined $gs->rev_map_get($max);
4738                        $gs->rev_map_set($max, 0 x40);
4739                }
4740                foreach my $g (@$globs) {
4741                        my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
4742                        Git::SVN::tmp_config($k, $max);
4743                }
4744                last if $max >= $head;
4745                $min = $max + 1;
4746                $max += $inc;
4747                $max = $head if ($max > $head);
4748        }
4749        Git::SVN::gc();
4750}
4751
4752sub get_dir_globbed {
4753        my ($self, $left, $depth, $r) = @_;
4754
4755        my @x = eval { $self->get_dir($left, $r) };
4756        return unless scalar @x == 3;
4757        my $dirents = $x[0];
4758        my @finalents;
4759        foreach my $de (keys %$dirents) {
4760                next if $dirents->{$de}->{kind} != $SVN::Node::dir;
4761                if ($depth > 1) {
4762                        my @args = ("$left/$de", $depth - 1, $r);
4763                        foreach my $dir ($self->get_dir_globbed(@args)) {
4764                                push @finalents, "$de/$dir";
4765                        }
4766                } else {
4767                        push @finalents, $de;
4768                }
4769        }
4770        @finalents;
4771}
4772
4773sub match_globs {
4774        my ($self, $exists, $paths, $globs, $r) = @_;
4775
4776        sub get_dir_check {
4777                my ($self, $exists, $g, $r) = @_;
4778
4779                my @dirs = $self->get_dir_globbed($g->{path}->{left},
4780                                                  $g->{path}->{depth},
4781                                                  $r);
4782
4783                foreach my $de (@dirs) {
4784                        my $p = $g->{path}->full_path($de);
4785                        next if $exists->{$p};
4786                        next if (length $g->{path}->{right} &&
4787                                 ($self->check_path($p, $r) !=
4788                                  $SVN::Node::dir));
4789                        $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
4790                                         $g->{ref}->full_path($de), 1);
4791                }
4792        }
4793        foreach my $g (@$globs) {
4794                if (my $path = $paths->{"/$g->{path}->{left}"}) {
4795                        if ($path->{action} =~ /^[AR]$/) {
4796                                get_dir_check($self, $exists, $g, $r);
4797                        }
4798                }
4799                foreach (keys %$paths) {
4800                        if (/$g->{path}->{left_regex}/ &&
4801                            !/$g->{path}->{regex}/) {
4802                                next if $paths->{$_}->{action} !~ /^[AR]$/;
4803                                get_dir_check($self, $exists, $g, $r);
4804                        }
4805                        next unless /$g->{path}->{regex}/;
4806                        my $p = $1;
4807                        my $pathname = $g->{path}->full_path($p);
4808                        next if $exists->{$pathname};
4809                        next if ($self->check_path($pathname, $r) !=
4810                                 $SVN::Node::dir);
4811                        $exists->{$pathname} = Git::SVN->init(
4812                                              $self->{url}, $pathname, undef,
4813                                              $g->{ref}->full_path($p), 1);
4814                }
4815                my $c = '';
4816                foreach (split m#/#, $g->{path}->{left}) {
4817                        $c .= "/$_";
4818                        next unless ($paths->{$c} &&
4819                                     ($paths->{$c}->{action} =~ /^[AR]$/));
4820                        get_dir_check($self, $exists, $g, $r);
4821                }
4822        }
4823        values %$exists;
4824}
4825
4826sub minimize_url {
4827        my ($self) = @_;
4828        return $self->{url} if ($self->{url} eq $self->{repos_root});
4829        my $url = $self->{repos_root};
4830        my @components = split(m!/!, $self->{svn_path});
4831        my $c = '';
4832        do {
4833                $url .= "/$c" if length $c;
4834                eval {
4835                        my $ra = (ref $self)->new($url);
4836                        my $latest = $ra->get_latest_revnum;
4837                        $ra->get_log("", $latest, 0, 1, 0, 1, sub {});
4838                };
4839        } while ($@ && ($c = shift @components));
4840        $url;
4841}
4842
4843sub can_do_switch {
4844        my $self = shift;
4845        unless (defined $can_do_switch) {
4846                my $pool = SVN::Pool->new;
4847                my $rep = eval {
4848                        $self->do_switch(1, '', 0, $self->{url},
4849                                         SVN::Delta::Editor->new, $pool);
4850                };
4851                if ($@) {
4852                        $can_do_switch = 0;
4853                } else {
4854                        $rep->abort_report($pool);
4855                        $can_do_switch = 1;
4856                }
4857                $pool->clear;
4858        }
4859        $can_do_switch;
4860}
4861
4862sub skip_unknown_revs {
4863        my ($err) = @_;
4864        my $errno = $err->apr_err();
4865        # Maybe the branch we're tracking didn't
4866        # exist when the repo started, so it's
4867        # not an error if it doesn't, just continue
4868        #
4869        # Wonderfully consistent library, eh?
4870        # 160013 - svn:// and file://
4871        # 175002 - http(s)://
4872        # 175007 - http(s):// (this repo required authorization, too...)
4873        #   More codes may be discovered later...
4874        if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
4875                my $err_key = $err->expanded_message;
4876                # revision numbers change every time, filter them out
4877                $err_key =~ s/\d+/\0/g;
4878                $err_key = "$errno\0$err_key";
4879                unless ($ignored_err{$err_key}) {
4880                        warn "W: Ignoring error from SVN, path probably ",
4881                             "does not exist: ($errno): ",
4882                             $err->expanded_message,"\n";
4883                        warn "W: Do not be alarmed at the above message ",
4884                             "git-svn is just searching aggressively for ",
4885                             "old history.\n",
4886                             "This may take a while on large repositories\n";
4887                        $ignored_err{$err_key} = 1;
4888                }
4889                return;
4890        }
4891        die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
4892}
4893
4894package Git::SVN::Log;
4895use strict;
4896use warnings;
4897use POSIX qw/strftime/;
4898use Time::Local;
4899use constant commit_log_separator => ('-' x 72) . "\n";
4900use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
4901            %rusers $show_commit $incremental/;
4902my $l_fmt;
4903
4904sub cmt_showable {
4905        my ($c) = @_;
4906        return 1 if defined $c->{r};
4907
4908        # big commit message got truncated by the 16k pretty buffer in rev-list
4909        if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
4910                                $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
4911                @{$c->{l}} = ();
4912                my @log = command(qw/cat-file commit/, $c->{c});
4913
4914                # shift off the headers
4915                shift @log while ($log[0] ne '');
4916                shift @log;
4917
4918                # TODO: make $c->{l} not have a trailing newline in the future
4919                @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
4920
4921                (undef, $c->{r}, undef) = ::extract_metadata(
4922                                (grep(/^git-svn-id: /, @log))[-1]);
4923        }
4924        return defined $c->{r};
4925}
4926
4927sub log_use_color {
4928        return $color || Git->repository->get_colorbool('color.diff');
4929}
4930
4931sub git_svn_log_cmd {
4932        my ($r_min, $r_max, @args) = @_;
4933        my $head = 'HEAD';
4934        my (@files, @log_opts);
4935        foreach my $x (@args) {
4936                if ($x eq '--' || @files) {
4937                        push @files, $x;
4938                } else {
4939                        if (::verify_ref("$x^0")) {
4940                                $head = $x;
4941                        } else {
4942                                push @log_opts, $x;
4943                        }
4944                }
4945        }
4946
4947        my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
4948        $gs ||= Git::SVN->_new;
4949        my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
4950                   $gs->refname);
4951        push @cmd, '-r' unless $non_recursive;
4952        push @cmd, qw/--raw --name-status/ if $verbose;
4953        push @cmd, '--color' if log_use_color();
4954        push @cmd, @log_opts;
4955        if (defined $r_max && $r_max == $r_min) {
4956                push @cmd, '--max-count=1';
4957                if (my $c = $gs->rev_map_get($r_max)) {
4958                        push @cmd, $c;
4959                }
4960        } elsif (defined $r_max) {
4961                if ($r_max < $r_min) {
4962                        ($r_min, $r_max) = ($r_max, $r_min);
4963                }
4964                my (undef, $c_max) = $gs->find_rev_before($r_max, 1, $r_min);
4965                my (undef, $c_min) = $gs->find_rev_after($r_min, 1, $r_max);
4966                # If there are no commits in the range, both $c_max and $c_min
4967                # will be undefined.  If there is at least 1 commit in the
4968                # range, both will be defined.
4969                return () if !defined $c_min || !defined $c_max;
4970                if ($c_min eq $c_max) {
4971                        push @cmd, '--max-count=1', $c_min;
4972                } else {
4973                        push @cmd, '--boundary', "$c_min..$c_max";
4974                }
4975        }
4976        return (@cmd, @files);
4977}
4978
4979# adapted from pager.c
4980sub config_pager {
4981        $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
4982        if (!defined $pager) {
4983                $pager = 'less';
4984        } elsif (length $pager == 0 || $pager eq 'cat') {
4985                $pager = undef;
4986        }
4987        $ENV{GIT_PAGER_IN_USE} = defined($pager);
4988}
4989
4990sub run_pager {
4991        return unless -t *STDOUT && defined $pager;
4992        pipe my ($rfd, $wfd) or return;
4993        defined(my $pid = fork) or ::fatal "Can't fork: $!";
4994        if (!$pid) {
4995                open STDOUT, '>&', $wfd or
4996                                     ::fatal "Can't redirect to stdout: $!";
4997                return;
4998        }
4999        open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!";
5000        $ENV{LESS} ||= 'FRSX';
5001        exec $pager or ::fatal "Can't run pager: $! ($pager)";
5002}
5003
5004sub format_svn_date {
5005        # some systmes don't handle or mishandle %z, so be creative.
5006        my $t = shift || time;
5007        my $gm = timelocal(gmtime($t));
5008        my $sign = qw( + + - )[ $t <=> $gm ];
5009        my $gmoff = sprintf("%s%02d%02d", $sign, (gmtime(abs($t - $gm)))[2,1]);
5010        return strftime("%Y-%m-%d %H:%M:%S $gmoff (%a, %d %b %Y)", localtime($t));
5011}
5012
5013sub parse_git_date {
5014        my ($t, $tz) = @_;
5015        # Date::Parse isn't in the standard Perl distro :(
5016        if ($tz =~ s/^\+//) {
5017                $t += tz_to_s_offset($tz);
5018        } elsif ($tz =~ s/^\-//) {
5019                $t -= tz_to_s_offset($tz);
5020        }
5021        return $t;
5022}
5023
5024sub set_local_timezone {
5025        if (defined $TZ) {
5026                $ENV{TZ} = $TZ;
5027        } else {
5028                delete $ENV{TZ};
5029        }
5030}
5031
5032sub tz_to_s_offset {
5033        my ($tz) = @_;
5034        $tz =~ s/(\d\d)$//;
5035        return ($1 * 60) + ($tz * 3600);
5036}
5037
5038sub get_author_info {
5039        my ($dest, $author, $t, $tz) = @_;
5040        $author =~ s/(?:^\s*|\s*$)//g;
5041        $dest->{a_raw} = $author;
5042        my $au;
5043        if ($::_authors) {
5044                $au = $rusers{$author} || undef;
5045        }
5046        if (!$au) {
5047                ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
5048        }
5049        $dest->{t} = $t;
5050        $dest->{tz} = $tz;
5051        $dest->{a} = $au;
5052        $dest->{t_utc} = parse_git_date($t, $tz);
5053}
5054
5055sub process_commit {
5056        my ($c, $r_min, $r_max, $defer) = @_;
5057        if (defined $r_min && defined $r_max) {
5058                if ($r_min == $c->{r} && $r_min == $r_max) {
5059                        show_commit($c);
5060                        return 0;
5061                }
5062                return 1 if $r_min == $r_max;
5063                if ($r_min < $r_max) {
5064                        # we need to reverse the print order
5065                        return 0 if (defined $limit && --$limit < 0);
5066                        push @$defer, $c;
5067                        return 1;
5068                }
5069                if ($r_min != $r_max) {
5070                        return 1 if ($r_min < $c->{r});
5071                        return 1 if ($r_max > $c->{r});
5072                }
5073        }
5074        return 0 if (defined $limit && --$limit < 0);
5075        show_commit($c);
5076        return 1;
5077}
5078
5079sub show_commit {
5080        my $c = shift;
5081        if ($oneline) {
5082                my $x = "\n";
5083                if (my $l = $c->{l}) {
5084                        while ($l->[0] =~ /^\s*$/) { shift @$l }
5085                        $x = $l->[0];
5086                }
5087                $l_fmt ||= 'A' . length($c->{r});
5088                print 'r',pack($l_fmt, $c->{r}),' | ';
5089                print "$c->{c} | " if $show_commit;
5090                print $x;
5091        } else {
5092                show_commit_normal($c);
5093        }
5094}
5095
5096sub show_commit_changed_paths {
5097        my ($c) = @_;
5098        return unless $c->{changed};
5099        print "Changed paths:\n", @{$c->{changed}};
5100}
5101
5102sub show_commit_normal {
5103        my ($c) = @_;
5104        print commit_log_separator, "r$c->{r} | ";
5105        print "$c->{c} | " if $show_commit;
5106        print "$c->{a} | ", format_svn_date($c->{t_utc}), ' | ';
5107        my $nr_line = 0;
5108
5109        if (my $l = $c->{l}) {
5110                while ($l->[$#$l] eq "\n" && $#$l > 0
5111                                          && $l->[($#$l - 1)] eq "\n") {
5112                        pop @$l;
5113                }
5114                $nr_line = scalar @$l;
5115                if (!$nr_line) {
5116                        print "1 line\n\n\n";
5117                } else {
5118                        if ($nr_line == 1) {
5119                                $nr_line = '1 line';
5120                        } else {
5121                                $nr_line .= ' lines';
5122                        }
5123                        print $nr_line, "\n";
5124                        show_commit_changed_paths($c);
5125                        print "\n";
5126                        print $_ foreach @$l;
5127                }
5128        } else {
5129                print "1 line\n";
5130                show_commit_changed_paths($c);
5131                print "\n";
5132
5133        }
5134        foreach my $x (qw/raw stat diff/) {
5135                if ($c->{$x}) {
5136                        print "\n";
5137                        print $_ foreach @{$c->{$x}}
5138                }
5139        }
5140}
5141
5142sub cmd_show_log {
5143        my (@args) = @_;
5144        my ($r_min, $r_max);
5145        my $r_last = -1; # prevent dupes
5146        set_local_timezone();
5147        if (defined $::_revision) {
5148                if ($::_revision =~ /^(\d+):(\d+)$/) {
5149                        ($r_min, $r_max) = ($1, $2);
5150                } elsif ($::_revision =~ /^\d+$/) {
5151                        $r_min = $r_max = $::_revision;
5152                } else {
5153                        ::fatal "-r$::_revision is not supported, use ",
5154                                "standard 'git log' arguments instead";
5155                }
5156        }
5157
5158        config_pager();
5159        @args = git_svn_log_cmd($r_min, $r_max, @args);
5160        if (!@args) {
5161                print commit_log_separator unless $incremental || $oneline;
5162                return;
5163        }
5164        my $log = command_output_pipe(@args);
5165        run_pager();
5166        my (@k, $c, $d, $stat);
5167        my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
5168        while (<$log>) {
5169                if (/^${esc_color}commit -?($::sha1_short)/o) {
5170                        my $cmt = $1;
5171                        if ($c && cmt_showable($c) && $c->{r} != $r_last) {
5172                                $r_last = $c->{r};
5173                                process_commit($c, $r_min, $r_max, \@k) or
5174                                                                goto out;
5175                        }
5176                        $d = undef;
5177                        $c = { c => $cmt };
5178                } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
5179                        get_author_info($c, $1, $2, $3);
5180                } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
5181                        # ignore
5182                } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
5183                        push @{$c->{raw}}, $_;
5184                } elsif (/^${esc_color}[ACRMDT]\t/) {
5185                        # we could add $SVN->{svn_path} here, but that requires
5186                        # remote access at the moment (repo_path_split)...
5187                        s#^(${esc_color})([ACRMDT])\t#$1   $2 #o;
5188                        push @{$c->{changed}}, $_;
5189                } elsif (/^${esc_color}diff /o) {
5190                        $d = 1;
5191                        push @{$c->{diff}}, $_;
5192                } elsif ($d) {
5193                        push @{$c->{diff}}, $_;
5194                } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
5195                          $esc_color*[\+\-]*$esc_color$/x) {
5196                        $stat = 1;
5197                        push @{$c->{stat}}, $_;
5198                } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
5199                        push @{$c->{stat}}, $_;
5200                        $stat = undef;
5201                } elsif (/^${esc_color}    (git-svn-id:.+)$/o) {
5202                        ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
5203                } elsif (s/^${esc_color}    //o) {
5204                        push @{$c->{l}}, $_;
5205                }
5206        }
5207        if ($c && defined $c->{r} && $c->{r} != $r_last) {
5208                $r_last = $c->{r};
5209                process_commit($c, $r_min, $r_max, \@k);
5210        }
5211        if (@k) {
5212                ($r_min, $r_max) = ($r_max, $r_min);
5213                process_commit($_, $r_min, $r_max) foreach reverse @k;
5214        }
5215out:
5216        close $log;
5217        print commit_log_separator unless $incremental || $oneline;
5218}
5219
5220sub cmd_blame {
5221        my $path = pop;
5222
5223        config_pager();
5224        run_pager();
5225
5226        my ($fh, $ctx, $rev);
5227
5228        if ($_git_format) {
5229                ($fh, $ctx) = command_output_pipe('blame', @_, $path);
5230                while (my $line = <$fh>) {
5231                        if ($line =~ /^\^?([[:xdigit:]]+)\s/) {
5232                                # Uncommitted edits show up as a rev ID of
5233                                # all zeros, which we can't look up with
5234                                # cmt_metadata
5235                                if ($1 !~ /^0+$/) {
5236                                        (undef, $rev, undef) =
5237                                                ::cmt_metadata($1);
5238                                        $rev = '0' if (!$rev);
5239                                } else {
5240                                        $rev = '0';
5241                                }
5242                                $rev = sprintf('%-10s', $rev);
5243                                $line =~ s/^\^?[[:xdigit:]]+(\s)/$rev$1/;
5244                        }
5245                        print $line;
5246                }
5247        } else {
5248                ($fh, $ctx) = command_output_pipe('blame', '-p', @_, 'HEAD',
5249                                                  '--', $path);
5250                my ($sha1);
5251                my %authors;
5252                my @buffer;
5253                my %dsha; #distinct sha keys
5254
5255                while (my $line = <$fh>) {
5256                        push @buffer, $line;
5257                        if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5258                                $dsha{$1} = 1;
5259                        }
5260                }
5261
5262                my $s2r = ::cmt_sha2rev_batch([keys %dsha]);
5263
5264                foreach my $line (@buffer) {
5265                        if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5266                                $rev = $s2r->{$1};
5267                                $rev = '0' if (!$rev)
5268                        }
5269                        elsif ($line =~ /^author (.*)/) {
5270                                $authors{$rev} = $1;
5271                                $authors{$rev} =~ s/\s/_/g;
5272                        }
5273                        elsif ($line =~ /^\t(.*)$/) {
5274                                printf("%6s %10s %s\n", $rev, $authors{$rev}, $1);
5275                        }
5276                }
5277        }
5278        command_close_pipe($fh, $ctx);
5279}
5280
5281package Git::SVN::Migration;
5282# these version numbers do NOT correspond to actual version numbers
5283# of git nor git-svn.  They are just relative.
5284#
5285# v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
5286#
5287# v1 layout: .git/$id/info/url, refs/remotes/$id
5288#
5289# v2 layout: .git/svn/$id/info/url, refs/remotes/$id
5290#
5291# v3 layout: .git/svn/$id, refs/remotes/$id
5292#            - info/url may remain for backwards compatibility
5293#            - this is what we migrate up to this layout automatically,
5294#            - this will be used by git svn init on single branches
5295# v3.1 layout (auto migrated):
5296#            - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
5297#              for backwards compatibility
5298#
5299# v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
5300#            - this is only created for newly multi-init-ed
5301#              repositories.  Similar in spirit to the
5302#              --use-separate-remotes option in git-clone (now default)
5303#            - we do not automatically migrate to this (following
5304#              the example set by core git)
5305#
5306# v5 layout: .rev_db.$UUID => .rev_map.$UUID
5307#            - newer, more-efficient format that uses 24-bytes per record
5308#              with no filler space.
5309#            - use xxd -c24 < .rev_map.$UUID to view and debug
5310#            - This is a one-way migration, repositories updated to the
5311#              new format will not be able to use old git-svn without
5312#              rebuilding the .rev_db.  Rebuilding the rev_db is not
5313#              possible if noMetadata or useSvmProps are set; but should
5314#              be no problem for users that use the (sensible) defaults.
5315use strict;
5316use warnings;
5317use Carp qw/croak/;
5318use File::Path qw/mkpath/;
5319use File::Basename qw/dirname basename/;
5320use vars qw/$_minimize/;
5321
5322sub migrate_from_v0 {
5323        my $git_dir = $ENV{GIT_DIR};
5324        return undef unless -d $git_dir;
5325        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5326        my $migrated = 0;
5327        while (<$fh>) {
5328                chomp;
5329                my ($id, $orig_ref) = ($_, $_);
5330                next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
5331                next unless -f "$git_dir/$id/info/url";
5332                my $new_ref = "refs/remotes/$id";
5333                if (::verify_ref("$new_ref^0")) {
5334                        print STDERR "W: $orig_ref is probably an old ",
5335                                     "branch used by an ancient version of ",
5336                                     "git-svn.\n",
5337                                     "However, $new_ref also exists.\n",
5338                                     "We will not be able ",
5339                                     "to use this branch until this ",
5340                                     "ambiguity is resolved.\n";
5341                        next;
5342                }
5343                print STDERR "Migrating from v0 layout...\n" if !$migrated;
5344                print STDERR "Renaming ref: $orig_ref => $new_ref\n";
5345                command_noisy('update-ref', $new_ref, $orig_ref);
5346                command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
5347                $migrated++;
5348        }
5349        command_close_pipe($fh, $ctx);
5350        print STDERR "Done migrating from v0 layout...\n" if $migrated;
5351        $migrated;
5352}
5353
5354sub migrate_from_v1 {
5355        my $git_dir = $ENV{GIT_DIR};
5356        my $migrated = 0;
5357        return $migrated unless -d $git_dir;
5358        my $svn_dir = "$git_dir/svn";
5359
5360        # just in case somebody used 'svn' as their $id at some point...
5361        return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
5362
5363        print STDERR "Migrating from a git-svn v1 layout...\n";
5364        mkpath([$svn_dir]);
5365        print STDERR "Data from a previous version of git-svn exists, but\n\t",
5366                     "$svn_dir\n\t(required for this version ",
5367                     "($::VERSION) of git-svn) does not exist.\n";
5368        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5369        while (<$fh>) {
5370                my $x = $_;
5371                next unless $x =~ s#^refs/remotes/##;
5372                chomp $x;
5373                next unless -f "$git_dir/$x/info/url";
5374                my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
5375                next unless $u;
5376                my $dn = dirname("$git_dir/svn/$x");
5377                mkpath([$dn]) unless -d $dn;
5378                if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
5379                        mkpath(["$git_dir/svn/svn"]);
5380                        print STDERR " - $git_dir/$x/info => ",
5381                                        "$git_dir/svn/$x/info\n";
5382                        rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
5383                               croak "$!: $x";
5384                        # don't worry too much about these, they probably
5385                        # don't exist with repos this old (save for index,
5386                        # and we can easily regenerate that)
5387                        foreach my $f (qw/unhandled.log index .rev_db/) {
5388                                rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
5389                        }
5390                } else {
5391                        print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
5392                        rename "$git_dir/$x", "$git_dir/svn/$x" or
5393                               croak "$!: $x";
5394                }
5395                $migrated++;
5396        }
5397        command_close_pipe($fh, $ctx);
5398        print STDERR "Done migrating from a git-svn v1 layout\n";
5399        $migrated;
5400}
5401
5402sub read_old_urls {
5403        my ($l_map, $pfx, $path) = @_;
5404        my @dir;
5405        foreach (<$path/*>) {
5406                if (-r "$_/info/url") {
5407                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
5408                        my $ref_id = $pfx . basename $_;
5409                        my $url = ::file_to_s("$_/info/url");
5410                        $l_map->{$ref_id} = $url;
5411                } elsif (-d $_) {
5412                        push @dir, $_;
5413                }
5414        }
5415        foreach (@dir) {
5416                my $x = $_;
5417                $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
5418                read_old_urls($l_map, $x, $_);
5419        }
5420}
5421
5422sub migrate_from_v2 {
5423        my @cfg = command(qw/config -l/);
5424        return if grep /^svn-remote\..+\.url=/, @cfg;
5425        my %l_map;
5426        read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
5427        my $migrated = 0;
5428
5429        foreach my $ref_id (sort keys %l_map) {
5430                eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
5431                if ($@) {
5432                        Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
5433                }
5434                $migrated++;
5435        }
5436        $migrated;
5437}
5438
5439sub minimize_connections {
5440        my $r = Git::SVN::read_all_remotes();
5441        my $new_urls = {};
5442        my $root_repos = {};
5443        foreach my $repo_id (keys %$r) {
5444                my $url = $r->{$repo_id}->{url} or next;
5445                my $fetch = $r->{$repo_id}->{fetch} or next;
5446                my $ra = Git::SVN::Ra->new($url);
5447
5448                # skip existing cases where we already connect to the root
5449                if (($ra->{url} eq $ra->{repos_root}) ||
5450                    ($ra->{repos_root} eq $repo_id)) {
5451                        $root_repos->{$ra->{url}} = $repo_id;
5452                        next;
5453                }
5454
5455                my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
5456                my $root_path = $ra->{url};
5457                $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
5458                foreach my $path (keys %$fetch) {
5459                        my $ref_id = $fetch->{$path};
5460                        my $gs = Git::SVN->new($ref_id, $repo_id, $path);
5461
5462                        # make sure we can read when connecting to
5463                        # a higher level of a repository
5464                        my ($last_rev, undef) = $gs->last_rev_commit;
5465                        if (!defined $last_rev) {
5466                                $last_rev = eval {
5467                                        $root_ra->get_latest_revnum;
5468                                };
5469                                next if $@;
5470                        }
5471                        my $new = $root_path;
5472                        $new .= length $path ? "/$path" : '';
5473                        eval {
5474                                $root_ra->get_log([$new], $last_rev, $last_rev,
5475                                                  0, 0, 1, sub { });
5476                        };
5477                        next if $@;
5478                        $new_urls->{$ra->{repos_root}}->{$new} =
5479                                { ref_id => $ref_id,
5480                                  old_repo_id => $repo_id,
5481                                  old_path => $path };
5482                }
5483        }
5484
5485        my @emptied;
5486        foreach my $url (keys %$new_urls) {
5487                # see if we can re-use an existing [svn-remote "repo_id"]
5488                # instead of creating a(n ugly) new section:
5489                my $repo_id = $root_repos->{$url} || $url;
5490
5491                my $fetch = $new_urls->{$url};
5492                foreach my $path (keys %$fetch) {
5493                        my $x = $fetch->{$path};
5494                        Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
5495                        my $pfx = "svn-remote.$x->{old_repo_id}";
5496
5497                        my $old_fetch = quotemeta("$x->{old_path}:".
5498                                                  "refs/remotes/$x->{ref_id}");
5499                        command_noisy(qw/config --unset/,
5500                                      "$pfx.fetch", '^'. $old_fetch . '$');
5501                        delete $r->{$x->{old_repo_id}}->
5502                               {fetch}->{$x->{old_path}};
5503                        if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
5504                                command_noisy(qw/config --unset/,
5505                                              "$pfx.url");
5506                                push @emptied, $x->{old_repo_id}
5507                        }
5508                }
5509        }
5510        if (@emptied) {
5511                my $file = $ENV{GIT_CONFIG} || "$ENV{GIT_DIR}/config";
5512                print STDERR <<EOF;
5513The following [svn-remote] sections in your config file ($file) are empty
5514and can be safely removed:
5515EOF
5516                print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
5517        }
5518}
5519
5520sub migration_check {
5521        migrate_from_v0();
5522        migrate_from_v1();
5523        migrate_from_v2();
5524        minimize_connections() if $_minimize;
5525}
5526
5527package Git::IndexInfo;
5528use strict;
5529use warnings;
5530use Git qw/command_input_pipe command_close_pipe/;
5531
5532sub new {
5533        my ($class) = @_;
5534        my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
5535        bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
5536}
5537
5538sub remove {
5539        my ($self, $path) = @_;
5540        if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
5541                return ++$self->{nr};
5542        }
5543        undef;
5544}
5545
5546sub update {
5547        my ($self, $mode, $hash, $path) = @_;
5548        if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
5549                return ++$self->{nr};
5550        }
5551        undef;
5552}
5553
5554sub DESTROY {
5555        my ($self) = @_;
5556        command_close_pipe($self->{gui}, $self->{ctx});
5557}
5558
5559package Git::SVN::GlobSpec;
5560use strict;
5561use warnings;
5562
5563sub new {
5564        my ($class, $glob) = @_;
5565        my $re = $glob;
5566        $re =~ s!/+$!!g; # no need for trailing slashes
5567        $re =~ m!^([^*]*)(\*(?:/\*)*)([^*]*)$!;
5568        my $temp = $re;
5569        my ($left, $right) = ($1, $3);
5570        $re = $2;
5571        my $depth = $re =~ tr/*/*/;
5572        if ($depth != $temp =~ tr/*/*/) {
5573                die "Only one set of wildcard directories " .
5574                        "(e.g. '*' or '*/*/*') is supported: '$glob'\n";
5575        }
5576        if ($depth == 0) {
5577                die "One '*' is needed for glob: '$glob'\n";
5578        }
5579        $re =~ s!\*!\[^/\]*!g;
5580        $re = quotemeta($left) . "($re)" . quotemeta($right);
5581        if (length $left && !($left =~ s!/+$!!g)) {
5582                die "Missing trailing '/' on left side of: '$glob' ($left)\n";
5583        }
5584        if (length $right && !($right =~ s!^/+!!g)) {
5585                die "Missing leading '/' on right side of: '$glob' ($right)\n";
5586        }
5587        my $left_re = qr/^\/\Q$left\E(\/|$)/;
5588        bless { left => $left, right => $right, left_regex => $left_re,
5589                regex => qr/$re/, glob => $glob, depth => $depth }, $class;
5590}
5591
5592sub full_path {
5593        my ($self, $path) = @_;
5594        return (length $self->{left} ? "$self->{left}/" : '') .
5595               $path . (length $self->{right} ? "/$self->{right}" : '');
5596}
5597
5598__END__
5599
5600Data structures:
5601
5602
5603$remotes = { # returned by read_all_remotes()
5604        'svn' => {
5605                # svn-remote.svn.url=https://svn.musicpd.org
5606                url => 'https://svn.musicpd.org',
5607                # svn-remote.svn.fetch=mpd/trunk:trunk
5608                fetch => {
5609                        'mpd/trunk' => 'trunk',
5610                },
5611                # svn-remote.svn.tags=mpd/tags/*:tags/*
5612                tags => {
5613                        path => {
5614                                left => 'mpd/tags',
5615                                right => '',
5616                                regex => qr!mpd/tags/([^/]+)$!,
5617                                glob => 'tags/*',
5618                        },
5619                        ref => {
5620                                left => 'tags',
5621                                right => '',
5622                                regex => qr!tags/([^/]+)$!,
5623                                glob => 'tags/*',
5624                        },
5625                }
5626        }
5627};
5628
5629$log_entry hashref as returned by libsvn_log_entry()
5630{
5631        log => 'whitespace-formatted log entry
5632',                                              # trailing newline is preserved
5633        revision => '8',                        # integer
5634        date => '2004-02-24T17:01:44.108345Z',  # commit date
5635        author => 'committer name'
5636};
5637
5638
5639# this is generated by generate_diff();
5640@mods = array of diff-index line hashes, each element represents one line
5641        of diff-index output
5642
5643diff-index line ($m hash)
5644{
5645        mode_a => first column of diff-index output, no leading ':',
5646        mode_b => second column of diff-index output,
5647        sha1_b => sha1sum of the final blob,
5648        chg => change type [MCRADT],
5649        file_a => original file name of a file (iff chg is 'C' or 'R')
5650        file_b => new/current file name of a file (any chg)
5651}
5652;
5653
5654# retval of read_url_paths{,_all}();
5655$l_map = {
5656        # repository root url
5657        'https://svn.musicpd.org' => {
5658                # repository path               # GIT_SVN_ID
5659                'mpd/trunk'             =>      'trunk',
5660                'mpd/tags/0.11.5'       =>      'tags/0.11.5',
5661        },
5662}
5663
5664Notes:
5665        I don't trust the each() function on unless I created %hash myself
5666        because the internal iterator may not have started at base.