Andrew's git - gitweb.git/blob

   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
  24if (! exists $ENV{SVN_SSH}) {
  25        if (exists $ENV{GIT_SSH}) {
  26                $ENV{SVN_SSH} = $ENV{GIT_SSH};
  27                if ($^O eq 'msys') {
  28                        $ENV{SVN_SSH} =~ s/\\/\\\\/g;
  29                }
  30        }
  31}
  32
  33$Git::SVN::Log::TZ = $ENV{TZ};
  34$ENV{TZ} = 'UTC';
  35$| = 1; # unbuffer STDOUT
  36
  37sub fatal (@) { print STDERR "@_\n"; exit 1 }
  38require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
  39require SVN::Ra;
  40require SVN::Delta;
  41if ($SVN::Core::VERSION lt '1.1.0') {
  42        fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)";
  43}
  44my $can_compress = eval { require Compress::Zlib; 1};
  45push @Git::SVN::Ra::ISA, 'SVN::Ra';
  46push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
  47push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
  48use Carp qw/croak/;
  49use Digest::MD5;
  50use IO::File qw//;
  51use File::Basename qw/dirname basename/;
  52use File::Path qw/mkpath/;
  53use File::Spec;
  54use File::Find;
  55use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev/;
  56use IPC::Open3;
  57use Git;
  58
  59BEGIN {
  60        # import functions from Git into our packages, en masse
  61        no strict 'refs';
  62        foreach (qw/command command_oneline command_noisy command_output_pipe
  63                    command_input_pipe command_close_pipe
  64                    command_bidi_pipe command_close_bidi_pipe/) {
  65                for my $package ( qw(SVN::Git::Editor SVN::Git::Fetcher
  66                        Git::SVN::Migration Git::SVN::Log Git::SVN),
  67                        __PACKAGE__) {
  68                        *{"${package}::$_"} = \&{"Git::$_"};
  69                }
  70        }
  71}
  72
  73my ($SVN);
  74
  75$sha1 = qr/[a-f\d]{40}/;
  76$sha1_short = qr/[a-f\d]{4,40}/;
  77my ($_stdin, $_help, $_edit,
  78        $_message, $_file, $_branch_dest,
  79        $_template, $_shared,
  80        $_version, $_fetch_all, $_no_rebase, $_fetch_parent,
  81        $_merge, $_strategy, $_dry_run, $_local,
  82        $_prefix, $_no_checkout, $_url, $_verbose,
  83        $_git_format, $_commit_url, $_tag);
  84$Git::SVN::_follow_parent = 1;
  85$_q ||= 0;
  86my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
  87                    'config-dir=s' => \$Git::SVN::Ra::config_dir,
  88                    'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache,
  89                    'ignore-paths=s' => \$SVN::Git::Fetcher::_ignore_regex );
  90my %fc_opts = ( 'follow-parent|follow!' => \$Git::SVN::_follow_parent,
  91                'authors-file|A=s' => \$_authors,
  92                'authors-prog=s' => \$_authors_prog,
  93                'repack:i' => \$Git::SVN::_repack,
  94                'noMetadata' => \$Git::SVN::_no_metadata,
  95                'useSvmProps' => \$Git::SVN::_use_svm_props,
  96                'useSvnsyncProps' => \$Git::SVN::_use_svnsync_props,
  97                'log-window-size=i' => \$Git::SVN::Ra::_log_window_size,
  98                'no-checkout' => \$_no_checkout,
  99                'quiet|q+' => \$_q,
 100                'repack-flags|repack-args|repack-opts=s' =>
 101                   \$Git::SVN::_repack_flags,
 102                'use-log-author' => \$Git::SVN::_use_log_author,
 103                'add-author-from' => \$Git::SVN::_add_author_from,
 104                'localtime' => \$Git::SVN::_localtime,
 105                %remote_opts );
 106
 107my ($_trunk, @_tags, @_branches, $_stdlayout);
 108my %icv;
 109my %init_opts = ( 'template=s' => \$_template, 'shared:s' => \$_shared,
 110                  'trunk|T=s' => \$_trunk, 'tags|t=s@' => \@_tags,
 111                  'branches|b=s@' => \@_branches, 'prefix=s' => \$_prefix,
 112                  'stdlayout|s' => \$_stdlayout,
 113                  'minimize-url|m!' => \$Git::SVN::_minimize_url,
 114                  'no-metadata' => sub { $icv{noMetadata} = 1 },
 115                  'use-svm-props' => sub { $icv{useSvmProps} = 1 },
 116                  'use-svnsync-props' => sub { $icv{useSvnsyncProps} = 1 },
 117                  'rewrite-root=s' => sub { $icv{rewriteRoot} = $_[1] },
 118                  %remote_opts );
 119my %cmt_opts = ( 'edit|e' => \$_edit,
 120                'rmdir' => \$SVN::Git::Editor::_rmdir,
 121                'find-copies-harder' => \$SVN::Git::Editor::_find_copies_harder,
 122                'l=i' => \$SVN::Git::Editor::_rename_limit,
 123                'copy-similarity|C=i'=> \$SVN::Git::Editor::_cp_similarity
 124);
 125
 126my %cmd = (
 127        fetch => [ \&cmd_fetch, "Download new revisions from SVN",
 128                        { 'revision|r=s' => \$_revision,
 129                          'fetch-all|all' => \$_fetch_all,
 130                          'parent|p' => \$_fetch_parent,
 131                           %fc_opts } ],
 132        clone => [ \&cmd_clone, "Initialize and fetch revisions",
 133                        { 'revision|r=s' => \$_revision,
 134                           %fc_opts, %init_opts } ],
 135        init => [ \&cmd_init, "Initialize a repo for tracking" .
 136                          " (requires URL argument)",
 137                          \%init_opts ],
 138        'multi-init' => [ \&cmd_multi_init,
 139                          "Deprecated alias for ".
 140                          "'$0 init -T<trunk> -b<branches> -t<tags>'",
 141                          \%init_opts ],
 142        dcommit => [ \&cmd_dcommit,
 143                     'Commit several diffs to merge with upstream',
 144                        { 'merge|m|M' => \$_merge,
 145                          'strategy|s=s' => \$_strategy,
 146                          'verbose|v' => \$_verbose,
 147                          'dry-run|n' => \$_dry_run,
 148                          'fetch-all|all' => \$_fetch_all,
 149                          'commit-url=s' => \$_commit_url,
 150                          'revision|r=i' => \$_revision,
 151                          'no-rebase' => \$_no_rebase,
 152                        %cmt_opts, %fc_opts } ],
 153        branch => [ \&cmd_branch,
 154                    'Create a branch in the SVN repository',
 155                    { 'message|m=s' => \$_message,
 156                      'destination|d=s' => \$_branch_dest,
 157                      'dry-run|n' => \$_dry_run,
 158                      'tag|t' => \$_tag } ],
 159        tag => [ sub { $_tag = 1; cmd_branch(@_) },
 160                 'Create a tag in the SVN repository',
 161                 { 'message|m=s' => \$_message,
 162                   'destination|d=s' => \$_branch_dest,
 163                   'dry-run|n' => \$_dry_run } ],
 164        'set-tree' => [ \&cmd_set_tree,
 165                        "Set an SVN repository to a git tree-ish",
 166                        { 'stdin' => \$_stdin, %cmt_opts, %fc_opts, } ],
 167        'create-ignore' => [ \&cmd_create_ignore,
 168                             'Create a .gitignore per svn:ignore',
 169                             { 'revision|r=i' => \$_revision
 170                             } ],
 171        'propget' => [ \&cmd_propget,
 172                       'Print the value of a property on a file or directory',
 173                       { 'revision|r=i' => \$_revision } ],
 174        'proplist' => [ \&cmd_proplist,
 175                       'List all properties of a file or directory',
 176                       { 'revision|r=i' => \$_revision } ],
 177        'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
 178                        { 'revision|r=i' => \$_revision
 179                        } ],
 180        'show-externals' => [ \&cmd_show_externals, "Show svn:externals listings",
 181                        { 'revision|r=i' => \$_revision
 182                        } ],
 183        'multi-fetch' => [ \&cmd_multi_fetch,
 184                           "Deprecated alias for $0 fetch --all",
 185                           { 'revision|r=s' => \$_revision, %fc_opts } ],
 186        'migrate' => [ sub { },
 187                       # no-op, we automatically run this anyways,
 188                       'Migrate configuration/metadata/layout from
 189                        previous versions of git-svn',
 190                       { 'minimize' => \$Git::SVN::Migration::_minimize,
 191                         %remote_opts } ],
 192        'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
 193                        { 'limit=i' => \$Git::SVN::Log::limit,
 194                          'revision|r=s' => \$_revision,
 195                          'verbose|v' => \$Git::SVN::Log::verbose,
 196                          'incremental' => \$Git::SVN::Log::incremental,
 197                          'oneline' => \$Git::SVN::Log::oneline,
 198                          'show-commit' => \$Git::SVN::Log::show_commit,
 199                          'non-recursive' => \$Git::SVN::Log::non_recursive,
 200                          'authors-file|A=s' => \$_authors,
 201                          'color' => \$Git::SVN::Log::color,
 202                          'pager=s' => \$Git::SVN::Log::pager
 203                        } ],
 204        'find-rev' => [ \&cmd_find_rev,
 205                        "Translate between SVN revision numbers and tree-ish",
 206                        {} ],
 207        'rebase' => [ \&cmd_rebase, "Fetch and rebase your working directory",
 208                        { 'merge|m|M' => \$_merge,
 209                          'verbose|v' => \$_verbose,
 210                          'strategy|s=s' => \$_strategy,
 211                          'local|l' => \$_local,
 212                          'fetch-all|all' => \$_fetch_all,
 213                          'dry-run|n' => \$_dry_run,
 214                          %fc_opts } ],
 215        'commit-diff' => [ \&cmd_commit_diff,
 216                           'Commit a diff between two trees',
 217                        { 'message|m=s' => \$_message,
 218                          'file|F=s' => \$_file,
 219                          'revision|r=s' => \$_revision,
 220                        %cmt_opts } ],
 221        'info' => [ \&cmd_info,
 222                    "Show info about the latest SVN revision
 223                     on the current branch",
 224                    { 'url' => \$_url, } ],
 225        'blame' => [ \&Git::SVN::Log::cmd_blame,
 226                    "Show what revision and author last modified each line of a file",
 227                    { 'git-format' => \$_git_format } ],
 228        'reset' => [ \&cmd_reset,
 229                     "Undo fetches back to the specified SVN revision",
 230                     { 'revision|r=s' => \$_revision,
 231                       'parent|p' => \$_fetch_parent } ],
 232        'gc' => [ \&cmd_gc,
 233                  "Compress unhandled.log files in .git/svn and remove " .
 234                  "index files in .git/svn",
 235                {} ],
 236);
 237
 238my $cmd;
 239for (my $i = 0; $i < @ARGV; $i++) {
 240        if (defined $cmd{$ARGV[$i]}) {
 241                $cmd = $ARGV[$i];
 242                splice @ARGV, $i, 1;
 243                last;
 244        } elsif ($ARGV[$i] eq 'help') {
 245                $cmd = $ARGV[$i+1];
 246                usage(0);
 247        }
 248};
 249
 250# make sure we're always running at the top-level working directory
 251unless ($cmd && $cmd =~ /(?:clone|init|multi-init)$/) {
 252        unless (-d $ENV{GIT_DIR}) {
 253                if ($git_dir_user_set) {
 254                        die "GIT_DIR=$ENV{GIT_DIR} explicitly set, ",
 255                            "but it is not a directory\n";
 256                }
 257                my $git_dir = delete $ENV{GIT_DIR};
 258                my $cdup = undef;
 259                git_cmd_try {
 260                        $cdup = command_oneline(qw/rev-parse --show-cdup/);
 261                        $git_dir = '.' unless ($cdup);
 262                        chomp $cdup if ($cdup);
 263                        $cdup = "." unless ($cdup && length $cdup);
 264                } "Already at toplevel, but $git_dir not found\n";
 265                chdir $cdup or die "Unable to chdir up to '$cdup'\n";
 266                unless (-d $git_dir) {
 267                        die "$git_dir still not found after going to ",
 268                            "'$cdup'\n";
 269                }
 270                $ENV{GIT_DIR} = $git_dir;
 271        }
 272        $_repository = Git->repository(Repository => $ENV{GIT_DIR});
 273}
 274
 275my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
 276
 277read_repo_config(\%opts);
 278if ($cmd && ($cmd eq 'log' || $cmd eq 'blame')) {
 279        Getopt::Long::Configure('pass_through');
 280}
 281my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
 282                    'minimize-connections' => \$Git::SVN::Migration::_minimize,
 283                    'id|i=s' => \$Git::SVN::default_ref_id,
 284                    'svn-remote|remote|R=s' => sub {
 285                       $Git::SVN::no_reuse_existing = 1;
 286                       $Git::SVN::default_repo_id = $_[1] });
 287exit 1 if (!$rv && $cmd && $cmd ne 'log');
 288
 289usage(0) if $_help;
 290version() if $_version;
 291usage(1) unless defined $cmd;
 292load_authors() if $_authors;
 293if (defined $_authors_prog) {
 294        $_authors_prog = "'" . File::Spec->rel2abs($_authors_prog) . "'";
 295}
 296
 297unless ($cmd =~ /^(?:clone|init|multi-init|commit-diff)$/) {
 298        Git::SVN::Migration::migration_check();
 299}
 300Git::SVN::init_vars();
 301eval {
 302        Git::SVN::verify_remotes_sanity();
 303        $cmd{$cmd}->[0]->(@ARGV);
 304};
 305fatal $@ if $@;
 306post_fetch_checkout();
 307exit 0;
 308
 309####################### primary functions ######################
 310sub usage {
 311        my $exit = shift || 0;
 312        my $fd = $exit ? \*STDERR : \*STDOUT;
 313        print $fd <<"";
 314git-svn - bidirectional operations between a single Subversion tree and git
 315Usage: git svn <command> [options] [arguments]\n
 316
 317        print $fd "Available commands:\n" unless $cmd;
 318
 319        foreach (sort keys %cmd) {
 320                next if $cmd && $cmd ne $_;
 321                next if /^multi-/; # don't show deprecated commands
 322                print $fd '  ',pack('A17',$_),$cmd{$_}->[1],"\n";
 323                foreach (sort keys %{$cmd{$_}->[2]}) {
 324                        # mixed-case options are for .git/config only
 325                        next if /[A-Z]/ && /^[a-z]+$/i;
 326                        # prints out arguments as they should be passed:
 327                        my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
 328                        print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
 329                                                        "--$_" : "-$_" }
 330                                                split /\|/,$_)," $x\n";
 331                }
 332        }
 333        print $fd <<"";
 334\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
 335arbitrary identifier if you're tracking multiple SVN branches/repositories in
 336one git repository and want to keep them separate.  See git-svn(1) for more
 337information.
 338
 339        exit $exit;
 340}
 341
 342sub version {
 343        print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
 344        exit 0;
 345}
 346
 347sub do_git_init_db {
 348        unless (-d $ENV{GIT_DIR}) {
 349                my @init_db = ('init');
 350                push @init_db, "--template=$_template" if defined $_template;
 351                if (defined $_shared) {
 352                        if ($_shared =~ /[a-z]/) {
 353                                push @init_db, "--shared=$_shared";
 354                        } else {
 355                                push @init_db, "--shared";
 356                        }
 357                }
 358                command_noisy(@init_db);
 359                $_repository = Git->repository(Repository => ".git");
 360        }
 361        command_noisy('config', 'core.autocrlf', 'false');
 362        my $set;
 363        my $pfx = "svn-remote.$Git::SVN::default_repo_id";
 364        foreach my $i (keys %icv) {
 365                die "'$set' and '$i' cannot both be set\n" if $set;
 366                next unless defined $icv{$i};
 367                command_noisy('config', "$pfx.$i", $icv{$i});
 368                $set = $i;
 369        }
 370        my $ignore_regex = \$SVN::Git::Fetcher::_ignore_regex;
 371        command_noisy('config', "$pfx.ignore-paths", $$ignore_regex)
 372                if defined $$ignore_regex;
 373}
 374
 375sub init_subdir {
 376        my $repo_path = shift or return;
 377        mkpath([$repo_path]) unless -d $repo_path;
 378        chdir $repo_path or die "Couldn't chdir to $repo_path: $!\n";
 379        $ENV{GIT_DIR} = '.git';
 380        $_repository = Git->repository(Repository => $ENV{GIT_DIR});
 381}
 382
 383sub cmd_clone {
 384        my ($url, $path) = @_;
 385        if (!defined $path &&
 386            (defined $_trunk || @_branches || @_tags ||
 387             defined $_stdlayout) &&
 388            $url !~ m#^[a-z\+]+://#) {
 389                $path = $url;
 390        }
 391        $path = basename($url) if !defined $path || !length $path;
 392        cmd_init($url, $path);
 393        Git::SVN::fetch_all($Git::SVN::default_repo_id);
 394        command_oneline('config', 'svn.authorsfile', $_authors) if $_authors;
 395}
 396
 397sub cmd_init {
 398        if (defined $_stdlayout) {
 399                $_trunk = 'trunk' if (!defined $_trunk);
 400                @_tags = 'tags' if (! @_tags);
 401                @_branches = 'branches' if (! @_branches);
 402        }
 403        if (defined $_trunk || @_branches || @_tags) {
 404                return cmd_multi_init(@_);
 405        }
 406        my $url = shift or die "SVN repository location required ",
 407                               "as a command-line argument\n";
 408        $url = canonicalize_url($url);
 409        init_subdir(@_);
 410        do_git_init_db();
 411
 412        if ($Git::SVN::_minimize_url eq 'unset') {
 413                $Git::SVN::_minimize_url = 0;
 414        }
 415
 416        Git::SVN->init($url);
 417}
 418
 419sub cmd_fetch {
 420        if (grep /^\d+=./, @_) {
 421                die "'<rev>=<commit>' fetch arguments are ",
 422                    "no longer supported.\n";
 423        }
 424        my ($remote) = @_;
 425        if (@_ > 1) {
 426                die "Usage: $0 fetch [--all] [--parent] [svn-remote]\n";
 427        }
 428        if ($_fetch_parent) {
 429                my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 430                unless ($gs) {
 431                        die "Unable to determine upstream SVN information from ",
 432                            "working tree history\n";
 433                }
 434                # just fetch, don't checkout.
 435                $_no_checkout = 'true';
 436                $_fetch_all ? $gs->fetch_all : $gs->fetch;
 437        } elsif ($_fetch_all) {
 438                cmd_multi_fetch();
 439        } else {
 440                $remote ||= $Git::SVN::default_repo_id;
 441                Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
 442        }
 443}
 444
 445sub cmd_set_tree {
 446        my (@commits) = @_;
 447        if ($_stdin || !@commits) {
 448                print "Reading from stdin...\n";
 449                @commits = ();
 450                while (<STDIN>) {
 451                        if (/\b($sha1_short)\b/o) {
 452                                unshift @commits, $1;
 453                        }
 454                }
 455        }
 456        my @revs;
 457        foreach my $c (@commits) {
 458                my @tmp = command('rev-parse',$c);
 459                if (scalar @tmp == 1) {
 460                        push @revs, $tmp[0];
 461                } elsif (scalar @tmp > 1) {
 462                        push @revs, reverse(command('rev-list',@tmp));
 463                } else {
 464                        fatal "Failed to rev-parse $c";
 465                }
 466        }
 467        my $gs = Git::SVN->new;
 468        my ($r_last, $cmt_last) = $gs->last_rev_commit;
 469        $gs->fetch;
 470        if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
 471                fatal "There are new revisions that were fetched ",
 472                      "and need to be merged (or acknowledged) ",
 473                      "before committing.\nlast rev: $r_last\n",
 474                      " current: $gs->{last_rev}";
 475        }
 476        $gs->set_tree($_) foreach @revs;
 477        print "Done committing ",scalar @revs," revisions to SVN\n";
 478        unlink $gs->{index};
 479}
 480
 481sub cmd_dcommit {
 482        my $head = shift;
 483        git_cmd_try { command_oneline(qw/diff-index --quiet HEAD/) }
 484                'Cannot dcommit with a dirty index.  Commit your changes first, '
 485                . "or stash them with `git stash'.\n";
 486        $head ||= 'HEAD';
 487
 488        my $old_head;
 489        if ($head ne 'HEAD') {
 490                $old_head = eval {
 491                        command_oneline([qw/symbolic-ref -q HEAD/])
 492                };
 493                if ($old_head) {
 494                        $old_head =~ s{^refs/heads/}{};
 495                } else {
 496                        $old_head = eval { command_oneline(qw/rev-parse HEAD/) };
 497                }
 498                command(['checkout', $head], STDERR => 0);
 499        }
 500
 501        my @refs;
 502        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD', \@refs);
 503        unless ($gs) {
 504                die "Unable to determine upstream SVN information from ",
 505                    "$head history.\nPerhaps the repository is empty.";
 506        }
 507
 508        if (defined $_commit_url) {
 509                $url = $_commit_url;
 510        } else {
 511                $url = eval { command_oneline('config', '--get',
 512                              "svn-remote.$gs->{repo_id}.commiturl") };
 513                if (!$url) {
 514                        $url = $gs->full_url
 515                }
 516        }
 517
 518        my $last_rev = $_revision if defined $_revision;
 519        if ($url) {
 520                print "Committing to $url ...\n";
 521        }
 522        my ($linear_refs, $parents) = linearize_history($gs, \@refs);
 523        if ($_no_rebase && scalar(@$linear_refs) > 1) {
 524                warn "Attempting to commit more than one change while ",
 525                     "--no-rebase is enabled.\n",
 526                     "If these changes depend on each other, re-running ",
 527                     "without --no-rebase may be required."
 528        }
 529        my $expect_url = $url;
 530        Git::SVN::remove_username($expect_url);
 531        while (1) {
 532                my $d = shift @$linear_refs or last;
 533                unless (defined $last_rev) {
 534                        (undef, $last_rev, undef) = cmt_metadata("$d~1");
 535                        unless (defined $last_rev) {
 536                                fatal "Unable to extract revision information ",
 537                                      "from commit $d~1";
 538                        }
 539                }
 540                if ($_dry_run) {
 541                        print "diff-tree $d~1 $d\n";
 542                } else {
 543                        my $cmt_rev;
 544                        my %ed_opts = ( r => $last_rev,
 545                                        log => get_commit_entry($d)->{log},
 546                                        ra => Git::SVN::Ra->new($url),
 547                                        config => SVN::Core::config_get_config(
 548                                                $Git::SVN::Ra::config_dir
 549                                        ),
 550                                        tree_a => "$d~1",
 551                                        tree_b => $d,
 552                                        editor_cb => sub {
 553                                               print "Committed r$_[0]\n";
 554                                               $cmt_rev = $_[0];
 555                                        },
 556                                        svn_path => '');
 557                        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
 558                                print "No changes\n$d~1 == $d\n";
 559                        } elsif ($parents->{$d} && @{$parents->{$d}}) {
 560                                $gs->{inject_parents_dcommit}->{$cmt_rev} =
 561                                                               $parents->{$d};
 562                        }
 563                        $_fetch_all ? $gs->fetch_all : $gs->fetch;
 564                        $last_rev = $cmt_rev;
 565                        next if $_no_rebase;
 566
 567                        # we always want to rebase against the current HEAD,
 568                        # not any head that was passed to us
 569                        my @diff = command('diff-tree', $d,
 570                                           $gs->refname, '--');
 571                        my @finish;
 572                        if (@diff) {
 573                                @finish = rebase_cmd();
 574                                print STDERR "W: $d and ", $gs->refname,
 575                                             " differ, using @finish:\n",
 576                                             join("\n", @diff), "\n";
 577                        } else {
 578                                print "No changes between current HEAD and ",
 579                                      $gs->refname,
 580                                      "\nResetting to the latest ",
 581                                      $gs->refname, "\n";
 582                                @finish = qw/reset --mixed/;
 583                        }
 584                        command_noisy(@finish, $gs->refname);
 585                        if (@diff) {
 586                                @refs = ();
 587                                my ($url_, $rev_, $uuid_, $gs_) =
 588                                              working_head_info('HEAD', \@refs);
 589                                my ($linear_refs_, $parents_) =
 590                                              linearize_history($gs_, \@refs);
 591                                if (scalar(@$linear_refs) !=
 592                                    scalar(@$linear_refs_)) {
 593                                        fatal "# of revisions changed ",
 594                                          "\nbefore:\n",
 595                                          join("\n", @$linear_refs),
 596                                          "\n\nafter:\n",
 597                                          join("\n", @$linear_refs_), "\n",
 598                                          'If you are attempting to commit ',
 599                                          "merges, try running:\n\t",
 600                                          'git rebase --interactive',
 601                                          '--preserve-merges ',
 602                                          $gs->refname,
 603                                          "\nBefore dcommitting";
 604                                }
 605                                if ($url_ ne $expect_url) {
 606                                        fatal "URL mismatch after rebase: ",
 607                                              "$url_ != $expect_url";
 608                                }
 609                                if ($uuid_ ne $uuid) {
 610                                        fatal "uuid mismatch after rebase: ",
 611                                              "$uuid_ != $uuid";
 612                                }
 613                                # remap parents
 614                                my (%p, @l, $i);
 615                                for ($i = 0; $i < scalar @$linear_refs; $i++) {
 616                                        my $new = $linear_refs_->[$i] or next;
 617                                        $p{$new} =
 618                                                $parents->{$linear_refs->[$i]};
 619                                        push @l, $new;
 620                                }
 621                                $parents = \%p;
 622                                $linear_refs = \@l;
 623                        }
 624                }
 625        }
 626
 627        if ($old_head) {
 628                my $new_head = command_oneline(qw/rev-parse HEAD/);
 629                my $new_is_symbolic = eval {
 630                        command_oneline(qw/symbolic-ref -q HEAD/);
 631                };
 632                if ($new_is_symbolic) {
 633                        print "dcommitted the branch ", $head, "\n";
 634                } else {
 635                        print "dcommitted on a detached HEAD because you gave ",
 636                              "a revision argument.\n",
 637                              "The rewritten commit is: ", $new_head, "\n";
 638                }
 639                command(['checkout', $old_head], STDERR => 0);
 640        }
 641
 642        unlink $gs->{index};
 643}
 644
 645sub cmd_branch {
 646        my ($branch_name, $head) = @_;
 647
 648        unless (defined $branch_name && length $branch_name) {
 649                die(($_tag ? "tag" : "branch") . " name required\n");
 650        }
 651        $head ||= 'HEAD';
 652
 653        my ($src, $rev, undef, $gs) = working_head_info($head);
 654
 655        my $remote = Git::SVN::read_all_remotes()->{$gs->{repo_id}};
 656        my $allglobs = $remote->{ $_tag ? 'tags' : 'branches' };
 657        my $glob;
 658        if ($#{$allglobs} == 0) {
 659                $glob = $allglobs->[0];
 660        } else {
 661                unless(defined $_branch_dest) {
 662                        die "Multiple ",
 663                            $_tag ? "tag" : "branch",
 664                            " paths defined for Subversion repository.\n",
 665                            "You must specify where you want to create the ",
 666                            $_tag ? "tag" : "branch",
 667                            " with the --destination argument.\n";
 668                }
 669                foreach my $g (@{$allglobs}) {
 670                        # SVN::Git::Editor could probably be moved to Git.pm..
 671                        my $re = SVN::Git::Editor::glob2pat($g->{path}->{left});
 672                        if ($_branch_dest =~ /$re/) {
 673                                $glob = $g;
 674                                last;
 675                        }
 676                }
 677                unless (defined $glob) {
 678                        my $dest_re = qr/\b\Q$_branch_dest\E\b/;
 679                        foreach my $g (@{$allglobs}) {
 680                                $g->{path}->{left} =~ /$dest_re/ or next;
 681                                if (defined $glob) {
 682                                        die "Ambiguous destination: ",
 683                                            $_branch_dest, "\nmatches both '",
 684                                            $glob->{path}->{left}, "' and '",
 685                                            $g->{path}->{left}, "'\n";
 686                                }
 687                                $glob = $g;
 688                        }
 689                        unless (defined $glob) {
 690                                die "Unknown ",
 691                                    $_tag ? "tag" : "branch",
 692                                    " destination $_branch_dest\n";
 693                        }
 694                }
 695        }
 696        my ($lft, $rgt) = @{ $glob->{path} }{qw/left right/};
 697        my $dst = join '/', $remote->{url}, $lft, $branch_name, ($rgt || ());
 698
 699        my $ctx = SVN::Client->new(
 700                auth    => Git::SVN::Ra::_auth_providers(),
 701                log_msg => sub {
 702                        ${ $_[0] } = defined $_message
 703                                ? $_message
 704                                : 'Create ' . ($_tag ? 'tag ' : 'branch ' )
 705                                . $branch_name;
 706                },
 707        );
 708
 709        eval {
 710                $ctx->ls($dst, 'HEAD', 0);
 711        } and die "branch ${branch_name} already exists\n";
 712
 713        print "Copying ${src} at r${rev} to ${dst}...\n";
 714        $ctx->copy($src, $rev, $dst)
 715                unless $_dry_run;
 716
 717        $gs->fetch_all;
 718}
 719
 720sub cmd_find_rev {
 721        my $revision_or_hash = shift or die "SVN or git revision required ",
 722                                            "as a command-line argument\n";
 723        my $result;
 724        if ($revision_or_hash =~ /^r\d+$/) {
 725                my $head = shift;
 726                $head ||= 'HEAD';
 727                my @refs;
 728                my (undef, undef, $uuid, $gs) = working_head_info($head, \@refs);
 729                unless ($gs) {
 730                        die "Unable to determine upstream SVN information from ",
 731                            "$head history\n";
 732                }
 733                my $desired_revision = substr($revision_or_hash, 1);
 734                $result = $gs->rev_map_get($desired_revision, $uuid);
 735        } else {
 736                my (undef, $rev, undef) = cmt_metadata($revision_or_hash);
 737                $result = $rev;
 738        }
 739        print "$result\n" if $result;
 740}
 741
 742sub cmd_rebase {
 743        command_noisy(qw/update-index --refresh/);
 744        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 745        unless ($gs) {
 746                die "Unable to determine upstream SVN information from ",
 747                    "working tree history\n";
 748        }
 749        if ($_dry_run) {
 750                print "Remote Branch: " . $gs->refname . "\n";
 751                print "SVN URL: " . $url . "\n";
 752                return;
 753        }
 754        if (command(qw/diff-index HEAD --/)) {
 755                print STDERR "Cannot rebase with uncommited changes:\n";
 756                command_noisy('status');
 757                exit 1;
 758        }
 759        unless ($_local) {
 760                # rebase will checkout for us, so no need to do it explicitly
 761                $_no_checkout = 'true';
 762                $_fetch_all ? $gs->fetch_all : $gs->fetch;
 763        }
 764        command_noisy(rebase_cmd(), $gs->refname);
 765}
 766
 767sub cmd_show_ignore {
 768        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 769        $gs ||= Git::SVN->new;
 770        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 771        $gs->prop_walk($gs->{path}, $r, sub {
 772                my ($gs, $path, $props) = @_;
 773                print STDOUT "\n# $path\n";
 774                my $s = $props->{'svn:ignore'} or return;
 775                $s =~ s/[\r\n]+/\n/g;
 776                $s =~ s/^\n+//;
 777                chomp $s;
 778                $s =~ s#^#$path#gm;
 779                print STDOUT "$s\n";
 780        });
 781}
 782
 783sub cmd_show_externals {
 784        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 785        $gs ||= Git::SVN->new;
 786        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 787        $gs->prop_walk($gs->{path}, $r, sub {
 788                my ($gs, $path, $props) = @_;
 789                print STDOUT "\n# $path\n";
 790                my $s = $props->{'svn:externals'} or return;
 791                $s =~ s/[\r\n]+/\n/g;
 792                chomp $s;
 793                $s =~ s#^#$path#gm;
 794                print STDOUT "$s\n";
 795        });
 796}
 797
 798sub cmd_create_ignore {
 799        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 800        $gs ||= Git::SVN->new;
 801        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 802        $gs->prop_walk($gs->{path}, $r, sub {
 803                my ($gs, $path, $props) = @_;
 804                # $path is of the form /path/to/dir/
 805                $path = '.' . $path;
 806                # SVN can have attributes on empty directories,
 807                # which git won't track
 808                mkpath([$path]) unless -d $path;
 809                my $ignore = $path . '.gitignore';
 810                my $s = $props->{'svn:ignore'} or return;
 811                open(GITIGNORE, '>', $ignore)
 812                  or fatal("Failed to open `$ignore' for writing: $!");
 813                $s =~ s/[\r\n]+/\n/g;
 814                $s =~ s/^\n+//;
 815                chomp $s;
 816                # Prefix all patterns so that the ignore doesn't apply
 817                # to sub-directories.
 818                $s =~ s#^#/#gm;
 819                print GITIGNORE "$s\n";
 820                close(GITIGNORE)
 821                  or fatal("Failed to close `$ignore': $!");
 822                command_noisy('add', '-f', $ignore);
 823        });
 824}
 825
 826sub canonicalize_path {
 827        my ($path) = @_;
 828        my $dot_slash_added = 0;
 829        if (substr($path, 0, 1) ne "/") {
 830                $path = "./" . $path;
 831                $dot_slash_added = 1;
 832        }
 833        # File::Spec->canonpath doesn't collapse x/../y into y (for a
 834        # good reason), so let's do this manually.
 835        $path =~ s#/+#/#g;
 836        $path =~ s#/\.(?:/|$)#/#g;
 837        $path =~ s#/[^/]+/\.\.##g;
 838        $path =~ s#/$##g;
 839        $path =~ s#^\./## if $dot_slash_added;
 840        $path =~ s#^/##;
 841        $path =~ s#^\.$##;
 842        return $path;
 843}
 844
 845sub canonicalize_url {
 846        my ($url) = @_;
 847        $url =~ s#^([^:]+://[^/]*/)(.*)$#$1 . canonicalize_path($2)#e;
 848        return $url;
 849}
 850
 851# get_svnprops(PATH)
 852# ------------------
 853# Helper for cmd_propget and cmd_proplist below.
 854sub get_svnprops {
 855        my $path = shift;
 856        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 857        $gs ||= Git::SVN->new;
 858
 859        # prefix THE PATH by the sub-directory from which the user
 860        # invoked us.
 861        $path = $cmd_dir_prefix . $path;
 862        fatal("No such file or directory: $path") unless -e $path;
 863        my $is_dir = -d $path ? 1 : 0;
 864        $path = $gs->{path} . '/' . $path;
 865
 866        # canonicalize the path (otherwise libsvn will abort or fail to
 867        # find the file)
 868        $path = canonicalize_path($path);
 869
 870        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 871        my $props;
 872        if ($is_dir) {
 873                (undef, undef, $props) = $gs->ra->get_dir($path, $r);
 874        }
 875        else {
 876                (undef, $props) = $gs->ra->get_file($path, $r, undef);
 877        }
 878        return $props;
 879}
 880
 881# cmd_propget (PROP, PATH)
 882# ------------------------
 883# Print the SVN property PROP for PATH.
 884sub cmd_propget {
 885        my ($prop, $path) = @_;
 886        $path = '.' if not defined $path;
 887        usage(1) if not defined $prop;
 888        my $props = get_svnprops($path);
 889        if (not defined $props->{$prop}) {
 890                fatal("`$path' does not have a `$prop' SVN property.");
 891        }
 892        print $props->{$prop} . "\n";
 893}
 894
 895# cmd_proplist (PATH)
 896# -------------------
 897# Print the list of SVN properties for PATH.
 898sub cmd_proplist {
 899        my $path = shift;
 900        $path = '.' if not defined $path;
 901        my $props = get_svnprops($path);
 902        print "Properties on '$path':\n";
 903        foreach (sort keys %{$props}) {
 904                print "  $_\n";
 905        }
 906}
 907
 908sub cmd_multi_init {
 909        my $url = shift;
 910        unless (defined $_trunk || @_branches || @_tags) {
 911                usage(1);
 912        }
 913
 914        $_prefix = '' unless defined $_prefix;
 915        if (defined $url) {
 916                $url = canonicalize_url($url);
 917                init_subdir(@_);
 918        }
 919        do_git_init_db();
 920        if (defined $_trunk) {
 921                my $trunk_ref = 'refs/remotes/' . $_prefix . 'trunk';
 922                # try both old-style and new-style lookups:
 923                my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
 924                unless ($gs_trunk) {
 925                        my ($trunk_url, $trunk_path) =
 926                                              complete_svn_url($url, $_trunk);
 927                        $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
 928                                                   undef, $trunk_ref);
 929                }
 930        }
 931        return unless @_branches || @_tags;
 932        my $ra = $url ? Git::SVN::Ra->new($url) : undef;
 933        foreach my $path (@_branches) {
 934                complete_url_ls_init($ra, $path, '--branches/-b', $_prefix);
 935        }
 936        foreach my $path (@_tags) {
 937                complete_url_ls_init($ra, $path, '--tags/-t', $_prefix.'tags/');
 938        }
 939}
 940
 941sub cmd_multi_fetch {
 942        my $remotes = Git::SVN::read_all_remotes();
 943        foreach my $repo_id (sort keys %$remotes) {
 944                if ($remotes->{$repo_id}->{url}) {
 945                        Git::SVN::fetch_all($repo_id, $remotes);
 946                }
 947        }
 948}
 949
 950# this command is special because it requires no metadata
 951sub cmd_commit_diff {
 952        my ($ta, $tb, $url) = @_;
 953        my $usage = "Usage: $0 commit-diff -r<revision> ".
 954                    "<tree-ish> <tree-ish> [<URL>]";
 955        fatal($usage) if (!defined $ta || !defined $tb);
 956        my $svn_path = '';
 957        if (!defined $url) {
 958                my $gs = eval { Git::SVN->new };
 959                if (!$gs) {
 960                        fatal("Needed URL or usable git-svn --id in ",
 961                              "the command-line\n", $usage);
 962                }
 963                $url = $gs->{url};
 964                $svn_path = $gs->{path};
 965        }
 966        unless (defined $_revision) {
 967                fatal("-r|--revision is a required argument\n", $usage);
 968        }
 969        if (defined $_message && defined $_file) {
 970                fatal("Both --message/-m and --file/-F specified ",
 971                      "for the commit message.\n",
 972                      "I have no idea what you mean");
 973        }
 974        if (defined $_file) {
 975                $_message = file_to_s($_file);
 976        } else {
 977                $_message ||= get_commit_entry($tb)->{log};
 978        }
 979        my $ra ||= Git::SVN::Ra->new($url);
 980        my $r = $_revision;
 981        if ($r eq 'HEAD') {
 982                $r = $ra->get_latest_revnum;
 983        } elsif ($r !~ /^\d+$/) {
 984                die "revision argument: $r not understood by git-svn\n";
 985        }
 986        my %ed_opts = ( r => $r,
 987                        log => $_message,
 988                        ra => $ra,
 989                        tree_a => $ta,
 990                        tree_b => $tb,
 991                        editor_cb => sub { print "Committed r$_[0]\n" },
 992                        svn_path => $svn_path );
 993        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
 994                print "No changes\n$ta == $tb\n";
 995        }
 996}
 997
 998sub escape_uri_only {
 999        my ($uri) = @_;
1000        my @tmp;
1001        foreach (split m{/}, $uri) {
1002                s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
1003                push @tmp, $_;
1004        }
1005        join('/', @tmp);
1006}
1007
1008sub escape_url {
1009        my ($url) = @_;
1010        if ($url =~ m#^([^:]+)://([^/]*)(.*)$#) {
1011                my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
1012                $url = "$scheme://$domain$uri";
1013        }
1014        $url;
1015}
1016
1017sub cmd_info {
1018        my $path = canonicalize_path(defined($_[0]) ? $_[0] : ".");
1019        my $fullpath = canonicalize_path($cmd_dir_prefix . $path);
1020        if (exists $_[1]) {
1021                die "Too many arguments specified\n";
1022        }
1023
1024        my ($file_type, $diff_status) = find_file_type_and_diff_status($path);
1025
1026        if (!$file_type && !$diff_status) {
1027                print STDERR "svn: '$path' is not under version control\n";
1028                exit 1;
1029        }
1030
1031        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1032        unless ($gs) {
1033                die "Unable to determine upstream SVN information from ",
1034                    "working tree history\n";
1035        }
1036
1037        # canonicalize_path() will return "" to make libsvn 1.5.x happy,
1038        $path = "." if $path eq "";
1039
1040        my $full_url = $url . ($fullpath eq "" ? "" : "/$fullpath");
1041
1042        if ($_url) {
1043                print escape_url($full_url), "\n";
1044                return;
1045        }
1046
1047        my $result = "Path: $path\n";
1048        $result .= "Name: " . basename($path) . "\n" if $file_type ne "dir";
1049        $result .= "URL: " . escape_url($full_url) . "\n";
1050
1051        eval {
1052                my $repos_root = $gs->repos_root;
1053                Git::SVN::remove_username($repos_root);
1054                $result .= "Repository Root: " . escape_url($repos_root) . "\n";
1055        };
1056        if ($@) {
1057                $result .= "Repository Root: (offline)\n";
1058        }
1059        $result .= "Repository UUID: $uuid\n" unless $diff_status eq "A" &&
1060                ($SVN::Core::VERSION le '1.5.4' || $file_type ne "dir");
1061        $result .= "Revision: " . ($diff_status eq "A" ? 0 : $rev) . "\n";
1062
1063        $result .= "Node Kind: " .
1064                   ($file_type eq "dir" ? "directory" : "file") . "\n";
1065
1066        my $schedule = $diff_status eq "A"
1067                       ? "add"
1068                       : ($diff_status eq "D" ? "delete" : "normal");
1069        $result .= "Schedule: $schedule\n";
1070
1071        if ($diff_status eq "A") {
1072                print $result, "\n";
1073                return;
1074        }
1075
1076        my ($lc_author, $lc_rev, $lc_date_utc);
1077        my @args = Git::SVN::Log::git_svn_log_cmd($rev, $rev, "--", $fullpath);
1078        my $log = command_output_pipe(@args);
1079        my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
1080        while (<$log>) {
1081                if (/^${esc_color}author (.+) <[^>]+> (\d+) ([\-\+]?\d+)$/o) {
1082                        $lc_author = $1;
1083                        $lc_date_utc = Git::SVN::Log::parse_git_date($2, $3);
1084                } elsif (/^${esc_color}    (git-svn-id:.+)$/o) {
1085                        (undef, $lc_rev, undef) = ::extract_metadata($1);
1086                }
1087        }
1088        close $log;
1089
1090        Git::SVN::Log::set_local_timezone();
1091
1092        $result .= "Last Changed Author: $lc_author\n";
1093        $result .= "Last Changed Rev: $lc_rev\n";
1094        $result .= "Last Changed Date: " .
1095                   Git::SVN::Log::format_svn_date($lc_date_utc) . "\n";
1096
1097        if ($file_type ne "dir") {
1098                my $text_last_updated_date =
1099                    ($diff_status eq "D" ? $lc_date_utc : (stat $path)[9]);
1100                $result .=
1101                    "Text Last Updated: " .
1102                    Git::SVN::Log::format_svn_date($text_last_updated_date) .
1103                    "\n";
1104                my $checksum;
1105                if ($diff_status eq "D") {
1106                        my ($fh, $ctx) =
1107                            command_output_pipe(qw(cat-file blob), "HEAD:$path");
1108                        if ($file_type eq "link") {
1109                                my $file_name = <$fh>;
1110                                $checksum = md5sum("link $file_name");
1111                        } else {
1112                                $checksum = md5sum($fh);
1113                        }
1114                        command_close_pipe($fh, $ctx);
1115                } elsif ($file_type eq "link") {
1116                        my $file_name =
1117                            command(qw(cat-file blob), "HEAD:$path");
1118                        $checksum =
1119                            md5sum("link " . $file_name);
1120                } else {
1121                        open FILE, "<", $path or die $!;
1122                        $checksum = md5sum(\*FILE);
1123                        close FILE or die $!;
1124                }
1125                $result .= "Checksum: " . $checksum . "\n";
1126        }
1127
1128        print $result, "\n";
1129}
1130
1131sub cmd_reset {
1132        my $target = shift || $_revision or die "SVN revision required\n";
1133        $target = $1 if $target =~ /^r(\d+)$/;
1134        $target =~ /^\d+$/ or die "Numeric SVN revision expected\n";
1135        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1136        unless ($gs) {
1137                die "Unable to determine upstream SVN information from ".
1138                    "history\n";
1139        }
1140        my ($r, $c) = $gs->find_rev_before($target, not $_fetch_parent);
1141        $gs->rev_map_set($r, $c, 'reset', $uuid);
1142        print "r$r = $c ($gs->{ref_id})\n";
1143}
1144
1145sub cmd_gc {
1146        if (!$can_compress) {
1147                warn "Compress::Zlib could not be found; unhandled.log " .
1148                     "files will not be compressed.\n";
1149        }
1150        find({ wanted => \&gc_directory, no_chdir => 1}, "$ENV{GIT_DIR}/svn");
1151}
1152
1153########################### utility functions #########################
1154
1155sub rebase_cmd {
1156        my @cmd = qw/rebase/;
1157        push @cmd, '-v' if $_verbose;
1158        push @cmd, qw/--merge/ if $_merge;
1159        push @cmd, "--strategy=$_strategy" if $_strategy;
1160        @cmd;
1161}
1162
1163sub post_fetch_checkout {
1164        return if $_no_checkout;
1165        my $gs = $Git::SVN::_head or return;
1166        return if verify_ref('refs/heads/master^0');
1167
1168        # look for "trunk" ref if it exists
1169        my $remote = Git::SVN::read_all_remotes()->{$gs->{repo_id}};
1170        my $fetch = $remote->{fetch};
1171        if ($fetch) {
1172                foreach my $p (keys %$fetch) {
1173                        basename($fetch->{$p}) eq 'trunk' or next;
1174                        $gs = Git::SVN->new($fetch->{$p}, $gs->{repo_id}, $p);
1175                        last;
1176                }
1177        }
1178
1179        my $valid_head = verify_ref('HEAD^0');
1180        command_noisy(qw(update-ref refs/heads/master), $gs->refname);
1181        return if ($valid_head || !verify_ref('HEAD^0'));
1182
1183        return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
1184        my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
1185        return if -f $index;
1186
1187        return if command_oneline(qw/rev-parse --is-inside-work-tree/) eq 'false';
1188        return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
1189        command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
1190        print STDERR "Checked out HEAD:\n  ",
1191                     $gs->full_url, " r", $gs->last_rev, "\n";
1192}
1193
1194sub complete_svn_url {
1195        my ($url, $path) = @_;
1196        $path =~ s#/+$##;
1197        if ($path !~ m#^[a-z\+]+://#) {
1198                if (!defined $url || $url !~ m#^[a-z\+]+://#) {
1199                        fatal("E: '$path' is not a complete URL ",
1200                              "and a separate URL is not specified");
1201                }
1202                return ($url, $path);
1203        }
1204        return ($path, '');
1205}
1206
1207sub complete_url_ls_init {
1208        my ($ra, $repo_path, $switch, $pfx) = @_;
1209        unless ($repo_path) {
1210                print STDERR "W: $switch not specified\n";
1211                return;
1212        }
1213        $repo_path =~ s#/+$##;
1214        if ($repo_path =~ m#^[a-z\+]+://#) {
1215                $ra = Git::SVN::Ra->new($repo_path);
1216                $repo_path = '';
1217        } else {
1218                $repo_path =~ s#^/+##;
1219                unless ($ra) {
1220                        fatal("E: '$repo_path' is not a complete URL ",
1221                              "and a separate URL is not specified");
1222                }
1223        }
1224        my $url = $ra->{url};
1225        my $gs = Git::SVN->init($url, undef, undef, undef, 1);
1226        my $k = "svn-remote.$gs->{repo_id}.url";
1227        my $orig_url = eval { command_oneline(qw/config --get/, $k) };
1228        if ($orig_url && ($orig_url ne $gs->{url})) {
1229                die "$k already set: $orig_url\n",
1230                    "wanted to set to: $gs->{url}\n";
1231        }
1232        command_oneline('config', $k, $gs->{url}) unless $orig_url;
1233        my $remote_path = "$gs->{path}/$repo_path";
1234        $remote_path =~ s{%([0-9A-F]{2})}{chr hex($1)}ieg;
1235        $remote_path =~ s#/+#/#g;
1236        $remote_path =~ s#^/##g;
1237        $remote_path .= "/*" if $remote_path !~ /\*/;
1238        my ($n) = ($switch =~ /^--(\w+)/);
1239        if (length $pfx && $pfx !~ m#/$#) {
1240                die "--prefix='$pfx' must have a trailing slash '/'\n";
1241        }
1242        command_noisy('config',
1243                      '--add',
1244                      "svn-remote.$gs->{repo_id}.$n",
1245                      "$remote_path:refs/remotes/$pfx*" .
1246                        ('/*' x (($remote_path =~ tr/*/*/) - 1)) );
1247}
1248
1249sub verify_ref {
1250        my ($ref) = @_;
1251        eval { command_oneline([ 'rev-parse', '--verify', $ref ],
1252                               { STDERR => 0 }); };
1253}
1254
1255sub get_tree_from_treeish {
1256        my ($treeish) = @_;
1257        # $treeish can be a symbolic ref, too:
1258        my $type = command_oneline(qw/cat-file -t/, $treeish);
1259        my $expected;
1260        while ($type eq 'tag') {
1261                ($treeish, $type) = command(qw/cat-file tag/, $treeish);
1262        }
1263        if ($type eq 'commit') {
1264                $expected = (grep /^tree /, command(qw/cat-file commit/,
1265                                                    $treeish))[0];
1266                ($expected) = ($expected =~ /^tree ($sha1)$/o);
1267                die "Unable to get tree from $treeish\n" unless $expected;
1268        } elsif ($type eq 'tree') {
1269                $expected = $treeish;
1270        } else {
1271                die "$treeish is a $type, expected tree, tag or commit\n";
1272        }
1273        return $expected;
1274}
1275
1276sub get_commit_entry {
1277        my ($treeish) = shift;
1278        my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
1279        my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
1280        my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
1281        open my $log_fh, '>', $commit_editmsg or croak $!;
1282
1283        my $type = command_oneline(qw/cat-file -t/, $treeish);
1284        if ($type eq 'commit' || $type eq 'tag') {
1285                my ($msg_fh, $ctx) = command_output_pipe('cat-file',
1286                                                         $type, $treeish);
1287                my $in_msg = 0;
1288                my $author;
1289                my $saw_from = 0;
1290                my $msgbuf = "";
1291                while (<$msg_fh>) {
1292                        if (!$in_msg) {
1293                                $in_msg = 1 if (/^\s*$/);
1294                                $author = $1 if (/^author (.*>)/);
1295                        } elsif (/^git-svn-id: /) {
1296                                # skip this for now, we regenerate the
1297                                # correct one on re-fetch anyways
1298                                # TODO: set *:merge properties or like...
1299                        } else {
1300                                if (/^From:/ || /^Signed-off-by:/) {
1301                                        $saw_from = 1;
1302                                }
1303                                $msgbuf .= $_;
1304                        }
1305                }
1306                $msgbuf =~ s/\s+$//s;
1307                if ($Git::SVN::_add_author_from && defined($author)
1308                    && !$saw_from) {
1309                        $msgbuf .= "\n\nFrom: $author";
1310                }
1311                print $log_fh $msgbuf or croak $!;
1312                command_close_pipe($msg_fh, $ctx);
1313        }
1314        close $log_fh or croak $!;
1315
1316        if ($_edit || ($type eq 'tree')) {
1317                my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
1318                # TODO: strip out spaces, comments, like git-commit.sh
1319                system($editor, $commit_editmsg);
1320        }
1321        rename $commit_editmsg, $commit_msg or croak $!;
1322        {
1323                require Encode;
1324                # SVN requires messages to be UTF-8 when entering the repo
1325                local $/;
1326                open $log_fh, '<', $commit_msg or croak $!;
1327                binmode $log_fh;
1328                chomp($log_entry{log} = <$log_fh>);
1329
1330                my $enc = Git::config('i18n.commitencoding') || 'UTF-8';
1331                my $msg = $log_entry{log};
1332
1333                eval { $msg = Encode::decode($enc, $msg, 1) };
1334                if ($@) {
1335                        die "Could not decode as $enc:\n", $msg,
1336                            "\nPerhaps you need to set i18n.commitencoding\n";
1337                }
1338
1339                eval { $msg = Encode::encode('UTF-8', $msg, 1) };
1340                die "Could not encode as UTF-8:\n$msg\n" if $@;
1341
1342                $log_entry{log} = $msg;
1343
1344                close $log_fh or croak $!;
1345        }
1346        unlink $commit_msg;
1347        \%log_entry;
1348}
1349
1350sub s_to_file {
1351        my ($str, $file, $mode) = @_;
1352        open my $fd,'>',$file or croak $!;
1353        print $fd $str,"\n" or croak $!;
1354        close $fd or croak $!;
1355        chmod ($mode &~ umask, $file) if (defined $mode);
1356}
1357
1358sub file_to_s {
1359        my $file = shift;
1360        open my $fd,'<',$file or croak "$!: file: $file\n";
1361        local $/;
1362        my $ret = <$fd>;
1363        close $fd or croak $!;
1364        $ret =~ s/\s*$//s;
1365        return $ret;
1366}
1367
1368# '<svn username> = real-name <email address>' mapping based on git-svnimport:
1369sub load_authors {
1370        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1371        my $log = $cmd eq 'log';
1372        while (<$authors>) {
1373                chomp;
1374                next unless /^(.+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
1375                my ($user, $name, $email) = ($1, $2, $3);
1376                if ($log) {
1377                        $Git::SVN::Log::rusers{"$name <$email>"} = $user;
1378                } else {
1379                        $users{$user} = [$name, $email];
1380                }
1381        }
1382        close $authors or croak $!;
1383}
1384
1385# convert GetOpt::Long specs for use by git-config
1386sub read_repo_config {
1387        return unless -d $ENV{GIT_DIR};
1388        my $opts = shift;
1389        my @config_only;
1390        foreach my $o (keys %$opts) {
1391                # if we have mixedCase and a long option-only, then
1392                # it's a config-only variable that we don't need for
1393                # the command-line.
1394                push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
1395                my $v = $opts->{$o};
1396                my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
1397                $key =~ s/-//g;
1398                my $arg = 'git config';
1399                $arg .= ' --int' if ($o =~ /[:=]i$/);
1400                $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
1401                if (ref $v eq 'ARRAY') {
1402                        chomp(my @tmp = `$arg --get-all svn.$key`);
1403                        @$v = @tmp if @tmp;
1404                } else {
1405                        chomp(my $tmp = `$arg --get svn.$key`);
1406                        if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
1407                                $$v = $tmp;
1408                        }
1409                }
1410        }
1411        delete @$opts{@config_only} if @config_only;
1412}
1413
1414sub extract_metadata {
1415        my $id = shift or return (undef, undef, undef);
1416        my ($url, $rev, $uuid) = ($id =~ /^\s*git-svn-id:\s+(.*)\@(\d+)
1417                                                        \s([a-f\d\-]+)$/ix);
1418        if (!defined $rev || !$uuid || !$url) {
1419                # some of the original repositories I made had
1420                # identifiers like this:
1421                ($rev, $uuid) = ($id =~/^\s*git-svn-id:\s(\d+)\@([a-f\d\-]+)/i);
1422        }
1423        return ($url, $rev, $uuid);
1424}
1425
1426sub cmt_metadata {
1427        return extract_metadata((grep(/^git-svn-id: /,
1428                command(qw/cat-file commit/, shift)))[-1]);
1429}
1430
1431sub cmt_sha2rev_batch {
1432        my %s2r;
1433        my ($pid, $in, $out, $ctx) = command_bidi_pipe(qw/cat-file --batch/);
1434        my $list = shift;
1435
1436        foreach my $sha (@{$list}) {
1437                my $first = 1;
1438                my $size = 0;
1439                print $out $sha, "\n";
1440
1441                while (my $line = <$in>) {
1442                        if ($first && $line =~ /^[[:xdigit:]]{40}\smissing$/) {
1443                                last;
1444                        } elsif ($first &&
1445                               $line =~ /^[[:xdigit:]]{40}\scommit\s(\d+)$/) {
1446                                $first = 0;
1447                                $size = $1;
1448                                next;
1449                        } elsif ($line =~ /^(git-svn-id: )/) {
1450                                my (undef, $rev, undef) =
1451                                                      extract_metadata($line);
1452                                $s2r{$sha} = $rev;
1453                        }
1454
1455                        $size -= length($line);
1456                        last if ($size == 0);
1457                }
1458        }
1459
1460        command_close_bidi_pipe($pid, $in, $out, $ctx);
1461
1462        return \%s2r;
1463}
1464
1465sub working_head_info {
1466        my ($head, $refs) = @_;
1467        my @args = ('log', '--no-color', '--first-parent', '--pretty=medium');
1468        my ($fh, $ctx) = command_output_pipe(@args, $head);
1469        my $hash;
1470        my %max;
1471        while (<$fh>) {
1472                if ( m{^commit ($::sha1)$} ) {
1473                        unshift @$refs, $hash if $hash and $refs;
1474                        $hash = $1;
1475                        next;
1476                }
1477                next unless s{^\s*(git-svn-id:)}{$1};
1478                my ($url, $rev, $uuid) = extract_metadata($_);
1479                if (defined $url && defined $rev) {
1480                        next if $max{$url} and $max{$url} < $rev;
1481                        if (my $gs = Git::SVN->find_by_url($url)) {
1482                                my $c = $gs->rev_map_get($rev, $uuid);
1483                                if ($c && $c eq $hash) {
1484                                        close $fh; # break the pipe
1485                                        return ($url, $rev, $uuid, $gs);
1486                                } else {
1487                                        $max{$url} ||= $gs->rev_map_max;
1488                                }
1489                        }
1490                }
1491        }
1492        command_close_pipe($fh, $ctx);
1493        (undef, undef, undef, undef);
1494}
1495
1496sub read_commit_parents {
1497        my ($parents, $c) = @_;
1498        chomp(my $p = command_oneline(qw/rev-list --parents -1/, $c));
1499        $p =~ s/^($c)\s*// or die "rev-list --parents -1 $c failed!\n";
1500        @{$parents->{$c}} = split(/ /, $p);
1501}
1502
1503sub linearize_history {
1504        my ($gs, $refs) = @_;
1505        my %parents;
1506        foreach my $c (@$refs) {
1507                read_commit_parents(\%parents, $c);
1508        }
1509
1510        my @linear_refs;
1511        my %skip = ();
1512        my $last_svn_commit = $gs->last_commit;
1513        foreach my $c (reverse @$refs) {
1514                next if $c eq $last_svn_commit;
1515                last if $skip{$c};
1516
1517                unshift @linear_refs, $c;
1518                $skip{$c} = 1;
1519
1520                # we only want the first parent to diff against for linear
1521                # history, we save the rest to inject when we finalize the
1522                # svn commit
1523                my $fp_a = verify_ref("$c~1");
1524                my $fp_b = shift @{$parents{$c}} if $parents{$c};
1525                if (!$fp_a || !$fp_b) {
1526                        die "Commit $c\n",
1527                            "has no parent commit, and therefore ",
1528                            "nothing to diff against.\n",
1529                            "You should be working from a repository ",
1530                            "originally created by git-svn\n";
1531                }
1532                if ($fp_a ne $fp_b) {
1533                        die "$c~1 = $fp_a, however parsing commit $c ",
1534                            "revealed that:\n$c~1 = $fp_b\nBUG!\n";
1535                }
1536
1537                foreach my $p (@{$parents{$c}}) {
1538                        $skip{$p} = 1;
1539                }
1540        }
1541        (\@linear_refs, \%parents);
1542}
1543
1544sub find_file_type_and_diff_status {
1545        my ($path) = @_;
1546        return ('dir', '') if $path eq '';
1547
1548        my $diff_output =
1549            command_oneline(qw(diff --cached --name-status --), $path) || "";
1550        my $diff_status = (split(' ', $diff_output))[0] || "";
1551
1552        my $ls_tree = command_oneline(qw(ls-tree HEAD), $path) || "";
1553
1554        return (undef, undef) if !$diff_status && !$ls_tree;
1555
1556        if ($diff_status eq "A") {
1557                return ("link", $diff_status) if -l $path;
1558                return ("dir", $diff_status) if -d $path;
1559                return ("file", $diff_status);
1560        }
1561
1562        my $mode = (split(' ', $ls_tree))[0] || "";
1563
1564        return ("link", $diff_status) if $mode eq "120000";
1565        return ("dir", $diff_status) if $mode eq "040000";
1566        return ("file", $diff_status);
1567}
1568
1569sub md5sum {
1570        my $arg = shift;
1571        my $ref = ref $arg;
1572        my $md5 = Digest::MD5->new();
1573        if ($ref eq 'GLOB' || $ref eq 'IO::File' || $ref eq 'File::Temp') {
1574                $md5->addfile($arg) or croak $!;
1575        } elsif ($ref eq 'SCALAR') {
1576                $md5->add($$arg) or croak $!;
1577        } elsif (!$ref) {
1578                $md5->add($arg) or croak $!;
1579        } else {
1580                ::fatal "Can't provide MD5 hash for unknown ref type: '", $ref, "'";
1581        }
1582        return $md5->hexdigest();
1583}
1584
1585sub gc_directory {
1586        if ($can_compress && -f $_ && basename($_) eq "unhandled.log") {
1587                my $out_filename = $_ . ".gz";
1588                open my $in_fh, "<", $_ or die "Unable to open $_: $!\n";
1589                binmode $in_fh;
1590                my $gz = Compress::Zlib::gzopen($out_filename, "ab") or
1591                                die "Unable to open $out_filename: $!\n";
1592
1593                my $res;
1594                while ($res = sysread($in_fh, my $str, 1024)) {
1595                        $gz->gzwrite($str) or
1596                                die "Unable to write: ".$gz->gzerror()."!\n";
1597                }
1598                unlink $_ or die "unlink $File::Find::name: $!\n";
1599        } elsif (-f $_ && basename($_) eq "index") {
1600                unlink $_ or die "unlink $_: $!\n";
1601        }
1602}
1603
1604package Git::SVN;
1605use strict;
1606use warnings;
1607use Fcntl qw/:DEFAULT :seek/;
1608use constant rev_map_fmt => 'NH40';
1609use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
1610            $_repack $_repack_flags $_use_svm_props $_head
1611            $_use_svnsync_props $no_reuse_existing $_minimize_url
1612            $_use_log_author $_add_author_from $_localtime/;
1613use Carp qw/croak/;
1614use File::Path qw/mkpath/;
1615use File::Copy qw/copy/;
1616use IPC::Open3;
1617
1618my ($_gc_nr, $_gc_period);
1619
1620# properties that we do not log:
1621my %SKIP_PROP;
1622BEGIN {
1623        %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
1624                                        svn:special svn:executable
1625                                        svn:entry:committed-rev
1626                                        svn:entry:last-author
1627                                        svn:entry:uuid
1628                                        svn:entry:committed-date/;
1629
1630        # some options are read globally, but can be overridden locally
1631        # per [svn-remote "..."] section.  Command-line options will *NOT*
1632        # override options set in an [svn-remote "..."] section
1633        no strict 'refs';
1634        for my $option (qw/follow_parent no_metadata use_svm_props
1635                           use_svnsync_props/) {
1636                my $key = $option;
1637                $key =~ tr/_//d;
1638                my $prop = "-$option";
1639                *$option = sub {
1640                        my ($self) = @_;
1641                        return $self->{$prop} if exists $self->{$prop};
1642                        my $k = "svn-remote.$self->{repo_id}.$key";
1643                        eval { command_oneline(qw/config --get/, $k) };
1644                        if ($@) {
1645                                $self->{$prop} = ${"Git::SVN::_$option"};
1646                        } else {
1647                                my $v = command_oneline(qw/config --bool/,$k);
1648                                $self->{$prop} = $v eq 'false' ? 0 : 1;
1649                        }
1650                        return $self->{$prop};
1651                }
1652        }
1653}
1654
1655
1656my (%LOCKFILES, %INDEX_FILES);
1657END {
1658        unlink keys %LOCKFILES if %LOCKFILES;
1659        unlink keys %INDEX_FILES if %INDEX_FILES;
1660}
1661
1662sub resolve_local_globs {
1663        my ($url, $fetch, $glob_spec) = @_;
1664        return unless defined $glob_spec;
1665        my $ref = $glob_spec->{ref};
1666        my $path = $glob_spec->{path};
1667        foreach (command(qw#for-each-ref --format=%(refname) refs/#)) {
1668                next unless m#^$ref->{regex}$#;
1669                my $p = $1;
1670                my $pathname = desanitize_refname($path->full_path($p));
1671                my $refname = desanitize_refname($ref->full_path($p));
1672                if (my $existing = $fetch->{$pathname}) {
1673                        if ($existing ne $refname) {
1674                                die "Refspec conflict:\n",
1675                                    "existing: $existing\n",
1676                                    " globbed: $refname\n";
1677                        }
1678                        my $u = (::cmt_metadata("$refname"))[0];
1679                        $u =~ s!^\Q$url\E(/|$)!! or die
1680                          "$refname: '$url' not found in '$u'\n";
1681                        if ($pathname ne $u) {
1682                                warn "W: Refspec glob conflict ",
1683                                     "(ref: $refname):\n",
1684                                     "expected path: $pathname\n",
1685                                     "    real path: $u\n",
1686                                     "Continuing ahead with $u\n";
1687                                next;
1688                        }
1689                } else {
1690                        $fetch->{$pathname} = $refname;
1691                }
1692        }
1693}
1694
1695sub parse_revision_argument {
1696        my ($base, $head) = @_;
1697        if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
1698                return ($base, $head);
1699        }
1700        return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
1701        return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
1702        return ($head, $head) if ($::_revision eq 'HEAD');
1703        return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
1704        return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
1705        die "revision argument: $::_revision not understood by git-svn\n";
1706}
1707
1708sub fetch_all {
1709        my ($repo_id, $remotes) = @_;
1710        if (ref $repo_id) {
1711                my $gs = $repo_id;
1712                $repo_id = undef;
1713                $repo_id = $gs->{repo_id};
1714        }
1715        $remotes ||= read_all_remotes();
1716        my $remote = $remotes->{$repo_id} or
1717                     die "[svn-remote \"$repo_id\"] unknown\n";
1718        my $fetch = $remote->{fetch};
1719        my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
1720        my (@gs, @globs);
1721        my $ra = Git::SVN::Ra->new($url);
1722        my $uuid = $ra->get_uuid;
1723        my $head = $ra->get_latest_revnum;
1724        $ra->get_log("", $head, 0, 1, 0, 1, sub { $head = $_[1] });
1725        my $base = defined $fetch ? $head : 0;
1726
1727        # read the max revs for wildcard expansion (branches/*, tags/*)
1728        foreach my $t (qw/branches tags/) {
1729                defined $remote->{$t} or next;
1730                push @globs, @{$remote->{$t}};
1731
1732                my $max_rev = eval { tmp_config(qw/--int --get/,
1733                                         "svn-remote.$repo_id.${t}-maxRev") };
1734                if (defined $max_rev && ($max_rev < $base)) {
1735                        $base = $max_rev;
1736                } elsif (!defined $max_rev) {
1737                        $base = 0;
1738                }
1739        }
1740
1741        if ($fetch) {
1742                foreach my $p (sort keys %$fetch) {
1743                        my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
1744                        my $lr = $gs->rev_map_max;
1745                        if (defined $lr) {
1746                                $base = $lr if ($lr < $base);
1747                        }
1748                        push @gs, $gs;
1749                }
1750        }
1751
1752        ($base, $head) = parse_revision_argument($base, $head);
1753        $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
1754}
1755
1756sub read_all_remotes {
1757        my $r = {};
1758        my $use_svm_props = eval { command_oneline(qw/config --bool
1759            svn.useSvmProps/) };
1760        $use_svm_props = $use_svm_props eq 'true' if $use_svm_props;
1761        my $svn_refspec = qr{\s*/?(.*?)\s*:\s*(.+?)\s*};
1762        foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
1763                if (m!^(.+)\.fetch=$svn_refspec$!) {
1764                        my ($remote, $local_ref, $remote_ref) = ($1, $2, $3);
1765                        die("svn-remote.$remote: remote ref '$remote_ref' "
1766                            . "must start with 'refs/'\n")
1767                                unless $remote_ref =~ m{^refs/};
1768                        $r->{$remote}->{fetch}->{$local_ref} = $remote_ref;
1769                        $r->{$remote}->{svm} = {} if $use_svm_props;
1770                } elsif (m!^(.+)\.usesvmprops=\s*(.*)\s*$!) {
1771                        $r->{$1}->{svm} = {};
1772                } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
1773                        $r->{$1}->{url} = $2;
1774                } elsif (m!^(.+)\.(branches|tags)=$svn_refspec$!) {
1775                        my ($remote, $t, $local_ref, $remote_ref) =
1776                                                             ($1, $2, $3, $4);
1777                        die("svn-remote.$remote: remote ref '$remote_ref' ($t) "
1778                            . "must start with 'refs/'\n")
1779                                unless $remote_ref =~ m{^refs/};
1780                        my $rs = {
1781                            t => $t,
1782                            remote => $remote,
1783                            path => Git::SVN::GlobSpec->new($local_ref),
1784                            ref => Git::SVN::GlobSpec->new($remote_ref) };
1785                        if (length($rs->{ref}->{right}) != 0) {
1786                                die "The '*' glob character must be the last ",
1787                                    "character of '$remote_ref'\n";
1788                        }
1789                        push @{ $r->{$remote}->{$t} }, $rs;
1790                }
1791        }
1792
1793        map {
1794                if (defined $r->{$_}->{svm}) {
1795                        my $svm;
1796                        eval {
1797                                my $section = "svn-remote.$_";
1798                                $svm = {
1799                                        source => tmp_config('--get',
1800                                            "$section.svm-source"),
1801                                        replace => tmp_config('--get',
1802                                            "$section.svm-replace"),
1803                                }
1804                        };
1805                        $r->{$_}->{svm} = $svm;
1806                }
1807        } keys %$r;
1808
1809        $r;
1810}
1811
1812sub init_vars {
1813        $_gc_nr = $_gc_period = 1000;
1814        if (defined $_repack || defined $_repack_flags) {
1815               warn "Repack options are obsolete; they have no effect.\n";
1816        }
1817}
1818
1819sub verify_remotes_sanity {
1820        return unless -d $ENV{GIT_DIR};
1821        my %seen;
1822        foreach (command(qw/config -l/)) {
1823                if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
1824                        if ($seen{$1}) {
1825                                die "Remote ref refs/remote/$1 is tracked by",
1826                                    "\n  \"$_\"\nand\n  \"$seen{$1}\"\n",
1827                                    "Please resolve this ambiguity in ",
1828                                    "your git configuration file before ",
1829                                    "continuing\n";
1830                        }
1831                        $seen{$1} = $_;
1832                }
1833        }
1834}
1835
1836sub find_existing_remote {
1837        my ($url, $remotes) = @_;
1838        return undef if $no_reuse_existing;
1839        my $existing;
1840        foreach my $repo_id (keys %$remotes) {
1841                my $u = $remotes->{$repo_id}->{url} or next;
1842                next if $u ne $url;
1843                $existing = $repo_id;
1844                last;
1845        }
1846        $existing;
1847}
1848
1849sub init_remote_config {
1850        my ($self, $url, $no_write) = @_;
1851        $url =~ s!/+$!!; # strip trailing slash
1852        my $r = read_all_remotes();
1853        my $existing = find_existing_remote($url, $r);
1854        if ($existing) {
1855                unless ($no_write) {
1856                        print STDERR "Using existing ",
1857                                     "[svn-remote \"$existing\"]\n";
1858                }
1859                $self->{repo_id} = $existing;
1860        } elsif ($_minimize_url) {
1861                my $min_url = Git::SVN::Ra->new($url)->minimize_url;
1862                $existing = find_existing_remote($min_url, $r);
1863                if ($existing) {
1864                        unless ($no_write) {
1865                                print STDERR "Using existing ",
1866                                             "[svn-remote \"$existing\"]\n";
1867                        }
1868                        $self->{repo_id} = $existing;
1869                }
1870                if ($min_url ne $url) {
1871                        unless ($no_write) {
1872                                print STDERR "Using higher level of URL: ",
1873                                             "$url => $min_url\n";
1874                        }
1875                        my $old_path = $self->{path};
1876                        $self->{path} = $url;
1877                        $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
1878                        if (length $old_path) {
1879                                $self->{path} .= "/$old_path";
1880                        }
1881                        $url = $min_url;
1882                }
1883        }
1884        my $orig_url;
1885        if (!$existing) {
1886                # verify that we aren't overwriting anything:
1887                $orig_url = eval {
1888                        command_oneline('config', '--get',
1889                                        "svn-remote.$self->{repo_id}.url")
1890                };
1891                if ($orig_url && ($orig_url ne $url)) {
1892                        die "svn-remote.$self->{repo_id}.url already set: ",
1893                            "$orig_url\nwanted to set to: $url\n";
1894                }
1895        }
1896        my ($xrepo_id, $xpath) = find_ref($self->refname);
1897        if (!$no_write && defined $xpath) {
1898                die "svn-remote.$xrepo_id.fetch already set to track ",
1899                    "$xpath:", $self->refname, "\n";
1900        }
1901        unless ($no_write) {
1902                command_noisy('config',
1903                              "svn-remote.$self->{repo_id}.url", $url);
1904                $self->{path} =~ s{^/}{};
1905                $self->{path} =~ s{%([0-9A-F]{2})}{chr hex($1)}ieg;
1906                command_noisy('config', '--add',
1907                              "svn-remote.$self->{repo_id}.fetch",
1908                              "$self->{path}:".$self->refname);
1909        }
1910        $self->{url} = $url;
1911}
1912
1913sub find_by_url { # repos_root and, path are optional
1914        my ($class, $full_url, $repos_root, $path) = @_;
1915
1916        return undef unless defined $full_url;
1917        remove_username($full_url);
1918        remove_username($repos_root) if defined $repos_root;
1919        my $remotes = read_all_remotes();
1920        if (defined $full_url && defined $repos_root && !defined $path) {
1921                $path = $full_url;
1922                $path =~ s#^\Q$repos_root\E(?:/|$)##;
1923        }
1924        foreach my $repo_id (keys %$remotes) {
1925                my $u = $remotes->{$repo_id}->{url} or next;
1926                remove_username($u);
1927                next if defined $repos_root && $repos_root ne $u;
1928
1929                my $fetch = $remotes->{$repo_id}->{fetch} || {};
1930                foreach my $t (qw/branches tags/) {
1931                        foreach my $globspec (@{$remotes->{$repo_id}->{$t}}) {
1932                                resolve_local_globs($u, $fetch, $globspec);
1933                        }
1934                }
1935                my $p = $path;
1936                my $rwr = rewrite_root({repo_id => $repo_id});
1937                my $svm = $remotes->{$repo_id}->{svm}
1938                        if defined $remotes->{$repo_id}->{svm};
1939                unless (defined $p) {
1940                        $p = $full_url;
1941                        my $z = $u;
1942                        my $prefix = '';
1943                        if ($rwr) {
1944                                $z = $rwr;
1945                                remove_username($z);
1946                        } elsif (defined $svm) {
1947                                $z = $svm->{source};
1948                                $prefix = $svm->{replace};
1949                                $prefix =~ s#^\Q$u\E(?:/|$)##;
1950                                $prefix =~ s#/$##;
1951                        }
1952                        $p =~ s#^\Q$z\E(?:/|$)#$prefix# or next;
1953                }
1954                foreach my $f (keys %$fetch) {
1955                        next if $f ne $p;
1956                        return Git::SVN->new($fetch->{$f}, $repo_id, $f);
1957                }
1958        }
1959        undef;
1960}
1961
1962sub init {
1963        my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
1964        my $self = _new($class, $repo_id, $ref_id, $path);
1965        if (defined $url) {
1966                $self->init_remote_config($url, $no_write);
1967        }
1968        $self;
1969}
1970
1971sub find_ref {
1972        my ($ref_id) = @_;
1973        foreach (command(qw/config -l/)) {
1974                next unless m!^svn-remote\.(.+)\.fetch=
1975                              \s*/?(.*?)\s*:\s*(.+?)\s*$!x;
1976                my ($repo_id, $path, $ref) = ($1, $2, $3);
1977                if ($ref eq $ref_id) {
1978                        $path = '' if ($path =~ m#^\./?#);
1979                        return ($repo_id, $path);
1980                }
1981        }
1982        (undef, undef, undef);
1983}
1984
1985sub new {
1986        my ($class, $ref_id, $repo_id, $path) = @_;
1987        if (defined $ref_id && !defined $repo_id && !defined $path) {
1988                ($repo_id, $path) = find_ref($ref_id);
1989                if (!defined $repo_id) {
1990                        die "Could not find a \"svn-remote.*.fetch\" key ",
1991                            "in the repository configuration matching: ",
1992                            "$ref_id\n";
1993                }
1994        }
1995        my $self = _new($class, $repo_id, $ref_id, $path);
1996        if (!defined $self->{path} || !length $self->{path}) {
1997                my $fetch = command_oneline('config', '--get',
1998                                            "svn-remote.$repo_id.fetch",
1999                                            ":$ref_id\$") or
2000                     die "Failed to read \"svn-remote.$repo_id.fetch\" ",
2001                         "\":$ref_id\$\" in config\n";
2002                ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
2003        }
2004        $self->{url} = command_oneline('config', '--get',
2005                                       "svn-remote.$repo_id.url") or
2006                  die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
2007        $self->rebuild;
2008        $self;
2009}
2010
2011sub refname {
2012        my ($refname) = $_[0]->{ref_id} ;
2013
2014        # It cannot end with a slash /, we'll throw up on this because
2015        # SVN can't have directories with a slash in their name, either:
2016        if ($refname =~ m{/$}) {
2017                die "ref: '$refname' ends with a trailing slash, this is ",
2018                    "not permitted by git nor Subversion\n";
2019        }
2020
2021        # It cannot have ASCII control character space, tilde ~, caret ^,
2022        # colon :, question-mark ?, asterisk *, space, or open bracket [
2023        # anywhere.
2024        #
2025        # Additionally, % must be escaped because it is used for escaping
2026        # and we want our escaped refname to be reversible
2027        $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
2028
2029        # no slash-separated component can begin with a dot .
2030        # /.* becomes /%2E*
2031        $refname =~ s{/\.}{/%2E}g;
2032
2033        # It cannot have two consecutive dots .. anywhere
2034        # .. becomes %2E%2E
2035        $refname =~ s{\.\.}{%2E%2E}g;
2036
2037        return $refname;
2038}
2039
2040sub desanitize_refname {
2041        my ($refname) = @_;
2042        $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
2043        return $refname;
2044}
2045
2046sub svm_uuid {
2047        my ($self) = @_;
2048        return $self->{svm}->{uuid} if $self->svm;
2049        $self->ra;
2050        unless ($self->{svm}) {
2051                die "SVM UUID not cached, and reading remotely failed\n";
2052        }
2053        $self->{svm}->{uuid};
2054}
2055
2056sub svm {
2057        my ($self) = @_;
2058        return $self->{svm} if $self->{svm};
2059        my $svm;
2060        # see if we have it in our config, first:
2061        eval {
2062                my $section = "svn-remote.$self->{repo_id}";
2063                $svm = {
2064                  source => tmp_config('--get', "$section.svm-source"),
2065                  uuid => tmp_config('--get', "$section.svm-uuid"),
2066                  replace => tmp_config('--get', "$section.svm-replace"),
2067                }
2068        };
2069        if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
2070                $self->{svm} = $svm;
2071        }
2072        $self->{svm};
2073}
2074
2075sub _set_svm_vars {
2076        my ($self, $ra) = @_;
2077        return $ra if $self->svm;
2078
2079        my @err = ( "useSvmProps set, but failed to read SVM properties\n",
2080                    "(svm:source, svm:uuid) ",
2081                    "from the following URLs:\n" );
2082        sub read_svm_props {
2083                my ($self, $ra, $path, $r) = @_;
2084                my $props = ($ra->get_dir($path, $r))[2];
2085                my $src = $props->{'svm:source'};
2086                my $uuid = $props->{'svm:uuid'};
2087                return undef if (!$src || !$uuid);
2088
2089                chomp($src, $uuid);
2090
2091                $uuid =~ m{^[0-9a-f\-]{30,}$}i
2092                    or die "doesn't look right - svm:uuid is '$uuid'\n";
2093
2094                # the '!' is used to mark the repos_root!/relative/path
2095                $src =~ s{/?!/?}{/};
2096                $src =~ s{/+$}{}; # no trailing slashes please
2097                # username is of no interest
2098                $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
2099
2100                my $replace = $ra->{url};
2101                $replace .= "/$path" if length $path;
2102
2103                my $section = "svn-remote.$self->{repo_id}";
2104                tmp_config("$section.svm-source", $src);
2105                tmp_config("$section.svm-replace", $replace);
2106                tmp_config("$section.svm-uuid", $uuid);
2107                $self->{svm} = {
2108                        source => $src,
2109                        uuid => $uuid,
2110                        replace => $replace
2111                };
2112        }
2113
2114        my $r = $ra->get_latest_revnum;
2115        my $path = $self->{path};
2116        my %tried;
2117        while (length $path) {
2118                unless ($tried{"$self->{url}/$path"}) {
2119                        return $ra if $self->read_svm_props($ra, $path, $r);
2120                        $tried{"$self->{url}/$path"} = 1;
2121                }
2122                $path =~ s#/?[^/]+$##;
2123        }
2124        die "Path: '$path' should be ''\n" if $path ne '';
2125        return $ra if $self->read_svm_props($ra, $path, $r);
2126        $tried{"$self->{url}/$path"} = 1;
2127
2128        if ($ra->{repos_root} eq $self->{url}) {
2129                die @err, (map { "  $_\n" } keys %tried), "\n";
2130        }
2131
2132        # nope, make sure we're connected to the repository root:
2133        my $ok;
2134        my @tried_b;
2135        $path = $ra->{svn_path};
2136        $ra = Git::SVN::Ra->new($ra->{repos_root});
2137        while (length $path) {
2138                unless ($tried{"$ra->{url}/$path"}) {
2139                        $ok = $self->read_svm_props($ra, $path, $r);
2140                        last if $ok;
2141                        $tried{"$ra->{url}/$path"} = 1;
2142                }
2143                $path =~ s#/?[^/]+$##;
2144        }
2145        die "Path: '$path' should be ''\n" if $path ne '';
2146        $ok ||= $self->read_svm_props($ra, $path, $r);
2147        $tried{"$ra->{url}/$path"} = 1;
2148        if (!$ok) {
2149                die @err, (map { "  $_\n" } keys %tried), "\n";
2150        }
2151        Git::SVN::Ra->new($self->{url});
2152}
2153
2154sub svnsync {
2155        my ($self) = @_;
2156        return $self->{svnsync} if $self->{svnsync};
2157
2158        if ($self->no_metadata) {
2159                die "Can't have both 'noMetadata' and ",
2160                    "'useSvnsyncProps' options set!\n";
2161        }
2162        if ($self->rewrite_root) {
2163                die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
2164                    "options set!\n";
2165        }
2166
2167        my $svnsync;
2168        # see if we have it in our config, first:
2169        eval {
2170                my $section = "svn-remote.$self->{repo_id}";
2171
2172                my $url = tmp_config('--get', "$section.svnsync-url");
2173                ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2174                   die "doesn't look right - svn:sync-from-url is '$url'\n";
2175
2176                my $uuid = tmp_config('--get', "$section.svnsync-uuid");
2177                ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2178                   die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2179
2180                $svnsync = { url => $url, uuid => $uuid }
2181        };
2182        if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
2183                return $self->{svnsync} = $svnsync;
2184        }
2185
2186        my $err = "useSvnsyncProps set, but failed to read " .
2187                  "svnsync property: svn:sync-from-";
2188        my $rp = $self->ra->rev_proplist(0);
2189
2190        my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
2191        ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2192                   die "doesn't look right - svn:sync-from-url is '$url'\n";
2193
2194        my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
2195        ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2196                   die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2197
2198        my $section = "svn-remote.$self->{repo_id}";
2199        tmp_config('--add', "$section.svnsync-uuid", $uuid);
2200        tmp_config('--add', "$section.svnsync-url", $url);
2201        return $self->{svnsync} = { url => $url, uuid => $uuid };
2202}
2203
2204# this allows us to memoize our SVN::Ra UUID locally and avoid a
2205# remote lookup (useful for 'git svn log').
2206sub ra_uuid {
2207        my ($self) = @_;
2208        unless ($self->{ra_uuid}) {
2209                my $key = "svn-remote.$self->{repo_id}.uuid";
2210                my $uuid = eval { tmp_config('--get', $key) };
2211                if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/i) {
2212                        $self->{ra_uuid} = $uuid;
2213                } else {
2214                        die "ra_uuid called without URL\n" unless $self->{url};
2215                        $self->{ra_uuid} = $self->ra->get_uuid;
2216                        tmp_config('--add', $key, $self->{ra_uuid});
2217                }
2218        }
2219        $self->{ra_uuid};
2220}
2221
2222sub _set_repos_root {
2223        my ($self, $repos_root) = @_;
2224        my $k = "svn-remote.$self->{repo_id}.reposRoot";
2225        $repos_root ||= $self->ra->{repos_root};
2226        tmp_config($k, $repos_root);
2227        $repos_root;
2228}
2229
2230sub repos_root {
2231        my ($self) = @_;
2232        my $k = "svn-remote.$self->{repo_id}.reposRoot";
2233        eval { tmp_config('--get', $k) } || $self->_set_repos_root;
2234}
2235
2236sub ra {
2237        my ($self) = shift;
2238        my $ra = Git::SVN::Ra->new($self->{url});
2239        $self->_set_repos_root($ra->{repos_root});
2240        if ($self->use_svm_props && !$self->{svm}) {
2241                if ($self->no_metadata) {
2242                        die "Can't have both 'noMetadata' and ",
2243                            "'useSvmProps' options set!\n";
2244                } elsif ($self->use_svnsync_props) {
2245                        die "Can't have both 'useSvnsyncProps' and ",
2246                            "'useSvmProps' options set!\n";
2247                }
2248                $ra = $self->_set_svm_vars($ra);
2249                $self->{-want_revprops} = 1;
2250        }
2251        $ra;
2252}
2253
2254# prop_walk(PATH, REV, SUB)
2255# -------------------------
2256# Recursively traverse PATH at revision REV and invoke SUB for each
2257# directory that contains a SVN property.  SUB will be invoked as
2258# follows:  &SUB(gs, path, props);  where `gs' is this instance of
2259# Git::SVN, `path' the path to the directory where the properties
2260# `props' were found.  The `path' will be relative to point of checkout,
2261# that is, if url://repo/trunk is the current Git branch, and that
2262# directory contains a sub-directory `d', SUB will be invoked with `/d/'
2263# as `path' (note the trailing `/').
2264sub prop_walk {
2265        my ($self, $path, $rev, $sub) = @_;
2266
2267        $path =~ s#^/##;
2268        my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev);
2269        $path =~ s#^/*#/#g;
2270        my $p = $path;
2271        # Strip the irrelevant part of the path.
2272        $p =~ s#^/+\Q$self->{path}\E(/|$)#/#;
2273        # Ensure the path is terminated by a `/'.
2274        $p =~ s#/*$#/#;
2275
2276        # The properties contain all the internal SVN stuff nobody
2277        # (usually) cares about.
2278        my $interesting_props = 0;
2279        foreach (keys %{$props}) {
2280                # If it doesn't start with `svn:', it must be a
2281                # user-defined property.
2282                ++$interesting_props and next if $_ !~ /^svn:/;
2283                # FIXME: Fragile, if SVN adds new public properties,
2284                # this needs to be updated.
2285                ++$interesting_props if /^svn:(?:ignore|keywords|executable
2286                                                 |eol-style|mime-type
2287                                                 |externals|needs-lock)$/x;
2288        }
2289        &$sub($self, $p, $props) if $interesting_props;
2290
2291        foreach (sort keys %$dirent) {
2292                next if $dirent->{$_}->{kind} != $SVN::Node::dir;
2293                $self->prop_walk($self->{path} . $p . $_, $rev, $sub);
2294        }
2295}
2296
2297sub last_rev { ($_[0]->last_rev_commit)[0] }
2298sub last_commit { ($_[0]->last_rev_commit)[1] }
2299
2300# returns the newest SVN revision number and newest commit SHA1
2301sub last_rev_commit {
2302        my ($self) = @_;
2303        if (defined $self->{last_rev} && defined $self->{last_commit}) {
2304                return ($self->{last_rev}, $self->{last_commit});
2305        }
2306        my $c = ::verify_ref($self->refname.'^0');
2307        if ($c && !$self->use_svm_props && !$self->no_metadata) {
2308                my $rev = (::cmt_metadata($c))[1];
2309                if (defined $rev) {
2310                        ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
2311                        return ($rev, $c);
2312                }
2313        }
2314        my $map_path = $self->map_path;
2315        unless (-e $map_path) {
2316                ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
2317                return (undef, undef);
2318        }
2319        my ($rev, $commit) = $self->rev_map_max(1);
2320        ($self->{last_rev}, $self->{last_commit}) = ($rev, $commit);
2321        return ($rev, $commit);
2322}
2323
2324sub get_fetch_range {
2325        my ($self, $min, $max) = @_;
2326        $max ||= $self->ra->get_latest_revnum;
2327        $min ||= $self->rev_map_max;
2328        (++$min, $max);
2329}
2330
2331sub tmp_config {
2332        my (@args) = @_;
2333        my $old_def_config = "$ENV{GIT_DIR}/svn/config";
2334        my $config = "$ENV{GIT_DIR}/svn/.metadata";
2335        if (! -f $config && -f $old_def_config) {
2336                rename $old_def_config, $config or
2337                       die "Failed rename $old_def_config => $config: $!\n";
2338        }
2339        my $old_config = $ENV{GIT_CONFIG};
2340        $ENV{GIT_CONFIG} = $config;
2341        $@ = undef;
2342        my @ret = eval {
2343                unless (-f $config) {
2344                        mkfile($config);
2345                        open my $fh, '>', $config or
2346                            die "Can't open $config: $!\n";
2347                        print $fh "; This file is used internally by ",
2348                                  "git-svn\n" or die
2349                                  "Couldn't write to $config: $!\n";
2350                        print $fh "; You should not have to edit it\n" or
2351                              die "Couldn't write to $config: $!\n";
2352                        close $fh or die "Couldn't close $config: $!\n";
2353                }
2354                command('config', @args);
2355        };
2356        my $err = $@;
2357        if (defined $old_config) {
2358                $ENV{GIT_CONFIG} = $old_config;
2359        } else {
2360                delete $ENV{GIT_CONFIG};
2361        }
2362        die $err if $err;
2363        wantarray ? @ret : $ret[0];
2364}
2365
2366sub tmp_index_do {
2367        my ($self, $sub) = @_;
2368        my $old_index = $ENV{GIT_INDEX_FILE};
2369        $ENV{GIT_INDEX_FILE} = $self->{index};
2370        $@ = undef;
2371        my @ret = eval {
2372                my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
2373                mkpath([$dir]) unless -d $dir;
2374                &$sub;
2375        };
2376        my $err = $@;
2377        if (defined $old_index) {
2378                $ENV{GIT_INDEX_FILE} = $old_index;
2379        } else {
2380                delete $ENV{GIT_INDEX_FILE};
2381        }
2382        die $err if $err;
2383        wantarray ? @ret : $ret[0];
2384}
2385
2386sub assert_index_clean {
2387        my ($self, $treeish) = @_;
2388
2389        $self->tmp_index_do(sub {
2390                command_noisy('read-tree', $treeish) unless -e $self->{index};
2391                my $x = command_oneline('write-tree');
2392                my ($y) = (command(qw/cat-file commit/, $treeish) =~
2393                           /^tree ($::sha1)/mo);
2394                return if $y eq $x;
2395
2396                warn "Index mismatch: $y != $x\nrereading $treeish\n";
2397                unlink $self->{index} or die "unlink $self->{index}: $!\n";
2398                command_noisy('read-tree', $treeish);
2399                $x = command_oneline('write-tree');
2400                if ($y ne $x) {
2401                        ::fatal "trees ($treeish) $y != $x\n",
2402                                "Something is seriously wrong...";
2403                }
2404        });
2405}
2406
2407sub get_commit_parents {
2408        my ($self, $log_entry) = @_;
2409        my (%seen, @ret, @tmp);
2410        # legacy support for 'set-tree'; this is only used by set_tree_cb:
2411        if (my $ip = $self->{inject_parents}) {
2412                if (my $commit = delete $ip->{$log_entry->{revision}}) {
2413                        push @tmp, $commit;
2414                }
2415        }
2416        if (my $cur = ::verify_ref($self->refname.'^0')) {
2417                push @tmp, $cur;
2418        }
2419        if (my $ipd = $self->{inject_parents_dcommit}) {
2420                if (my $commit = delete $ipd->{$log_entry->{revision}}) {
2421                        push @tmp, @$commit;
2422                }
2423        }
2424        push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
2425        while (my $p = shift @tmp) {
2426                next if $seen{$p};
2427                $seen{$p} = 1;
2428                push @ret, $p;
2429                # MAXPARENT is defined to 16 in commit-tree.c:
2430                last if @ret >= 16;
2431        }
2432        if (@tmp) {
2433                die "r$log_entry->{revision}: No room for parents:\n\t",
2434                    join("\n\t", @tmp), "\n";
2435        }
2436        @ret;
2437}
2438
2439sub rewrite_root {
2440        my ($self) = @_;
2441        return $self->{-rewrite_root} if exists $self->{-rewrite_root};
2442        my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
2443        my $rwr = eval { command_oneline(qw/config --get/, $k) };
2444        if ($rwr) {
2445                $rwr =~ s#/+$##;
2446                if ($rwr !~ m#^[a-z\+]+://#) {
2447                        die "$rwr is not a valid URL (key: $k)\n";
2448                }
2449        }
2450        $self->{-rewrite_root} = $rwr;
2451}
2452
2453sub metadata_url {
2454        my ($self) = @_;
2455        ($self->rewrite_root || $self->{url}) .
2456           (length $self->{path} ? '/' . $self->{path} : '');
2457}
2458
2459sub full_url {
2460        my ($self) = @_;
2461        $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
2462}
2463
2464
2465sub set_commit_header_env {
2466        my ($log_entry) = @_;
2467        my %env;
2468        foreach my $ned (qw/NAME EMAIL DATE/) {
2469                foreach my $ac (qw/AUTHOR COMMITTER/) {
2470                        $env{"GIT_${ac}_${ned}"} = $ENV{"GIT_${ac}_${ned}"};
2471                }
2472        }
2473
2474        $ENV{GIT_AUTHOR_NAME} = $log_entry->{name};
2475        $ENV{GIT_AUTHOR_EMAIL} = $log_entry->{email};
2476        $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
2477
2478        $ENV{GIT_COMMITTER_NAME} = (defined $log_entry->{commit_name})
2479                                                ? $log_entry->{commit_name}
2480                                                : $log_entry->{name};
2481        $ENV{GIT_COMMITTER_EMAIL} = (defined $log_entry->{commit_email})
2482                                                ? $log_entry->{commit_email}
2483                                                : $log_entry->{email};
2484        \%env;
2485}
2486
2487sub restore_commit_header_env {
2488        my ($env) = @_;
2489        foreach my $ned (qw/NAME EMAIL DATE/) {
2490                foreach my $ac (qw/AUTHOR COMMITTER/) {
2491                        my $k = "GIT_${ac}_${ned}";
2492                        if (defined $env->{$k}) {
2493                                $ENV{$k} = $env->{$k};
2494                        } else {
2495                                delete $ENV{$k};
2496                        }
2497                }
2498        }
2499}
2500
2501sub gc {
2502        command_noisy('gc', '--auto');
2503};
2504
2505sub do_git_commit {
2506        my ($self, $log_entry) = @_;
2507        my $lr = $self->last_rev;
2508        if (defined $lr && $lr >= $log_entry->{revision}) {
2509                die "Last fetched revision of ", $self->refname,
2510                    " was r$lr, but we are about to fetch: ",
2511                    "r$log_entry->{revision}!\n";
2512        }
2513        if (my $c = $self->rev_map_get($log_entry->{revision})) {
2514                croak "$log_entry->{revision} = $c already exists! ",
2515                      "Why are we refetching it?\n";
2516        }
2517        my $old_env = set_commit_header_env($log_entry);
2518        my $tree = $log_entry->{tree};
2519        if (!defined $tree) {
2520                $tree = $self->tmp_index_do(sub {
2521                                            command_oneline('write-tree') });
2522        }
2523        die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
2524
2525        my @exec = ('git', 'commit-tree', $tree);
2526        foreach ($self->get_commit_parents($log_entry)) {
2527                push @exec, '-p', $_;
2528        }
2529        defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
2530                                                                   or croak $!;
2531        binmode $msg_fh;
2532
2533        # we always get UTF-8 from SVN, but we may want our commits in
2534        # a different encoding.
2535        if (my $enc = Git::config('i18n.commitencoding')) {
2536                require Encode;
2537                Encode::from_to($log_entry->{log}, 'UTF-8', $enc);
2538        }
2539        print $msg_fh $log_entry->{log} or croak $!;
2540        restore_commit_header_env($old_env);
2541        unless ($self->no_metadata) {
2542                print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
2543                              or croak $!;
2544        }
2545        $msg_fh->flush == 0 or croak $!;
2546        close $msg_fh or croak $!;
2547        chomp(my $commit = do { local $/; <$out_fh> });
2548        close $out_fh or croak $!;
2549        waitpid $pid, 0;
2550        croak $? if $?;
2551        if ($commit !~ /^$::sha1$/o) {
2552                die "Failed to commit, invalid sha1: $commit\n";
2553        }
2554
2555        $self->rev_map_set($log_entry->{revision}, $commit, 1);
2556
2557        $self->{last_rev} = $log_entry->{revision};
2558        $self->{last_commit} = $commit;
2559        print "r$log_entry->{revision}" unless $::_q > 1;
2560        if (defined $log_entry->{svm_revision}) {
2561                 print " (\@$log_entry->{svm_revision})" unless $::_q > 1;
2562                 $self->rev_map_set($log_entry->{svm_revision}, $commit,
2563                                   0, $self->svm_uuid);
2564        }
2565        print " = $commit ($self->{ref_id})\n" unless $::_q > 1;
2566        if (--$_gc_nr == 0) {
2567                $_gc_nr = $_gc_period;
2568                gc();
2569        }
2570        return $commit;
2571}
2572
2573sub match_paths {
2574        my ($self, $paths, $r) = @_;
2575        return 1 if $self->{path} eq '';
2576        if (my $path = $paths->{"/$self->{path}"}) {
2577                return ($path->{action} eq 'D') ? 0 : 1;
2578        }
2579        $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
2580        if (grep /$self->{path_regex}/, keys %$paths) {
2581                return 1;
2582        }
2583        my $c = '';
2584        foreach (split m#/#, $self->{path}) {
2585                $c .= "/$_";
2586                next unless ($paths->{$c} &&
2587                             ($paths->{$c}->{action} =~ /^[AR]$/));
2588                if ($self->ra->check_path($self->{path}, $r) ==
2589                    $SVN::Node::dir) {
2590                        return 1;
2591                }
2592        }
2593        return 0;
2594}
2595
2596sub find_parent_branch {
2597        my ($self, $paths, $rev) = @_;
2598        return undef unless $self->follow_parent;
2599        unless (defined $paths) {
2600                my $err_handler = $SVN::Error::handler;
2601                $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
2602                $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1,
2603                                   sub { $paths = $_[0] });
2604                $SVN::Error::handler = $err_handler;
2605        }
2606        return undef unless defined $paths;
2607
2608        # look for a parent from another branch:
2609        my @b_path_components = split m#/#, $self->{path};
2610        my @a_path_components;
2611        my $i;
2612        while (@b_path_components) {
2613                $i = $paths->{'/'.join('/', @b_path_components)};
2614                last if $i && defined $i->{copyfrom_path};
2615                unshift(@a_path_components, pop(@b_path_components));
2616        }
2617        return undef unless defined $i && defined $i->{copyfrom_path};
2618        my $branch_from = $i->{copyfrom_path};
2619        if (@a_path_components) {
2620                print STDERR "branch_from: $branch_from => ";
2621                $branch_from .= '/'.join('/', @a_path_components);
2622                print STDERR $branch_from, "\n";
2623        }
2624        my $r = $i->{copyfrom_rev};
2625        my $repos_root = $self->ra->{repos_root};
2626        my $url = $self->ra->{url};
2627        my $new_url = $url . $branch_from;
2628        print STDERR  "Found possible branch point: ",
2629                      "$new_url => ", $self->full_url, ", $r\n";
2630        $branch_from =~ s#^/##;
2631        my $gs = $self->other_gs($new_url, $url,
2632                                 $branch_from, $r, $self->{ref_id});
2633        my ($r0, $parent) = $gs->find_rev_before($r, 1);
2634        {
2635                my ($base, $head);
2636                if (!defined $r0 || !defined $parent) {
2637                        ($base, $head) = parse_revision_argument(0, $r);
2638                } else {
2639                        if ($r0 < $r) {
2640                                $gs->ra->get_log([$gs->{path}], $r0 + 1, $r, 1,
2641                                        0, 1, sub { $base = $_[1] - 1 });
2642                        }
2643                }
2644                if (defined $base && $base <= $r) {
2645                        $gs->fetch($base, $r);
2646                }
2647                ($r0, $parent) = $gs->find_rev_before($r, 1);
2648        }
2649        if (defined $r0 && defined $parent) {
2650                print STDERR "Found branch parent: ($self->{ref_id}) $parent\n";
2651                my $ed;
2652                if ($self->ra->can_do_switch) {
2653                        $self->assert_index_clean($parent);
2654                        print STDERR "Following parent with do_switch\n";
2655                        # do_switch works with svn/trunk >= r22312, but that
2656                        # is not included with SVN 1.4.3 (the latest version
2657                        # at the moment), so we can't rely on it
2658                        $self->{last_rev} = $r0;
2659                        $self->{last_commit} = $parent;
2660                        $ed = SVN::Git::Fetcher->new($self, $gs->{path});
2661                        $gs->ra->gs_do_switch($r0, $rev, $gs,
2662                                              $self->full_url, $ed)
2663                          or die "SVN connection failed somewhere...\n";
2664                } elsif ($self->ra->trees_match($new_url, $r0,
2665                                                $self->full_url, $rev)) {
2666                        print STDERR "Trees match:\n",
2667                                     "  $new_url\@$r0\n",
2668                                     "  ${\$self->full_url}\@$rev\n",
2669                                     "Following parent with no changes\n";
2670                        $self->tmp_index_do(sub {
2671                            command_noisy('read-tree', $parent);
2672                        });
2673                        $self->{last_commit} = $parent;
2674                } else {
2675                        print STDERR "Following parent with do_update\n";
2676                        $ed = SVN::Git::Fetcher->new($self);
2677                        $self->ra->gs_do_update($rev, $rev, $self, $ed)
2678                          or die "SVN connection failed somewhere...\n";
2679                }
2680                print STDERR "Successfully followed parent\n";
2681                return $self->make_log_entry($rev, [$parent], $ed);
2682        }
2683        return undef;
2684}
2685
2686sub do_fetch {
2687        my ($self, $paths, $rev) = @_;
2688        my $ed;
2689        my ($last_rev, @parents);
2690        if (my $lc = $self->last_commit) {
2691                # we can have a branch that was deleted, then re-added
2692                # under the same name but copied from another path, in
2693                # which case we'll have multiple parents (we don't
2694                # want to break the original ref, nor lose copypath info):
2695                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2696                        push @{$log_entry->{parents}}, $lc;
2697                        return $log_entry;
2698                }
2699                $ed = SVN::Git::Fetcher->new($self);
2700                $last_rev = $self->{last_rev};
2701                $ed->{c} = $lc;
2702                @parents = ($lc);
2703        } else {
2704                $last_rev = $rev;
2705                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2706                        return $log_entry;
2707                }
2708                $ed = SVN::Git::Fetcher->new($self);
2709        }
2710        unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
2711                die "SVN connection failed somewhere...\n";
2712        }
2713        $self->make_log_entry($rev, \@parents, $ed);
2714}
2715
2716sub get_untracked {
2717        my ($self, $ed) = @_;
2718        my @out;
2719        my $h = $ed->{empty};
2720        foreach (sort keys %$h) {
2721                my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
2722                push @out, "  $act: " . uri_encode($_);
2723                warn "W: $act: $_\n";
2724        }
2725        foreach my $t (qw/dir_prop file_prop/) {
2726                $h = $ed->{$t} or next;
2727                foreach my $path (sort keys %$h) {
2728                        my $ppath = $path eq '' ? '.' : $path;
2729                        foreach my $prop (sort keys %{$h->{$path}}) {
2730                                next if $SKIP_PROP{$prop};
2731                                my $v = $h->{$path}->{$prop};
2732                                my $t_ppath_prop = "$t: " .
2733                                                    uri_encode($ppath) . ' ' .
2734                                                    uri_encode($prop);
2735                                if (defined $v) {
2736                                        push @out, "  +$t_ppath_prop " .
2737                                                   uri_encode($v);
2738                                } else {
2739                                        push @out, "  -$t_ppath_prop";
2740                                }
2741                        }
2742                }
2743        }
2744        foreach my $t (qw/absent_file absent_directory/) {
2745                $h = $ed->{$t} or next;
2746                foreach my $parent (sort keys %$h) {
2747                        foreach my $path (sort @{$h->{$parent}}) {
2748                                push @out, "  $t: " .
2749                                           uri_encode("$parent/$path");
2750                                warn "W: $t: $parent/$path ",
2751                                     "Insufficient permissions?\n";
2752                        }
2753                }
2754        }
2755        \@out;
2756}
2757
2758# parse_svn_date(DATE)
2759# --------------------
2760# Given a date (in UTC) from Subversion, return a string in the format
2761# "<TZ Offset> <local date/time>" that Git will use.
2762#
2763# By default the parsed date will be in UTC; if $Git::SVN::_localtime
2764# is true we'll convert it to the local timezone instead.
2765sub parse_svn_date {
2766        my $date = shift || return '+0000 1970-01-01 00:00:00';
2767        my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2768                                            (\d\d)\:(\d\d)\:(\d\d)\.\d*Z$/x) or
2769                                         croak "Unable to parse date: $date\n";
2770        my $parsed_date;    # Set next.
2771
2772        if ($Git::SVN::_localtime) {
2773                # Translate the Subversion datetime to an epoch time.
2774                # Begin by switching ourselves to $date's timezone, UTC.
2775                my $old_env_TZ = $ENV{TZ};
2776                $ENV{TZ} = 'UTC';
2777
2778                my $epoch_in_UTC =
2779                    POSIX::strftime('%s', $S, $M, $H, $d, $m - 1, $Y - 1900);
2780
2781                # Determine our local timezone (including DST) at the
2782                # time of $epoch_in_UTC.  $Git::SVN::Log::TZ stored the
2783                # value of TZ, if any, at the time we were run.
2784                if (defined $Git::SVN::Log::TZ) {
2785                        $ENV{TZ} = $Git::SVN::Log::TZ;
2786                } else {
2787                        delete $ENV{TZ};
2788                }
2789
2790                my $our_TZ =
2791                    POSIX::strftime('%Z', $S, $M, $H, $d, $m - 1, $Y - 1900);
2792
2793                # This converts $epoch_in_UTC into our local timezone.
2794                my ($sec, $min, $hour, $mday, $mon, $year,
2795                    $wday, $yday, $isdst) = localtime($epoch_in_UTC);
2796
2797                $parsed_date = sprintf('%s %04d-%02d-%02d %02d:%02d:%02d',
2798                                       $our_TZ, $year + 1900, $mon + 1,
2799                                       $mday, $hour, $min, $sec);
2800
2801                # Reset us to the timezone in effect when we entered
2802                # this routine.
2803                if (defined $old_env_TZ) {
2804                        $ENV{TZ} = $old_env_TZ;
2805                } else {
2806                        delete $ENV{TZ};
2807                }
2808        } else {
2809                $parsed_date = "+0000 $Y-$m-$d $H:$M:$S";
2810        }
2811
2812        return $parsed_date;
2813}
2814
2815sub other_gs {
2816        my ($self, $new_url, $url,
2817            $branch_from, $r, $old_ref_id) = @_;
2818        my $gs = Git::SVN->find_by_url($new_url, $url, $branch_from);
2819        unless ($gs) {
2820                my $ref_id = $old_ref_id;
2821                $ref_id =~ s/\@\d+$//;
2822                $ref_id .= "\@$r";
2823                # just grow a tail if we're not unique enough :x
2824                $ref_id .= '-' while find_ref($ref_id);
2825                print STDERR "Initializing parent: $ref_id\n";
2826                my ($u, $p, $repo_id) = ($new_url, '', $ref_id);
2827                if ($u =~ s#^\Q$url\E(/|$)##) {
2828                        $p = $u;
2829                        $u = $url;
2830                        $repo_id = $self->{repo_id};
2831                }
2832                $gs = Git::SVN->init($u, $p, $repo_id, $ref_id, 1);
2833        }
2834        $gs
2835}
2836
2837sub call_authors_prog {
2838        my ($orig_author) = @_;
2839        my $author = `$::_authors_prog $orig_author`;
2840        if ($? != 0) {
2841                die "$::_authors_prog failed with exit code $?\n"
2842        }
2843        if ($author =~ /^\s*(.+?)\s*<(.*)>\s*$/) {
2844                my ($name, $email) = ($1, $2);
2845                $email = undef if length $2 == 0;
2846                return [$name, $email];
2847        } else {
2848                die "Author: $orig_author: $::_authors_prog returned "
2849                        . "invalid author format: $author\n";
2850        }
2851}
2852
2853sub check_author {
2854        my ($author) = @_;
2855        if (!defined $author || length $author == 0) {
2856                $author = '(no author)';
2857        }
2858        if (!defined $::users{$author}) {
2859                if (defined $::_authors_prog) {
2860                        $::users{$author} = call_authors_prog($author);
2861                } elsif (defined $::_authors) {
2862                        die "Author: $author not defined in $::_authors file\n";
2863                }
2864        }
2865        $author;
2866}
2867
2868sub make_log_entry {
2869        my ($self, $rev, $parents, $ed) = @_;
2870        my $untracked = $self->get_untracked($ed);
2871
2872        open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
2873        print $un "r$rev\n" or croak $!;
2874        print $un $_, "\n" foreach @$untracked;
2875        my %log_entry = ( parents => $parents || [], revision => $rev,
2876                          log => '');
2877
2878        my $headrev;
2879        my $logged = delete $self->{logged_rev_props};
2880        if (!$logged || $self->{-want_revprops}) {
2881                my $rp = $self->ra->rev_proplist($rev);
2882                foreach (sort keys %$rp) {
2883                        my $v = $rp->{$_};
2884                        if (/^svn:(author|date|log)$/) {
2885                                $log_entry{$1} = $v;
2886                        } elsif ($_ eq 'svm:headrev') {
2887                                $headrev = $v;
2888                        } else {
2889                                print $un "  rev_prop: ", uri_encode($_), ' ',
2890                                          uri_encode($v), "\n";
2891                        }
2892                }
2893        } else {
2894                map { $log_entry{$_} = $logged->{$_} } keys %$logged;
2895        }
2896        close $un or croak $!;
2897
2898        $log_entry{date} = parse_svn_date($log_entry{date});
2899        $log_entry{log} .= "\n";
2900        my $author = $log_entry{author} = check_author($log_entry{author});
2901        my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
2902                                                       : ($author, undef);
2903
2904        my ($commit_name, $commit_email) = ($name, $email);
2905        if ($_use_log_author) {
2906                my $name_field;
2907                if ($log_entry{log} =~ /From:\s+(.*\S)\s*\n/i) {
2908                        $name_field = $1;
2909                } elsif ($log_entry{log} =~ /Signed-off-by:\s+(.*\S)\s*\n/i) {
2910                        $name_field = $1;
2911                }
2912                if (!defined $name_field) {
2913                        if (!defined $email) {
2914                                $email = $name;
2915                        }
2916                } elsif ($name_field =~ /(.*?)\s+<(.*)>/) {
2917                        ($name, $email) = ($1, $2);
2918                } elsif ($name_field =~ /(.*)@/) {
2919                        ($name, $email) = ($1, $name_field);
2920                } else {
2921                        ($name, $email) = ($name_field, $name_field);
2922                }
2923        }
2924        if (defined $headrev && $self->use_svm_props) {
2925                if ($self->rewrite_root) {
2926                        die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
2927                            "options set!\n";
2928                }
2929                my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$}i;
2930                # we don't want "SVM: initializing mirror for junk" ...
2931                return undef if $r == 0;
2932                my $svm = $self->svm;
2933                if ($uuid ne $svm->{uuid}) {
2934                        die "UUID mismatch on SVM path:\n",
2935                            "expected: $svm->{uuid}\n",
2936                            "     got: $uuid\n";
2937                }
2938                my $full_url = $self->full_url;
2939                $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
2940                             die "Failed to replace '$svm->{replace}' with ",
2941                                 "'$svm->{source}' in $full_url\n";
2942                # throw away username for storing in records
2943                remove_username($full_url);
2944                $log_entry{metadata} = "$full_url\@$r $uuid";
2945                $log_entry{svm_revision} = $r;
2946                $email ||= "$author\@$uuid";
2947                $commit_email ||= "$author\@$uuid";
2948        } elsif ($self->use_svnsync_props) {
2949                my $full_url = $self->svnsync->{url};
2950                $full_url .= "/$self->{path}" if length $self->{path};
2951                remove_username($full_url);
2952                my $uuid = $self->svnsync->{uuid};
2953                $log_entry{metadata} = "$full_url\@$rev $uuid";
2954                $email ||= "$author\@$uuid";
2955                $commit_email ||= "$author\@$uuid";
2956        } else {
2957                my $url = $self->metadata_url;
2958                remove_username($url);
2959                $log_entry{metadata} = "$url\@$rev " .
2960                                       $self->ra->get_uuid;
2961                $email ||= "$author\@" . $self->ra->get_uuid;
2962                $commit_email ||= "$author\@" . $self->ra->get_uuid;
2963        }
2964        $log_entry{name} = $name;
2965        $log_entry{email} = $email;
2966        $log_entry{commit_name} = $commit_name;
2967        $log_entry{commit_email} = $commit_email;
2968        \%log_entry;
2969}
2970
2971sub fetch {
2972        my ($self, $min_rev, $max_rev, @parents) = @_;
2973        my ($last_rev, $last_commit) = $self->last_rev_commit;
2974        my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
2975        $self->ra->gs_fetch_loop_common($base, $head, [$self]);
2976}
2977
2978sub set_tree_cb {
2979        my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
2980        $self->{inject_parents} = { $rev => $tree };
2981        $self->fetch(undef, undef);
2982}
2983
2984sub set_tree {
2985        my ($self, $tree) = (shift, shift);
2986        my $log_entry = ::get_commit_entry($tree);
2987        unless ($self->{last_rev}) {
2988                ::fatal("Must have an existing revision to commit");
2989        }
2990        my %ed_opts = ( r => $self->{last_rev},
2991                        log => $log_entry->{log},
2992                        ra => $self->ra,
2993                        tree_a => $self->{last_commit},
2994                        tree_b => $tree,
2995                        editor_cb => sub {
2996                               $self->set_tree_cb($log_entry, $tree, @_) },
2997                        svn_path => $self->{path} );
2998        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
2999                print "No changes\nr$self->{last_rev} = $tree\n";
3000        }
3001}
3002
3003sub rebuild_from_rev_db {
3004        my ($self, $path) = @_;
3005        my $r = -1;
3006        open my $fh, '<', $path or croak "open: $!";
3007        binmode $fh or croak "binmode: $!";
3008        while (<$fh>) {
3009                length($_) == 41 or croak "inconsistent size in ($_) != 41";
3010                chomp($_);
3011                ++$r;
3012                next if $_ eq ('0' x 40);
3013                $self->rev_map_set($r, $_);
3014                print "r$r = $_\n";
3015        }
3016        close $fh or croak "close: $!";
3017        unlink $path or croak "unlink: $!";
3018}
3019
3020sub rebuild {
3021        my ($self) = @_;
3022        my $map_path = $self->map_path;
3023        my $partial = (-e $map_path && ! -z $map_path);
3024        return unless ::verify_ref($self->refname.'^0');
3025        if (!$partial && ($self->use_svm_props || $self->no_metadata)) {
3026                my $rev_db = $self->rev_db_path;
3027                $self->rebuild_from_rev_db($rev_db);
3028                if ($self->use_svm_props) {
3029                        my $svm_rev_db = $self->rev_db_path($self->svm_uuid);
3030                        $self->rebuild_from_rev_db($svm_rev_db);
3031                }
3032                $self->unlink_rev_db_symlink;
3033                return;
3034        }
3035        print "Rebuilding $map_path ...\n" if (!$partial);
3036        my ($base_rev, $head) = ($partial ? $self->rev_map_max_norebuild(1) :
3037                (undef, undef));
3038        my ($log, $ctx) =
3039            command_output_pipe(qw/rev-list --pretty=raw --no-color --reverse/,
3040                                ($head ? "$head.." : "") . $self->refname,
3041                                '--');
3042        my $metadata_url = $self->metadata_url;
3043        remove_username($metadata_url);
3044        my $svn_uuid = $self->ra_uuid;
3045        my $c;
3046        while (<$log>) {
3047                if ( m{^commit ($::sha1)$} ) {
3048                        $c = $1;
3049                        next;
3050                }
3051                next unless s{^\s*(git-svn-id:)}{$1};
3052                my ($url, $rev, $uuid) = ::extract_metadata($_);
3053                remove_username($url);
3054
3055                # ignore merges (from set-tree)
3056                next if (!defined $rev || !$uuid);
3057
3058                # if we merged or otherwise started elsewhere, this is
3059                # how we break out of it
3060                if (($uuid ne $svn_uuid) ||
3061                    ($metadata_url && $url && ($url ne $metadata_url))) {
3062                        next;
3063                }
3064                if ($partial && $head) {
3065                        print "Partial-rebuilding $map_path ...\n";
3066                        print "Currently at $base_rev = $head\n";
3067                        $head = undef;
3068                }
3069
3070                $self->rev_map_set($rev, $c);
3071                print "r$rev = $c\n";
3072        }
3073        command_close_pipe($log, $ctx);
3074        print "Done rebuilding $map_path\n" if (!$partial || !$head);
3075        my $rev_db_path = $self->rev_db_path;
3076        if (-f $self->rev_db_path) {
3077                unlink $self->rev_db_path or croak "unlink: $!";
3078        }
3079        $self->unlink_rev_db_symlink;
3080}
3081
3082# rev_map:
3083# Tie::File seems to be prone to offset errors if revisions get sparse,
3084# it's not that fast, either.  Tie::File is also not in Perl 5.6.  So
3085# one of my favorite modules is out :<  Next up would be one of the DBM
3086# modules, but I'm not sure which is most portable...
3087#
3088# This is the replacement for the rev_db format, which was too big
3089# and inefficient for large repositories with a lot of sparse history
3090# (mainly tags)
3091#
3092# The format is this:
3093#   - 24 bytes for every record,
3094#     * 4 bytes for the integer representing an SVN revision number
3095#     * 20 bytes representing the sha1 of a git commit
3096#   - No empty padding records like the old format
3097#     (except the last record, which can be overwritten)
3098#   - new records are written append-only since SVN revision numbers
3099#     increase monotonically
3100#   - lookups on SVN revision number are done via a binary search
3101#   - Piping the file to xxd -c24 is a good way of dumping it for
3102#     viewing or editing (piped back through xxd -r), should the need
3103#     ever arise.
3104#   - The last record can be padding revision with an all-zero sha1
3105#     This is used to optimize fetch performance when using multiple
3106#     "fetch" directives in .git/config
3107#
3108# These files are disposable unless noMetadata or useSvmProps is set
3109
3110sub _rev_map_set {
3111        my ($fh, $rev, $commit) = @_;
3112
3113        binmode $fh or croak "binmode: $!";
3114        my $size = (stat($fh))[7];
3115        ($size % 24) == 0 or croak "inconsistent size: $size";
3116
3117        my $wr_offset = 0;
3118        if ($size > 0) {
3119                sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3120                my $read = sysread($fh, my $buf, 24) or croak "read: $!";
3121                $read == 24 or croak "read only $read bytes (!= 24)";
3122                my ($last_rev, $last_commit) = unpack(rev_map_fmt, $buf);
3123                if ($last_commit eq ('0' x40)) {
3124                        if ($size >= 48) {
3125                                sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3126                                $read = sysread($fh, $buf, 24) or
3127                                    croak "read: $!";
3128                                $read == 24 or
3129                                    croak "read only $read bytes (!= 24)";
3130                                ($last_rev, $last_commit) =
3131                                    unpack(rev_map_fmt, $buf);
3132                                if ($last_commit eq ('0' x40)) {
3133                                        croak "inconsistent .rev_map\n";
3134                                }
3135                        }
3136                        if ($last_rev >= $rev) {
3137                                croak "last_rev is higher!: $last_rev >= $rev";
3138                        }
3139                        $wr_offset = -24;
3140                }
3141        }
3142        sysseek($fh, $wr_offset, SEEK_END) or croak "seek: $!";
3143        syswrite($fh, pack(rev_map_fmt, $rev, $commit), 24) == 24 or
3144          croak "write: $!";
3145}
3146
3147sub _rev_map_reset {
3148        my ($fh, $rev, $commit) = @_;
3149        my $c = _rev_map_get($fh, $rev);
3150        $c eq $commit or die "_rev_map_reset(@_) commit $c does not match!\n";
3151        my $offset = sysseek($fh, 0, SEEK_CUR) or croak "seek: $!";
3152        truncate $fh, $offset or croak "truncate: $!";
3153}
3154
3155sub mkfile {
3156        my ($path) = @_;
3157        unless (-e $path) {
3158                my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
3159                mkpath([$dir]) unless -d $dir;
3160                open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
3161                close $fh or die "Couldn't close (create) $path: $!\n";
3162        }
3163}
3164
3165sub rev_map_set {
3166        my ($self, $rev, $commit, $update_ref, $uuid) = @_;
3167        length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
3168        my $db = $self->map_path($uuid);
3169        my $db_lock = "$db.lock";
3170        my $sig;
3171        $update_ref ||= 0;
3172        if ($update_ref) {
3173                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3174                            $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
3175        }
3176        mkfile($db);
3177
3178        $LOCKFILES{$db_lock} = 1;
3179        my $sync;
3180        # both of these options make our .rev_db file very, very important
3181        # and we can't afford to lose it because rebuild() won't work
3182        if ($self->use_svm_props || $self->no_metadata) {
3183                $sync = 1;
3184                copy($db, $db_lock) or die "rev_map_set(@_): ",
3185                                           "Failed to copy: ",
3186                                           "$db => $db_lock ($!)\n";
3187        } else {
3188                rename $db, $db_lock or die "rev_map_set(@_): ",
3189                                            "Failed to rename: ",
3190                                            "$db => $db_lock ($!)\n";
3191        }
3192
3193        sysopen(my $fh, $db_lock, O_RDWR | O_CREAT)
3194             or croak "Couldn't open $db_lock: $!\n";
3195        $update_ref eq 'reset' ? _rev_map_reset($fh, $rev, $commit) :
3196                                 _rev_map_set($fh, $rev, $commit);
3197        if ($sync) {
3198                $fh->flush or die "Couldn't flush $db_lock: $!\n";
3199                $fh->sync or die "Couldn't sync $db_lock: $!\n";
3200        }
3201        close $fh or croak $!;
3202        if ($update_ref) {
3203                $_head = $self;
3204                my $note = "";
3205                $note = " ($update_ref)" if ($update_ref !~ /^\d*$/);
3206                command_noisy('update-ref', '-m', "r$rev$note",
3207                              $self->refname, $commit);
3208        }
3209        rename $db_lock, $db or die "rev_map_set(@_): ", "Failed to rename: ",
3210                                    "$db_lock => $db ($!)\n";
3211        delete $LOCKFILES{$db_lock};
3212        if ($update_ref) {
3213                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3214                            $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
3215                kill $sig, $$ if defined $sig;
3216        }
3217}
3218
3219# If want_commit, this will return an array of (rev, commit) where
3220# commit _must_ be a valid commit in the archive.
3221# Otherwise, it'll return the max revision (whether or not the
3222# commit is valid or just a 0x40 placeholder).
3223sub rev_map_max {
3224        my ($self, $want_commit) = @_;
3225        $self->rebuild;
3226        my ($r, $c) = $self->rev_map_max_norebuild($want_commit);
3227        $want_commit ? ($r, $c) : $r;
3228}
3229
3230sub rev_map_max_norebuild {
3231        my ($self, $want_commit) = @_;
3232        my $map_path = $self->map_path;
3233        stat $map_path or return $want_commit ? (0, undef) : 0;
3234        sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3235        binmode $fh or croak "binmode: $!";
3236        my $size = (stat($fh))[7];
3237        ($size % 24) == 0 or croak "inconsistent size: $size";
3238
3239        if ($size == 0) {
3240                close $fh or croak "close: $!";
3241                return $want_commit ? (0, undef) : 0;
3242        }
3243
3244        sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3245        sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3246        my ($r, $c) = unpack(rev_map_fmt, $buf);
3247        if ($want_commit && $c eq ('0' x40)) {
3248                if ($size < 48) {
3249                        return $want_commit ? (0, undef) : 0;
3250                }
3251                sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3252                sysread($fh, $buf, 24) == 24 or croak "read: $!";
3253                ($r, $c) = unpack(rev_map_fmt, $buf);
3254                if ($c eq ('0'x40)) {
3255                        croak "Penultimate record is all-zeroes in $map_path";
3256                }
3257        }
3258        close $fh or croak "close: $!";
3259        $want_commit ? ($r, $c) : $r;
3260}
3261
3262sub rev_map_get {
3263        my ($self, $rev, $uuid) = @_;
3264        my $map_path = $self->map_path($uuid);
3265        return undef unless -e $map_path;
3266
3267        sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3268        my $c = _rev_map_get($fh, $rev);
3269        close($fh) or croak "close: $!";
3270        $c
3271}
3272
3273sub _rev_map_get {
3274        my ($fh, $rev) = @_;
3275
3276        binmode $fh or croak "binmode: $!";
3277        my $size = (stat($fh))[7];
3278        ($size % 24) == 0 or croak "inconsistent size: $size";
3279
3280        if ($size == 0) {
3281                return undef;
3282        }
3283
3284        my ($l, $u) = (0, $size - 24);
3285        my ($r, $c, $buf);
3286
3287        while ($l <= $u) {
3288                my $i = int(($l/24 + $u/24) / 2) * 24;
3289                sysseek($fh, $i, SEEK_SET) or croak "seek: $!";
3290                sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3291                my ($r, $c) = unpack(rev_map_fmt, $buf);
3292
3293                if ($r < $rev) {
3294                        $l = $i + 24;
3295                } elsif ($r > $rev) {
3296                        $u = $i - 24;
3297                } else { # $r == $rev
3298                        return $c eq ('0' x 40) ? undef : $c;
3299                }
3300        }
3301        undef;
3302}
3303
3304# Finds the first svn revision that exists on (if $eq_ok is true) or
3305# before $rev for the current branch.  It will not search any lower
3306# than $min_rev.  Returns the git commit hash and svn revision number
3307# if found, else (undef, undef).
3308sub find_rev_before {
3309        my ($self, $rev, $eq_ok, $min_rev) = @_;
3310        --$rev unless $eq_ok;
3311        $min_rev ||= 1;
3312        my $max_rev = $self->rev_map_max;
3313        $rev = $max_rev if ($rev > $max_rev);
3314        while ($rev >= $min_rev) {
3315                if (my $c = $self->rev_map_get($rev)) {
3316                        return ($rev, $c);
3317                }
3318                --$rev;
3319        }
3320        return (undef, undef);
3321}
3322
3323# Finds the first svn revision that exists on (if $eq_ok is true) or
3324# after $rev for the current branch.  It will not search any higher
3325# than $max_rev.  Returns the git commit hash and svn revision number
3326# if found, else (undef, undef).
3327sub find_rev_after {
3328        my ($self, $rev, $eq_ok, $max_rev) = @_;
3329        ++$rev unless $eq_ok;
3330        $max_rev ||= $self->rev_map_max;
3331        while ($rev <= $max_rev) {
3332                if (my $c = $self->rev_map_get($rev)) {
3333                        return ($rev, $c);
3334                }
3335                ++$rev;
3336        }
3337        return (undef, undef);
3338}
3339
3340sub _new {
3341        my ($class, $repo_id, $ref_id, $path) = @_;
3342        unless (defined $repo_id && length $repo_id) {
3343                $repo_id = $Git::SVN::default_repo_id;
3344        }
3345        unless (defined $ref_id && length $ref_id) {
3346                $_prefix = '' unless defined($_prefix);
3347                $_[2] = $ref_id =
3348                             "refs/remotes/$_prefix$Git::SVN::default_ref_id";
3349        }
3350        $_[1] = $repo_id;
3351        my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
3352
3353        # Older repos imported by us used $GIT_DIR/svn/foo instead of
3354        # $GIT_DIR/svn/refs/remotes/foo when tracking refs/remotes/foo
3355        if ($ref_id =~ m{^refs/remotes/(.*)}) {
3356                my $old_dir = "$ENV{GIT_DIR}/svn/$1";
3357                if (-d $old_dir && ! -d $dir) {
3358                        $dir = $old_dir;
3359                }
3360        }
3361
3362        $_[3] = $path = '' unless (defined $path);
3363        mkpath([$dir]);
3364        bless {
3365                ref_id => $ref_id, dir => $dir, index => "$dir/index",
3366                path => $path, config => "$ENV{GIT_DIR}/svn/config",
3367                map_root => "$dir/.rev_map", repo_id => $repo_id }, $class;
3368}
3369
3370# for read-only access of old .rev_db formats
3371sub unlink_rev_db_symlink {
3372        my ($self) = @_;
3373        my $link = $self->rev_db_path;
3374        $link =~ s/\.[\w-]+$// or croak "missing UUID at the end of $link";
3375        if (-l $link) {
3376                unlink $link or croak "unlink: $link failed!";
3377        }
3378}
3379
3380sub rev_db_path {
3381        my ($self, $uuid) = @_;
3382        my $db_path = $self->map_path($uuid);
3383        $db_path =~ s{/\.rev_map\.}{/\.rev_db\.}
3384            or croak "map_path: $db_path does not contain '/.rev_map.' !";
3385        $db_path;
3386}
3387
3388# the new replacement for .rev_db
3389sub map_path {
3390        my ($self, $uuid) = @_;
3391        $uuid ||= $self->ra_uuid;
3392        "$self->{map_root}.$uuid";
3393}
3394
3395sub uri_encode {
3396        my ($f) = @_;
3397        $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
3398        $f
3399}
3400
3401sub remove_username {
3402        $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
3403}
3404
3405package Git::SVN::Prompt;
3406use strict;
3407use warnings;
3408require SVN::Core;
3409use vars qw/$_no_auth_cache $_username/;
3410
3411sub simple {
3412        my ($cred, $realm, $default_username, $may_save, $pool) = @_;
3413        $may_save = undef if $_no_auth_cache;
3414        $default_username = $_username if defined $_username;
3415        if (defined $default_username && length $default_username) {
3416                if (defined $realm && length $realm) {
3417                        print STDERR "Authentication realm: $realm\n";
3418                        STDERR->flush;
3419                }
3420                $cred->username($default_username);
3421        } else {
3422                username($cred, $realm, $may_save, $pool);
3423        }
3424        $cred->password(_read_password("Password for '" .
3425                                       $cred->username . "': ", $realm));
3426        $cred->may_save($may_save);
3427        $SVN::_Core::SVN_NO_ERROR;
3428}
3429
3430sub ssl_server_trust {
3431        my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
3432        $may_save = undef if $_no_auth_cache;
3433        print STDERR "Error validating server certificate for '$realm':\n";
3434        {
3435                no warnings 'once';
3436                # All variables SVN::Auth::SSL::* are used only once,
3437                # so we're shutting up Perl warnings about this.
3438                if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
3439                        print STDERR " - The certificate is not issued ",
3440                            "by a trusted authority. Use the\n",
3441                            "   fingerprint to validate ",
3442                            "the certificate manually!\n";
3443                }
3444                if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
3445                        print STDERR " - The certificate hostname ",
3446                            "does not match.\n";
3447                }
3448                if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
3449                        print STDERR " - The certificate is not yet valid.\n";
3450                }
3451                if ($failures & $SVN::Auth::SSL::EXPIRED) {
3452                        print STDERR " - The certificate has expired.\n";
3453                }
3454                if ($failures & $SVN::Auth::SSL::OTHER) {
3455                        print STDERR " - The certificate has ",
3456                            "an unknown error.\n";
3457                }
3458        } # no warnings 'once'
3459        printf STDERR
3460                "Certificate information:\n".
3461                " - Hostname: %s\n".
3462                " - Valid: from %s until %s\n".
3463                " - Issuer: %s\n".
3464                " - Fingerprint: %s\n",
3465                map $cert_info->$_, qw(hostname valid_from valid_until
3466                                       issuer_dname fingerprint);
3467        my $choice;
3468prompt:
3469        print STDERR $may_save ?
3470              "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
3471              "(R)eject or accept (t)emporarily? ";
3472        STDERR->flush;
3473        $choice = lc(substr(<STDIN> || 'R', 0, 1));
3474        if ($choice =~ /^t$/i) {
3475                $cred->may_save(undef);
3476        } elsif ($choice =~ /^r$/i) {
3477                return -1;
3478        } elsif ($may_save && $choice =~ /^p$/i) {
3479                $cred->may_save($may_save);
3480        } else {
3481                goto prompt;
3482        }
3483        $cred->accepted_failures($failures);
3484        $SVN::_Core::SVN_NO_ERROR;
3485}
3486
3487sub ssl_client_cert {
3488        my ($cred, $realm, $may_save, $pool) = @_;
3489        $may_save = undef if $_no_auth_cache;
3490        print STDERR "Client certificate filename: ";
3491        STDERR->flush;
3492        chomp(my $filename = <STDIN>);
3493        $cred->cert_file($filename);
3494        $cred->may_save($may_save);
3495        $SVN::_Core::SVN_NO_ERROR;
3496}
3497
3498sub ssl_client_cert_pw {
3499        my ($cred, $realm, $may_save, $pool) = @_;
3500        $may_save = undef if $_no_auth_cache;
3501        $cred->password(_read_password("Password: ", $realm));
3502        $cred->may_save($may_save);
3503        $SVN::_Core::SVN_NO_ERROR;
3504}
3505
3506sub username {
3507        my ($cred, $realm, $may_save, $pool) = @_;
3508        $may_save = undef if $_no_auth_cache;
3509        if (defined $realm && length $realm) {
3510                print STDERR "Authentication realm: $realm\n";
3511        }
3512        my $username;
3513        if (defined $_username) {
3514                $username = $_username;
3515        } else {
3516                print STDERR "Username: ";
3517                STDERR->flush;
3518                chomp($username = <STDIN>);
3519        }
3520        $cred->username($username);
3521        $cred->may_save($may_save);
3522        $SVN::_Core::SVN_NO_ERROR;
3523}
3524
3525sub _read_password {
3526        my ($prompt, $realm) = @_;
3527        print STDERR $prompt;
3528        STDERR->flush;
3529        require Term::ReadKey;
3530        Term::ReadKey::ReadMode('noecho');
3531        my $password = '';
3532        while (defined(my $key = Term::ReadKey::ReadKey(0))) {
3533                last if $key =~ /[\012\015]/; # \n\r
3534                $password .= $key;
3535        }
3536        Term::ReadKey::ReadMode('restore');
3537        print STDERR "\n";
3538        STDERR->flush;
3539        $password;
3540}
3541
3542package SVN::Git::Fetcher;
3543use vars qw/@ISA/;
3544use strict;
3545use warnings;
3546use Carp qw/croak/;
3547use File::Temp qw/tempfile/;
3548use IO::File qw//;
3549use vars qw/$_ignore_regex/;
3550
3551# file baton members: path, mode_a, mode_b, pool, fh, blob, base
3552sub new {
3553        my ($class, $git_svn, $switch_path) = @_;
3554        my $self = SVN::Delta::Editor->new;
3555        bless $self, $class;
3556        if (exists $git_svn->{last_commit}) {
3557                $self->{c} = $git_svn->{last_commit};
3558                $self->{empty_symlinks} =
3559                                  _mark_empty_symlinks($git_svn, $switch_path);
3560        }
3561        $self->{ignore_regex} = eval { command_oneline('config', '--get',
3562                             "svn-remote.$git_svn->{repo_id}.ignore-paths") };
3563        $self->{empty} = {};
3564        $self->{dir_prop} = {};
3565        $self->{file_prop} = {};
3566        $self->{absent_dir} = {};
3567        $self->{absent_file} = {};
3568        $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
3569        $self;
3570}
3571
3572# this uses the Ra object, so it must be called before do_{switch,update},
3573# not inside them (when the Git::SVN::Fetcher object is passed) to
3574# do_{switch,update}
3575sub _mark_empty_symlinks {
3576        my ($git_svn, $switch_path) = @_;
3577        my $bool = Git::config_bool('svn.brokenSymlinkWorkaround');
3578        return {} if (!defined($bool)) || (defined($bool) && ! $bool);
3579
3580        my %ret;
3581        my ($rev, $cmt) = $git_svn->last_rev_commit;
3582        return {} unless ($rev && $cmt);
3583
3584        # allow the warning to be printed for each revision we fetch to
3585        # ensure the user sees it.  The user can also disable the workaround
3586        # on the repository even while git svn is running and the next
3587        # revision fetched will skip this expensive function.
3588        my $printed_warning;
3589        chomp(my $empty_blob = `git hash-object -t blob --stdin < /dev/null`);
3590        my ($ls, $ctx) = command_output_pipe(qw/ls-tree -r -z/, $cmt);
3591        local $/ = "\0";
3592        my $pfx = defined($switch_path) ? $switch_path : $git_svn->{path};
3593        $pfx .= '/' if length($pfx);
3594        while (<$ls>) {
3595                chomp;
3596                s/\A100644 blob $empty_blob\t//o or next;
3597                unless ($printed_warning) {
3598                        print STDERR "Scanning for empty symlinks, ",
3599                                     "this may take a while if you have ",
3600                                     "many empty files\n",
3601                                     "You may disable this with `",
3602                                     "git config svn.brokenSymlinkWorkaround ",
3603                                     "false'.\n",
3604                                     "This may be done in a different ",
3605                                     "terminal without restarting ",
3606                                     "git svn\n";
3607                        $printed_warning = 1;
3608                }
3609                my $path = $_;
3610                my (undef, $props) =
3611                               $git_svn->ra->get_file($pfx.$path, $rev, undef);
3612                if ($props->{'svn:special'}) {
3613                        $ret{$path} = 1;
3614                }
3615        }
3616        command_close_pipe($ls, $ctx);
3617        \%ret;
3618}
3619
3620# returns true if a given path is inside a ".git" directory
3621sub in_dot_git {
3622        $_[0] =~ m{(?:^|/)\.git(?:/|$)};
3623}
3624
3625# return value: 0 -- don't ignore, 1 -- ignore
3626sub is_path_ignored {
3627        my ($self, $path) = @_;
3628        return 1 if in_dot_git($path);
3629        return 1 if defined($self->{ignore_regex}) &&
3630                    $path =~ m!$self->{ignore_regex}!;
3631        return 0 unless defined($_ignore_regex);
3632        return 1 if $path =~ m!$_ignore_regex!o;
3633        return 0;
3634}
3635
3636sub set_path_strip {
3637        my ($self, $path) = @_;
3638        $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
3639}
3640
3641sub open_root {
3642        { path => '' };
3643}
3644
3645sub open_directory {
3646        my ($self, $path, $pb, $rev) = @_;
3647        { path => $path };
3648}
3649
3650sub git_path {
3651        my ($self, $path) = @_;
3652        if ($self->{path_strip}) {
3653                $path =~ s!$self->{path_strip}!! or
3654                  die "Failed to strip path '$path' ($self->{path_strip})\n";
3655        }
3656        $path;
3657}
3658
3659sub delete_entry {
3660        my ($self, $path, $rev, $pb) = @_;
3661        return undef if $self->is_path_ignored($path);
3662
3663        my $gpath = $self->git_path($path);
3664        return undef if ($gpath eq '');
3665
3666        # remove entire directories.
3667        my ($tree) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
3668                         =~ /\A040000 tree ([a-f\d]{40})\t\Q$gpath\E\0/);
3669        if ($tree) {
3670                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3671                                                     -r --name-only -z/,
3672                                                     $tree);
3673                local $/ = "\0";
3674                while (<$ls>) {
3675                        chomp;
3676                        my $rmpath = "$gpath/$_";
3677                        $self->{gii}->remove($rmpath);
3678                        print "\tD\t$rmpath\n" unless $::_q;
3679                }
3680                print "\tD\t$gpath/\n" unless $::_q;
3681                command_close_pipe($ls, $ctx);
3682                $self->{empty}->{$path} = 0
3683        } else {
3684                $self->{gii}->remove($gpath);
3685                print "\tD\t$gpath\n" unless $::_q;
3686        }
3687        undef;
3688}
3689
3690sub open_file {
3691        my ($self, $path, $pb, $rev) = @_;
3692        my ($mode, $blob);
3693
3694        goto out if $self->is_path_ignored($path);
3695
3696        my $gpath = $self->git_path($path);
3697        ($mode, $blob) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
3698                             =~ /\A(\d{6}) blob ([a-f\d]{40})\t\Q$gpath\E\0/);
3699        unless (defined $mode && defined $blob) {
3700                die "$path was not found in commit $self->{c} (r$rev)\n";
3701        }
3702        if ($mode eq '100644' && $self->{empty_symlinks}->{$path}) {
3703                $mode = '120000';
3704        }
3705out:
3706        { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
3707          pool => SVN::Pool->new, action => 'M' };
3708}
3709
3710sub add_file {
3711        my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
3712        my $mode;
3713
3714        if (!$self->is_path_ignored($path)) {
3715                my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3716                delete $self->{empty}->{$dir};
3717                $mode = '100644';
3718        }
3719        { path => $path, mode_a => $mode, mode_b => $mode,
3720          pool => SVN::Pool->new, action => 'A' };
3721}
3722
3723sub add_directory {
3724        my ($self, $path, $cp_path, $cp_rev) = @_;
3725        goto out if $self->is_path_ignored($path);
3726        my $gpath = $self->git_path($path);
3727        if ($gpath eq '') {
3728                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3729                                                     -r --name-only -z/,
3730                                                     $self->{c});
3731                local $/ = "\0";
3732                while (<$ls>) {
3733                        chomp;
3734                        $self->{gii}->remove($_);
3735                        print "\tD\t$_\n" unless $::_q;
3736                }
3737                command_close_pipe($ls, $ctx);
3738                $self->{empty}->{$path} = 0;
3739        }
3740        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3741        delete $self->{empty}->{$dir};
3742        $self->{empty}->{$path} = 1;
3743out:
3744        { path => $path };
3745}
3746
3747sub change_dir_prop {
3748        my ($self, $db, $prop, $value) = @_;
3749        return undef if $self->is_path_ignored($db->{path});
3750        $self->{dir_prop}->{$db->{path}} ||= {};
3751        $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
3752        undef;
3753}
3754
3755sub absent_directory {
3756        my ($self, $path, $pb) = @_;
3757        return undef if $self->is_path_ignored($path);
3758        $self->{absent_dir}->{$pb->{path}} ||= [];
3759        push @{$self->{absent_dir}->{$pb->{path}}}, $path;
3760        undef;
3761}
3762
3763sub absent_file {
3764        my ($self, $path, $pb) = @_;
3765        return undef if $self->is_path_ignored($path);
3766        $self->{absent_file}->{$pb->{path}} ||= [];
3767        push @{$self->{absent_file}->{$pb->{path}}}, $path;
3768        undef;
3769}
3770
3771sub change_file_prop {
3772        my ($self, $fb, $prop, $value) = @_;
3773        return undef if $self->is_path_ignored($fb->{path});
3774        if ($prop eq 'svn:executable') {
3775                if ($fb->{mode_b} != 120000) {
3776                        $fb->{mode_b} = defined $value ? 100755 : 100644;
3777                }
3778        } elsif ($prop eq 'svn:special') {
3779                $fb->{mode_b} = defined $value ? 120000 : 100644;
3780        } else {
3781                $self->{file_prop}->{$fb->{path}} ||= {};
3782                $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
3783        }
3784        undef;
3785}
3786
3787sub apply_textdelta {
3788        my ($self, $fb, $exp) = @_;
3789        return undef if $self->is_path_ignored($fb->{path});
3790        my $fh = $::_repository->temp_acquire('svn_delta');
3791        # $fh gets auto-closed() by SVN::TxDelta::apply(),
3792        # (but $base does not,) so dup() it for reading in close_file
3793        open my $dup, '<&', $fh or croak $!;
3794        my $base = $::_repository->temp_acquire('git_blob');
3795
3796        if ($fb->{blob}) {
3797                my ($base_is_link, $size);
3798
3799                if ($fb->{mode_a} eq '120000' &&
3800                    ! $self->{empty_symlinks}->{$fb->{path}}) {
3801                        print $base 'link ' or die "print $!\n";
3802                        $base_is_link = 1;
3803                }
3804        retry:
3805                $size = $::_repository->cat_blob($fb->{blob}, $base);
3806                die "Failed to read object $fb->{blob}" if ($size < 0);
3807
3808                if (defined $exp) {
3809                        seek $base, 0, 0 or croak $!;
3810                        my $got = ::md5sum($base);
3811                        if ($got ne $exp) {
3812                                my $err = "Checksum mismatch: ".
3813                                       "$fb->{path} $fb->{blob}\n" .
3814                                       "expected: $exp\n" .
3815                                       "     got: $got\n";
3816                                if ($base_is_link) {
3817                                        warn $err,
3818                                             "Retrying... (possibly ",
3819                                             "a bad symlink from SVN)\n";
3820                                        $::_repository->temp_reset($base);
3821                                        $base_is_link = 0;
3822                                        goto retry;
3823                                }
3824                                die $err;
3825                        }
3826                }
3827        }
3828        seek $base, 0, 0 or croak $!;
3829        $fb->{fh} = $fh;
3830        $fb->{base} = $base;
3831        [ SVN::TxDelta::apply($base, $dup, undef, $fb->{path}, $fb->{pool}) ];
3832}
3833
3834sub close_file {
3835        my ($self, $fb, $exp) = @_;
3836        return undef if $self->is_path_ignored($fb->{path});
3837
3838        my $hash;
3839        my $path = $self->git_path($fb->{path});
3840        if (my $fh = $fb->{fh}) {
3841                if (defined $exp) {
3842                        seek($fh, 0, 0) or croak $!;
3843                        my $got = ::md5sum($fh);
3844                        if ($got ne $exp) {
3845                                die "Checksum mismatch: $path\n",
3846                                    "expected: $exp\n    got: $got\n";
3847                        }
3848                }
3849                if ($fb->{mode_b} == 120000) {
3850                        sysseek($fh, 0, 0) or croak $!;
3851                        my $rd = sysread($fh, my $buf, 5);
3852
3853                        if (!defined $rd) {
3854                                croak "sysread: $!\n";
3855                        } elsif ($rd == 0) {
3856                                warn "$path has mode 120000",
3857                                     " but it points to nothing\n",
3858                                     "converting to an empty file with mode",
3859                                     " 100644\n";
3860                                $fb->{mode_b} = '100644';
3861                        } elsif ($buf ne 'link ') {
3862                                warn "$path has mode 120000",
3863                                     " but is not a link\n";
3864                        } else {
3865                                my $tmp_fh = $::_repository->temp_acquire(
3866                                        'svn_hash');
3867                                my $res;
3868                                while ($res = sysread($fh, my $str, 1024)) {
3869                                        my $out = syswrite($tmp_fh, $str, $res);
3870                                        defined($out) && $out == $res
3871                                                or croak("write ",
3872                                                        Git::temp_path($tmp_fh),
3873                                                        ": $!\n");
3874                                }
3875                                defined $res or croak $!;
3876
3877                                ($fh, $tmp_fh) = ($tmp_fh, $fh);
3878                                Git::temp_release($tmp_fh, 1);
3879                        }
3880                }
3881
3882                $hash = $::_repository->hash_and_insert_object(
3883                                Git::temp_path($fh));
3884                $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
3885
3886                Git::temp_release($fb->{base}, 1);
3887                Git::temp_release($fh, 1);
3888        } else {
3889                $hash = $fb->{blob} or die "no blob information\n";
3890        }
3891        $fb->{pool}->clear;
3892        $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
3893        print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
3894        undef;
3895}
3896
3897sub abort_edit {
3898        my $self = shift;
3899        $self->{nr} = $self->{gii}->{nr};
3900        delete $self->{gii};
3901        $self->SUPER::abort_edit(@_);
3902}
3903
3904sub close_edit {
3905        my $self = shift;
3906        $self->{git_commit_ok} = 1;
3907        $self->{nr} = $self->{gii}->{nr};
3908        delete $self->{gii};
3909        $self->SUPER::close_edit(@_);
3910}
3911
3912package SVN::Git::Editor;
3913use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
3914use strict;
3915use warnings;
3916use Carp qw/croak/;
3917use IO::File;
3918
3919sub new {
3920        my ($class, $opts) = @_;
3921        foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
3922                die "$_ required!\n" unless (defined $opts->{$_});
3923        }
3924
3925        my $pool = SVN::Pool->new;
3926        my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
3927        my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
3928                                     $opts->{r}, $mods);
3929
3930        # $opts->{ra} functions should not be used after this:
3931        my @ce  = $opts->{ra}->get_commit_editor($opts->{log},
3932                                                $opts->{editor_cb}, $pool);
3933        my $self = SVN::Delta::Editor->new(@ce, $pool);
3934        bless $self, $class;
3935        foreach (qw/svn_path r tree_a tree_b/) {
3936                $self->{$_} = $opts->{$_};
3937        }
3938        $self->{url} = $opts->{ra}->{url};
3939        $self->{mods} = $mods;
3940        $self->{types} = $types;
3941        $self->{pool} = $pool;
3942        $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
3943        $self->{rm} = { };
3944        $self->{path_prefix} = length $self->{svn_path} ?
3945                               "$self->{svn_path}/" : '';
3946        $self->{config} = $opts->{config};
3947        return $self;
3948}
3949
3950sub generate_diff {
3951        my ($tree_a, $tree_b) = @_;
3952        my @diff_tree = qw(diff-tree -z -r);
3953        if ($_cp_similarity) {
3954                push @diff_tree, "-C$_cp_similarity";
3955        } else {
3956                push @diff_tree, '-C';
3957        }
3958        push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
3959        push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
3960        push @diff_tree, $tree_a, $tree_b;
3961        my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
3962        local $/ = "\0";
3963        my $state = 'meta';
3964        my @mods;
3965        while (<$diff_fh>) {
3966                chomp $_; # this gets rid of the trailing "\0"
3967                if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
3968                                        ($::sha1)\s($::sha1)\s
3969                                        ([MTCRAD])\d*$/xo) {
3970                        push @mods, {   mode_a => $1, mode_b => $2,
3971                                        sha1_a => $3, sha1_b => $4,
3972                                        chg => $5 };
3973                        if ($5 =~ /^(?:C|R)$/) {
3974                                $state = 'file_a';
3975                        } else {
3976                                $state = 'file_b';
3977                        }
3978                } elsif ($state eq 'file_a') {
3979                        my $x = $mods[$#mods] or croak "Empty array\n";
3980                        if ($x->{chg} !~ /^(?:C|R)$/) {
3981                                croak "Error parsing $_, $x->{chg}\n";
3982                        }
3983                        $x->{file_a} = $_;
3984                        $state = 'file_b';
3985                } elsif ($state eq 'file_b') {
3986                        my $x = $mods[$#mods] or croak "Empty array\n";
3987                        if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
3988                                croak "Error parsing $_, $x->{chg}\n";
3989                        }
3990                        if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
3991                                croak "Error parsing $_, $x->{chg}\n";
3992                        }
3993                        $x->{file_b} = $_;
3994                        $state = 'meta';
3995                } else {
3996                        croak "Error parsing $_\n";
3997                }
3998        }
3999        command_close_pipe($diff_fh, $ctx);
4000        \@mods;
4001}
4002
4003sub check_diff_paths {
4004        my ($ra, $pfx, $rev, $mods) = @_;
4005        my %types;
4006        $pfx .= '/' if length $pfx;
4007
4008        sub type_diff_paths {
4009                my ($ra, $types, $path, $rev) = @_;
4010                my @p = split m#/+#, $path;
4011                my $c = shift @p;
4012                unless (defined $types->{$c}) {
4013                        $types->{$c} = $ra->check_path($c, $rev);
4014                }
4015                while (@p) {
4016                        $c .= '/' . shift @p;
4017                        next if defined $types->{$c};
4018                        $types->{$c} = $ra->check_path($c, $rev);
4019                }
4020        }
4021
4022        foreach my $m (@$mods) {
4023                foreach my $f (qw/file_a file_b/) {
4024                        next unless defined $m->{$f};
4025                        my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
4026                        if (length $pfx.$dir && ! defined $types{$dir}) {
4027                                type_diff_paths($ra, \%types, $pfx.$dir, $rev);
4028                        }
4029                }
4030        }
4031        \%types;
4032}
4033
4034sub split_path {
4035        return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
4036}
4037
4038sub repo_path {
4039        my ($self, $path) = @_;
4040        $self->{path_prefix}.(defined $path ? $path : '');
4041}
4042
4043sub url_path {
4044        my ($self, $path) = @_;
4045        if ($self->{url} =~ m#^https?://#) {
4046                $path =~ s!([^~a-zA-Z0-9_./-])!uc sprintf("%%%02x",ord($1))!eg;
4047        }
4048        $self->{url} . '/' . $self->repo_path($path);
4049}
4050
4051sub rmdirs {
4052        my ($self) = @_;
4053        my $rm = $self->{rm};
4054        delete $rm->{''}; # we never delete the url we're tracking
4055        return unless %$rm;
4056
4057        foreach (keys %$rm) {
4058                my @d = split m#/#, $_;
4059                my $c = shift @d;
4060                $rm->{$c} = 1;
4061                while (@d) {
4062                        $c .= '/' . shift @d;
4063                        $rm->{$c} = 1;
4064                }
4065        }
4066        delete $rm->{$self->{svn_path}};
4067        delete $rm->{''}; # we never delete the url we're tracking
4068        return unless %$rm;
4069
4070        my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
4071                                             $self->{tree_b});
4072        local $/ = "\0";
4073        while (<$fh>) {
4074                chomp;
4075                my @dn = split m#/#, $_;
4076                while (pop @dn) {
4077                        delete $rm->{join '/', @dn};
4078                }
4079                unless (%$rm) {
4080                        close $fh;
4081                        return;
4082                }
4083        }
4084        command_close_pipe($fh, $ctx);
4085
4086        my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
4087        foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
4088                $self->close_directory($bat->{$d}, $p);
4089                my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
4090                print "\tD+\t$d/\n" unless $::_q;
4091                $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
4092                delete $bat->{$d};
4093        }
4094}
4095
4096sub open_or_add_dir {
4097        my ($self, $full_path, $baton) = @_;
4098        my $t = $self->{types}->{$full_path};
4099        if (!defined $t) {
4100                die "$full_path not known in r$self->{r} or we have a bug!\n";
4101        }
4102        {
4103                no warnings 'once';
4104                # SVN::Node::none and SVN::Node::file are used only once,
4105                # so we're shutting up Perl's warnings about them.
4106                if ($t == $SVN::Node::none) {
4107                        return $self->add_directory($full_path, $baton,
4108                            undef, -1, $self->{pool});
4109                } elsif ($t == $SVN::Node::dir) {
4110                        return $self->open_directory($full_path, $baton,
4111                            $self->{r}, $self->{pool});
4112                } # no warnings 'once'
4113                print STDERR "$full_path already exists in repository at ",
4114                    "r$self->{r} and it is not a directory (",
4115                    ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
4116        } # no warnings 'once'
4117        exit 1;
4118}
4119
4120sub ensure_path {
4121        my ($self, $path) = @_;
4122        my $bat = $self->{bat};
4123        my $repo_path = $self->repo_path($path);
4124        return $bat->{''} unless (length $repo_path);
4125        my @p = split m#/+#, $repo_path;
4126        my $c = shift @p;
4127        $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
4128        while (@p) {
4129                my $c0 = $c;
4130                $c .= '/' . shift @p;
4131                $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
4132        }
4133        return $bat->{$c};
4134}
4135
4136# Subroutine to convert a globbing pattern to a regular expression.
4137# From perl cookbook.
4138sub glob2pat {
4139        my $globstr = shift;
4140        my %patmap = ('*' => '.*', '?' => '.', '[' => '[', ']' => ']');
4141        $globstr =~ s{(.)} { $patmap{$1} || "\Q$1" }ge;
4142        return '^' . $globstr . '$';
4143}
4144
4145sub check_autoprop {
4146        my ($self, $pattern, $properties, $file, $fbat) = @_;
4147        # Convert the globbing pattern to a regular expression.
4148        my $regex = glob2pat($pattern);
4149        # Check if the pattern matches the file name.
4150        if($file =~ m/($regex)/) {
4151                # Parse the list of properties to set.
4152                my @props = split(/;/, $properties);
4153                foreach my $prop (@props) {
4154                        # Parse 'name=value' syntax and set the property.
4155                        if ($prop =~ /([^=]+)=(.*)/) {
4156                                my ($n,$v) = ($1,$2);
4157                                for ($n, $v) {
4158                                        s/^\s+//; s/\s+$//;
4159                                }
4160                                $self->change_file_prop($fbat, $n, $v);
4161                        }
4162                }
4163        }
4164}
4165
4166sub apply_autoprops {
4167        my ($self, $file, $fbat) = @_;
4168        my $conf_t = ${$self->{config}}{'config'};
4169        no warnings 'once';
4170        # Check [miscellany]/enable-auto-props in svn configuration.
4171        if (SVN::_Core::svn_config_get_bool(
4172                $conf_t,
4173                $SVN::_Core::SVN_CONFIG_SECTION_MISCELLANY,
4174                $SVN::_Core::SVN_CONFIG_OPTION_ENABLE_AUTO_PROPS,
4175                0)) {
4176                # Auto-props are enabled.  Enumerate them to look for matches.
4177                my $callback = sub {
4178                        $self->check_autoprop($_[0], $_[1], $file, $fbat);
4179                };
4180                SVN::_Core::svn_config_enumerate(
4181                        $conf_t,
4182                        $SVN::_Core::SVN_CONFIG_SECTION_AUTO_PROPS,
4183                        $callback);
4184        }
4185}
4186
4187sub A {
4188        my ($self, $m) = @_;
4189        my ($dir, $file) = split_path($m->{file_b});
4190        my $pbat = $self->ensure_path($dir);
4191        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4192                                        undef, -1);
4193        print "\tA\t$m->{file_b}\n" unless $::_q;
4194        $self->apply_autoprops($file, $fbat);
4195        $self->chg_file($fbat, $m);
4196        $self->close_file($fbat,undef,$self->{pool});
4197}
4198
4199sub C {
4200        my ($self, $m) = @_;
4201        my ($dir, $file) = split_path($m->{file_b});
4202        my $pbat = $self->ensure_path($dir);
4203        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4204                                $self->url_path($m->{file_a}), $self->{r});
4205        print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
4206        $self->chg_file($fbat, $m);
4207        $self->close_file($fbat,undef,$self->{pool});
4208}
4209
4210sub delete_entry {
4211        my ($self, $path, $pbat) = @_;
4212        my $rpath = $self->repo_path($path);
4213        my ($dir, $file) = split_path($rpath);
4214        $self->{rm}->{$dir} = 1;
4215        $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
4216}
4217
4218sub R {
4219        my ($self, $m) = @_;
4220        my ($dir, $file) = split_path($m->{file_b});
4221        my $pbat = $self->ensure_path($dir);
4222        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4223                                $self->url_path($m->{file_a}), $self->{r});
4224        print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
4225        $self->apply_autoprops($file, $fbat);
4226        $self->chg_file($fbat, $m);
4227        $self->close_file($fbat,undef,$self->{pool});
4228
4229        ($dir, $file) = split_path($m->{file_a});
4230        $pbat = $self->ensure_path($dir);
4231        $self->delete_entry($m->{file_a}, $pbat);
4232}
4233
4234sub M {
4235        my ($self, $m) = @_;
4236        my ($dir, $file) = split_path($m->{file_b});
4237        my $pbat = $self->ensure_path($dir);
4238        my $fbat = $self->open_file($self->repo_path($m->{file_b}),
4239                                $pbat,$self->{r},$self->{pool});
4240        print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
4241        $self->chg_file($fbat, $m);
4242        $self->close_file($fbat,undef,$self->{pool});
4243}
4244
4245sub T { shift->M(@_) }
4246
4247sub change_file_prop {
4248        my ($self, $fbat, $pname, $pval) = @_;
4249        $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
4250}
4251
4252sub _chg_file_get_blob ($$$$) {
4253        my ($self, $fbat, $m, $which) = @_;
4254        my $fh = $::_repository->temp_acquire("git_blob_$which");
4255        if ($m->{"mode_$which"} =~ /^120/) {
4256                print $fh 'link ' or croak $!;
4257                $self->change_file_prop($fbat,'svn:special','*');
4258        } elsif ($m->{mode_a} =~ /^120/ && $m->{"mode_$which"} !~ /^120/) {
4259                $self->change_file_prop($fbat,'svn:special',undef);
4260        }
4261        my $blob = $m->{"sha1_$which"};
4262        return ($fh,) if ($blob =~ /^0{40}$/);
4263        my $size = $::_repository->cat_blob($blob, $fh);
4264        croak "Failed to read object $blob" if ($size < 0);
4265        $fh->flush == 0 or croak $!;
4266        seek $fh, 0, 0 or croak $!;
4267
4268        my $exp = ::md5sum($fh);
4269        seek $fh, 0, 0 or croak $!;
4270        return ($fh, $exp);
4271}
4272
4273sub chg_file {
4274        my ($self, $fbat, $m) = @_;
4275        if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
4276                $self->change_file_prop($fbat,'svn:executable','*');
4277        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
4278                $self->change_file_prop($fbat,'svn:executable',undef);
4279        }
4280        my ($fh_a, $exp_a) = _chg_file_get_blob $self, $fbat, $m, 'a';
4281        my ($fh_b, $exp_b) = _chg_file_get_blob $self, $fbat, $m, 'b';
4282        my $pool = SVN::Pool->new;
4283        my $atd = $self->apply_textdelta($fbat, $exp_a, $pool);
4284        if (-s $fh_a) {
4285                my $txstream = SVN::TxDelta::new ($fh_a, $fh_b, $pool);
4286                my $res = SVN::TxDelta::send_txstream($txstream, @$atd, $pool);
4287                if (defined $res) {
4288                        die "Unexpected result from send_txstream: $res\n",
4289                            "(SVN::Core::VERSION: $SVN::Core::VERSION)\n";
4290                }
4291        } else {
4292                my $got = SVN::TxDelta::send_stream($fh_b, @$atd, $pool);
4293                die "Checksum mismatch\nexpected: $exp_b\ngot: $got\n"
4294                    if ($got ne $exp_b);
4295        }
4296        Git::temp_release($fh_b, 1);
4297        Git::temp_release($fh_a, 1);
4298        $pool->clear;
4299}
4300
4301sub D {
4302        my ($self, $m) = @_;
4303        my ($dir, $file) = split_path($m->{file_b});
4304        my $pbat = $self->ensure_path($dir);
4305        print "\tD\t$m->{file_b}\n" unless $::_q;
4306        $self->delete_entry($m->{file_b}, $pbat);
4307}
4308
4309sub close_edit {
4310        my ($self) = @_;
4311        my ($p,$bat) = ($self->{pool}, $self->{bat});
4312        foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
4313                next if $_ eq '';
4314                $self->close_directory($bat->{$_}, $p);
4315        }
4316        $self->close_directory($bat->{''}, $p);
4317        $self->SUPER::close_edit($p);
4318        $p->clear;
4319}
4320
4321sub abort_edit {
4322        my ($self) = @_;
4323        $self->SUPER::abort_edit($self->{pool});
4324}
4325
4326sub DESTROY {
4327        my $self = shift;
4328        $self->SUPER::DESTROY(@_);
4329        $self->{pool}->clear;
4330}
4331
4332# this drives the editor
4333sub apply_diff {
4334        my ($self) = @_;
4335        my $mods = $self->{mods};
4336        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
4337        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
4338                my $f = $m->{chg};
4339                if (defined $o{$f}) {
4340                        $self->$f($m);
4341                } else {
4342                        fatal("Invalid change type: $f");
4343                }
4344        }
4345        $self->rmdirs if $_rmdir;
4346        if (@$mods == 0) {
4347                $self->abort_edit;
4348        } else {
4349                $self->close_edit;
4350        }
4351        return scalar @$mods;
4352}
4353
4354package Git::SVN::Ra;
4355use vars qw/@ISA $config_dir $_log_window_size/;
4356use strict;
4357use warnings;
4358my ($ra_invalid, $can_do_switch, %ignored_err, $RA);
4359
4360BEGIN {
4361        # enforce temporary pool usage for some simple functions
4362        no strict 'refs';
4363        for my $f (qw/rev_proplist get_latest_revnum get_uuid get_repos_root
4364                      get_file/) {
4365                my $SUPER = "SUPER::$f";
4366                *$f = sub {
4367                        my $self = shift;
4368                        my $pool = SVN::Pool->new;
4369                        my @ret = $self->$SUPER(@_,$pool);
4370                        $pool->clear;
4371                        wantarray ? @ret : $ret[0];
4372                };
4373        }
4374}
4375
4376sub _auth_providers () {
4377        [
4378          SVN::Client::get_simple_provider(),
4379          SVN::Client::get_ssl_server_trust_file_provider(),
4380          SVN::Client::get_simple_prompt_provider(
4381            \&Git::SVN::Prompt::simple, 2),
4382          SVN::Client::get_ssl_client_cert_file_provider(),
4383          SVN::Client::get_ssl_client_cert_prompt_provider(
4384            \&Git::SVN::Prompt::ssl_client_cert, 2),
4385          SVN::Client::get_ssl_client_cert_pw_file_provider(),
4386          SVN::Client::get_ssl_client_cert_pw_prompt_provider(
4387            \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
4388          SVN::Client::get_username_provider(),
4389          SVN::Client::get_ssl_server_trust_prompt_provider(
4390            \&Git::SVN::Prompt::ssl_server_trust),
4391          SVN::Client::get_username_prompt_provider(
4392            \&Git::SVN::Prompt::username, 2)
4393        ]
4394}
4395
4396sub escape_uri_only {
4397        my ($uri) = @_;
4398        my @tmp;
4399        foreach (split m{/}, $uri) {
4400                s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
4401                push @tmp, $_;
4402        }
4403        join('/', @tmp);
4404}
4405
4406sub escape_url {
4407        my ($url) = @_;
4408        if ($url =~ m#^(https?)://([^/]+)(.*)$#) {
4409                my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
4410                $url = "$scheme://$domain$uri";
4411        }
4412        $url;
4413}
4414
4415sub new {
4416        my ($class, $url) = @_;
4417        $url =~ s!/+$!!;
4418        return $RA if ($RA && $RA->{url} eq $url);
4419
4420        SVN::_Core::svn_config_ensure($config_dir, undef);
4421        my ($baton, $callbacks) = SVN::Core::auth_open_helper(_auth_providers);
4422        my $config = SVN::Core::config_get_config($config_dir);
4423        $RA = undef;
4424        my $dont_store_passwords = 1;
4425        my $conf_t = ${$config}{'config'};
4426        {
4427                no warnings 'once';
4428                # The usage of $SVN::_Core::SVN_CONFIG_* variables
4429                # produces warnings that variables are used only once.
4430                # I had not found the better way to shut them up, so
4431                # the warnings of type 'once' are disabled in this block.
4432                if (SVN::_Core::svn_config_get_bool($conf_t,
4433                    $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4434                    $SVN::_Core::SVN_CONFIG_OPTION_STORE_PASSWORDS,
4435                    1) == 0) {
4436                        SVN::_Core::svn_auth_set_parameter($baton,
4437                            $SVN::_Core::SVN_AUTH_PARAM_DONT_STORE_PASSWORDS,
4438                            bless (\$dont_store_passwords, "_p_void"));
4439                }
4440                if (SVN::_Core::svn_config_get_bool($conf_t,
4441                    $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4442                    $SVN::_Core::SVN_CONFIG_OPTION_STORE_AUTH_CREDS,
4443                    1) == 0) {
4444                        $Git::SVN::Prompt::_no_auth_cache = 1;
4445                }
4446        } # no warnings 'once'
4447        my $self = SVN::Ra->new(url => escape_url($url), auth => $baton,
4448                              config => $config,
4449                              pool => SVN::Pool->new,
4450                              auth_provider_callbacks => $callbacks);
4451        $self->{url} = $url;
4452        $self->{svn_path} = $url;
4453        $self->{repos_root} = $self->get_repos_root;
4454        $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
4455        $self->{cache} = { check_path => { r => 0, data => {} },
4456                           get_dir => { r => 0, data => {} } };
4457        $RA = bless $self, $class;
4458}
4459
4460sub check_path {
4461        my ($self, $path, $r) = @_;
4462        my $cache = $self->{cache}->{check_path};
4463        if ($r == $cache->{r} && exists $cache->{data}->{$path}) {
4464                return $cache->{data}->{$path};
4465        }
4466        my $pool = SVN::Pool->new;
4467        my $t = $self->SUPER::check_path($path, $r, $pool);
4468        $pool->clear;
4469        if ($r != $cache->{r}) {
4470                %{$cache->{data}} = ();
4471                $cache->{r} = $r;
4472        }
4473        $cache->{data}->{$path} = $t;
4474}
4475
4476sub get_dir {
4477        my ($self, $dir, $r) = @_;
4478        my $cache = $self->{cache}->{get_dir};
4479        if ($r == $cache->{r}) {
4480                if (my $x = $cache->{data}->{$dir}) {
4481                        return wantarray ? @$x : $x->[0];
4482                }
4483        }
4484        my $pool = SVN::Pool->new;
4485        my ($d, undef, $props) = $self->SUPER::get_dir($dir, $r, $pool);
4486        my %dirents = map { $_ => { kind => $d->{$_}->kind } } keys %$d;
4487        $pool->clear;
4488        if ($r != $cache->{r}) {
4489                %{$cache->{data}} = ();
4490                $cache->{r} = $r;
4491        }
4492        $cache->{data}->{$dir} = [ \%dirents, $r, $props ];
4493        wantarray ? (\%dirents, $r, $props) : \%dirents;
4494}
4495
4496sub DESTROY {
4497        # do not call the real DESTROY since we store ourselves in $RA
4498}
4499
4500# get_log(paths, start, end, limit,
4501#         discover_changed_paths, strict_node_history, receiver)
4502sub get_log {
4503        my ($self, @args) = @_;
4504        my $pool = SVN::Pool->new;
4505
4506        # svn_log_changed_path_t objects passed to get_log are likely to be
4507        # overwritten even if only the refs are copied to an external variable,
4508        # so we should dup the structures in their entirety.  Using an
4509        # externally passed pool (instead of our temporary and quickly cleared
4510        # pool in Git::SVN::Ra) does not help matters at all...
4511        my $receiver = pop @args;
4512        my $prefix = "/".$self->{svn_path};
4513        $prefix =~ s#/+($)##;
4514        my $prefix_regex = qr#^\Q$prefix\E#;
4515        push(@args, sub {
4516                my ($paths) = $_[0];
4517                return &$receiver(@_) unless $paths;
4518                $_[0] = ();
4519                foreach my $p (keys %$paths) {
4520                        my $i = $paths->{$p};
4521                        # Make path relative to our url, not repos_root
4522                        $p =~ s/$prefix_regex//;
4523                        my %s = map { $_ => $i->$_; }
4524                                qw/copyfrom_path copyfrom_rev action/;
4525                        if ($s{'copyfrom_path'}) {
4526                                $s{'copyfrom_path'} =~ s/$prefix_regex//;
4527                        }
4528                        $_[0]{$p} = \%s;
4529                }
4530                &$receiver(@_);
4531        });
4532
4533
4534        # the limit parameter was not supported in SVN 1.1.x, so we
4535        # drop it.  Therefore, the receiver callback passed to it
4536        # is made aware of this limitation by being wrapped if
4537        # the limit passed to is being wrapped.
4538        if ($SVN::Core::VERSION le '1.2.0') {
4539                my $limit = splice(@args, 3, 1);
4540                if ($limit > 0) {
4541                        my $receiver = pop @args;
4542                        push(@args, sub { &$receiver(@_) if (--$limit >= 0) });
4543                }
4544        }
4545        my $ret = $self->SUPER::get_log(@args, $pool);
4546        $pool->clear;
4547        $ret;
4548}
4549
4550sub trees_match {
4551        my ($self, $url1, $rev1, $url2, $rev2) = @_;
4552        my $ctx = SVN::Client->new(auth => _auth_providers);
4553        my $out = IO::File->new_tmpfile;
4554
4555        # older SVN (1.1.x) doesn't take $pool as the last parameter for
4556        # $ctx->diff(), so we'll create a default one
4557        my $pool = SVN::Pool->new_default_sub;
4558
4559        $ra_invalid = 1; # this will open a new SVN::Ra connection to $url1
4560        $ctx->diff([], $url1, $rev1, $url2, $rev2, 1, 1, 0, $out, $out);
4561        $out->flush;
4562        my $ret = (($out->stat)[7] == 0);
4563        close $out or croak $!;
4564
4565        $ret;
4566}
4567
4568sub get_commit_editor {
4569        my ($self, $log, $cb, $pool) = @_;
4570        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
4571        $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
4572}
4573
4574sub gs_do_update {
4575        my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
4576        my $new = ($rev_a == $rev_b);
4577        my $path = $gs->{path};
4578
4579        if ($new && -e $gs->{index}) {
4580                unlink $gs->{index} or die
4581                  "Couldn't unlink index: $gs->{index}: $!\n";
4582        }
4583        my $pool = SVN::Pool->new;
4584        $editor->set_path_strip($path);
4585        my (@pc) = split m#/#, $path;
4586        my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
4587                                        1, $editor, $pool);
4588        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
4589
4590        # Since we can't rely on svn_ra_reparent being available, we'll
4591        # just have to do some magic with set_path to make it so
4592        # we only want a partial path.
4593        my $sp = '';
4594        my $final = join('/', @pc);
4595        while (@pc) {
4596                $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
4597                $sp .= '/' if length $sp;
4598                $sp .= shift @pc;
4599        }
4600        die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
4601
4602        $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
4603
4604        $reporter->finish_report($pool);
4605        $pool->clear;
4606        $editor->{git_commit_ok};
4607}
4608
4609# this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
4610# svn_ra_reparent didn't work before 1.4)
4611sub gs_do_switch {
4612        my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
4613        my $path = $gs->{path};
4614        my $pool = SVN::Pool->new;
4615
4616        my $full_url = $self->{url};
4617        my $old_url = $full_url;
4618        $full_url .= '/' . $path if length $path;
4619        my ($ra, $reparented);
4620
4621        if ($old_url =~ m#^svn(\+ssh)?://# ||
4622            ($full_url =~ m#^https?://# &&
4623             escape_url($full_url) ne $full_url)) {
4624                $_[0] = undef;
4625                $self = undef;
4626                $RA = undef;
4627                $ra = Git::SVN::Ra->new($full_url);
4628                $ra_invalid = 1;
4629        } elsif ($old_url ne $full_url) {
4630                SVN::_Ra::svn_ra_reparent($self->{session}, $full_url, $pool);
4631                $self->{url} = $full_url;
4632                $reparented = 1;
4633        }
4634
4635        $ra ||= $self;
4636        $url_b = escape_url($url_b);
4637        my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
4638        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
4639        $reporter->set_path('', $rev_a, 0, @lock, $pool);
4640        $reporter->finish_report($pool);
4641
4642        if ($reparented) {
4643                SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
4644                $self->{url} = $old_url;
4645        }
4646
4647        $pool->clear;
4648        $editor->{git_commit_ok};
4649}
4650
4651sub longest_common_path {
4652        my ($gsv, $globs) = @_;
4653        my %common;
4654        my $common_max = scalar @$gsv;
4655
4656        foreach my $gs (@$gsv) {
4657                my @tmp = split m#/#, $gs->{path};
4658                my $p = '';
4659                foreach (@tmp) {
4660                        $p .= length($p) ? "/$_" : $_;
4661                        $common{$p} ||= 0;
4662                        $common{$p}++;
4663                }
4664        }
4665        $globs ||= [];
4666        $common_max += scalar @$globs;
4667        foreach my $glob (@$globs) {
4668                my @tmp = split m#/#, $glob->{path}->{left};
4669                my $p = '';
4670                foreach (@tmp) {
4671                        $p .= length($p) ? "/$_" : $_;
4672                        $common{$p} ||= 0;
4673                        $common{$p}++;
4674                }
4675        }
4676
4677        my $longest_path = '';
4678        foreach (sort {length $b <=> length $a} keys %common) {
4679                if ($common{$_} == $common_max) {
4680                        $longest_path = $_;
4681                        last;
4682                }
4683        }
4684        $longest_path;
4685}
4686
4687sub gs_fetch_loop_common {
4688        my ($self, $base, $head, $gsv, $globs) = @_;
4689        return if ($base > $head);
4690        my $inc = $_log_window_size;
4691        my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
4692        my $longest_path = longest_common_path($gsv, $globs);
4693        my $ra_url = $self->{url};
4694        my $find_trailing_edge;
4695        while (1) {
4696                my %revs;
4697                my $err;
4698                my $err_handler = $SVN::Error::handler;
4699                $SVN::Error::handler = sub {
4700                        ($err) = @_;
4701                        skip_unknown_revs($err);
4702                };
4703                sub _cb {
4704                        my ($paths, $r, $author, $date, $log) = @_;
4705                        [ $paths,
4706                          { author => $author, date => $date, log => $log } ];
4707                }
4708                $self->get_log([$longest_path], $min, $max, 0, 1, 1,
4709                               sub { $revs{$_[1]} = _cb(@_) });
4710                if ($err) {
4711                        print "Checked through r$max\r";
4712                } else {
4713                        $find_trailing_edge = 1;
4714                }
4715                if ($err and $find_trailing_edge) {
4716                        print STDERR "Path '$longest_path' ",
4717                                     "was probably deleted:\n",
4718                                     $err->expanded_message,
4719                                     "\nWill attempt to follow ",
4720                                     "revisions r$min .. r$max ",
4721                                     "committed before the deletion\n";
4722                        my $hi = $max;
4723                        while (--$hi >= $min) {
4724                                my $ok;
4725                                $self->get_log([$longest_path], $min, $hi,
4726                                               0, 1, 1, sub {
4727                                               $ok = $_[1];
4728                                               $revs{$_[1]} = _cb(@_) });
4729                                if ($ok) {
4730                                        print STDERR "r$min .. r$ok OK\n";
4731                                        last;
4732                                }
4733                        }
4734                        $find_trailing_edge = 0;
4735                }
4736                $SVN::Error::handler = $err_handler;
4737
4738                my %exists = map { $_->{path} => $_ } @$gsv;
4739                foreach my $r (sort {$a <=> $b} keys %revs) {
4740                        my ($paths, $logged) = @{$revs{$r}};
4741
4742                        foreach my $gs ($self->match_globs(\%exists, $paths,
4743                                                           $globs, $r)) {
4744                                if ($gs->rev_map_max >= $r) {
4745                                        next;
4746                                }
4747                                next unless $gs->match_paths($paths, $r);
4748                                $gs->{logged_rev_props} = $logged;
4749                                if (my $last_commit = $gs->last_commit) {
4750                                        $gs->assert_index_clean($last_commit);
4751                                }
4752                                my $log_entry = $gs->do_fetch($paths, $r);
4753                                if ($log_entry) {
4754                                        $gs->do_git_commit($log_entry);
4755                                }
4756                                $INDEX_FILES{$gs->{index}} = 1;
4757                        }
4758                        foreach my $g (@$globs) {
4759                                my $k = "svn-remote.$g->{remote}." .
4760                                        "$g->{t}-maxRev";
4761                                Git::SVN::tmp_config($k, $r);
4762                        }
4763                        if ($ra_invalid) {
4764                                $_[0] = undef;
4765                                $self = undef;
4766                                $RA = undef;
4767                                $self = Git::SVN::Ra->new($ra_url);
4768                                $ra_invalid = undef;
4769                        }
4770                }
4771                # pre-fill the .rev_db since it'll eventually get filled in
4772                # with '0' x40 if something new gets committed
4773                foreach my $gs (@$gsv) {
4774                        next if $gs->rev_map_max >= $max;
4775                        next if defined $gs->rev_map_get($max);
4776                        $gs->rev_map_set($max, 0 x40);
4777                }
4778                foreach my $g (@$globs) {
4779                        my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
4780                        Git::SVN::tmp_config($k, $max);
4781                }
4782                last if $max >= $head;
4783                $min = $max + 1;
4784                $max += $inc;
4785                $max = $head if ($max > $head);
4786        }
4787        Git::SVN::gc();
4788}
4789
4790sub get_dir_globbed {
4791        my ($self, $left, $depth, $r) = @_;
4792
4793        my @x = eval { $self->get_dir($left, $r) };
4794        return unless scalar @x == 3;
4795        my $dirents = $x[0];
4796        my @finalents;
4797        foreach my $de (keys %$dirents) {
4798                next if $dirents->{$de}->{kind} != $SVN::Node::dir;
4799                if ($depth > 1) {
4800                        my @args = ("$left/$de", $depth - 1, $r);
4801                        foreach my $dir ($self->get_dir_globbed(@args)) {
4802                                push @finalents, "$de/$dir";
4803                        }
4804                } else {
4805                        push @finalents, $de;
4806                }
4807        }
4808        @finalents;
4809}
4810
4811sub match_globs {
4812        my ($self, $exists, $paths, $globs, $r) = @_;
4813
4814        sub get_dir_check {
4815                my ($self, $exists, $g, $r) = @_;
4816
4817                my @dirs = $self->get_dir_globbed($g->{path}->{left},
4818                                                  $g->{path}->{depth},
4819                                                  $r);
4820
4821                foreach my $de (@dirs) {
4822                        my $p = $g->{path}->full_path($de);
4823                        next if $exists->{$p};
4824                        next if (length $g->{path}->{right} &&
4825                                 ($self->check_path($p, $r) !=
4826                                  $SVN::Node::dir));
4827                        $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
4828                                         $g->{ref}->full_path($de), 1);
4829                }
4830        }
4831        foreach my $g (@$globs) {
4832                if (my $path = $paths->{"/$g->{path}->{left}"}) {
4833                        if ($path->{action} =~ /^[AR]$/) {
4834                                get_dir_check($self, $exists, $g, $r);
4835                        }
4836                }
4837                foreach (keys %$paths) {
4838                        if (/$g->{path}->{left_regex}/ &&
4839                            !/$g->{path}->{regex}/) {
4840                                next if $paths->{$_}->{action} !~ /^[AR]$/;
4841                                get_dir_check($self, $exists, $g, $r);
4842                        }
4843                        next unless /$g->{path}->{regex}/;
4844                        my $p = $1;
4845                        my $pathname = $g->{path}->full_path($p);
4846                        next if $exists->{$pathname};
4847                        next if ($self->check_path($pathname, $r) !=
4848                                 $SVN::Node::dir);
4849                        $exists->{$pathname} = Git::SVN->init(
4850                                              $self->{url}, $pathname, undef,
4851                                              $g->{ref}->full_path($p), 1);
4852                }
4853                my $c = '';
4854                foreach (split m#/#, $g->{path}->{left}) {
4855                        $c .= "/$_";
4856                        next unless ($paths->{$c} &&
4857                                     ($paths->{$c}->{action} =~ /^[AR]$/));
4858                        get_dir_check($self, $exists, $g, $r);
4859                }
4860        }
4861        values %$exists;
4862}
4863
4864sub minimize_url {
4865        my ($self) = @_;
4866        return $self->{url} if ($self->{url} eq $self->{repos_root});
4867        my $url = $self->{repos_root};
4868        my @components = split(m!/!, $self->{svn_path});
4869        my $c = '';
4870        do {
4871                $url .= "/$c" if length $c;
4872                eval {
4873                        my $ra = (ref $self)->new($url);
4874                        my $latest = $ra->get_latest_revnum;
4875                        $ra->get_log("", $latest, 0, 1, 0, 1, sub {});
4876                };
4877        } while ($@ && ($c = shift @components));
4878        $url;
4879}
4880
4881sub can_do_switch {
4882        my $self = shift;
4883        unless (defined $can_do_switch) {
4884                my $pool = SVN::Pool->new;
4885                my $rep = eval {
4886                        $self->do_switch(1, '', 0, $self->{url},
4887                                         SVN::Delta::Editor->new, $pool);
4888                };
4889                if ($@) {
4890                        $can_do_switch = 0;
4891                } else {
4892                        $rep->abort_report($pool);
4893                        $can_do_switch = 1;
4894                }
4895                $pool->clear;
4896        }
4897        $can_do_switch;
4898}
4899
4900sub skip_unknown_revs {
4901        my ($err) = @_;
4902        my $errno = $err->apr_err();
4903        # Maybe the branch we're tracking didn't
4904        # exist when the repo started, so it's
4905        # not an error if it doesn't, just continue
4906        #
4907        # Wonderfully consistent library, eh?
4908        # 160013 - svn:// and file://
4909        # 175002 - http(s)://
4910        # 175007 - http(s):// (this repo required authorization, too...)
4911        #   More codes may be discovered later...
4912        if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
4913                my $err_key = $err->expanded_message;
4914                # revision numbers change every time, filter them out
4915                $err_key =~ s/\d+/\0/g;
4916                $err_key = "$errno\0$err_key";
4917                unless ($ignored_err{$err_key}) {
4918                        warn "W: Ignoring error from SVN, path probably ",
4919                             "does not exist: ($errno): ",
4920                             $err->expanded_message,"\n";
4921                        warn "W: Do not be alarmed at the above message ",
4922                             "git-svn is just searching aggressively for ",
4923                             "old history.\n",
4924                             "This may take a while on large repositories\n";
4925                        $ignored_err{$err_key} = 1;
4926                }
4927                return;
4928        }
4929        die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
4930}
4931
4932package Git::SVN::Log;
4933use strict;
4934use warnings;
4935use POSIX qw/strftime/;
4936use Time::Local;
4937use constant commit_log_separator => ('-' x 72) . "\n";
4938use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
4939            %rusers $show_commit $incremental/;
4940my $l_fmt;
4941
4942sub cmt_showable {
4943        my ($c) = @_;
4944        return 1 if defined $c->{r};
4945
4946        # big commit message got truncated by the 16k pretty buffer in rev-list
4947        if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
4948                                $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
4949                @{$c->{l}} = ();
4950                my @log = command(qw/cat-file commit/, $c->{c});
4951
4952                # shift off the headers
4953                shift @log while ($log[0] ne '');
4954                shift @log;
4955
4956                # TODO: make $c->{l} not have a trailing newline in the future
4957                @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
4958
4959                (undef, $c->{r}, undef) = ::extract_metadata(
4960                                (grep(/^git-svn-id: /, @log))[-1]);
4961        }
4962        return defined $c->{r};
4963}
4964
4965sub log_use_color {
4966        return $color || Git->repository->get_colorbool('color.diff');
4967}
4968
4969sub git_svn_log_cmd {
4970        my ($r_min, $r_max, @args) = @_;
4971        my $head = 'HEAD';
4972        my (@files, @log_opts);
4973        foreach my $x (@args) {
4974                if ($x eq '--' || @files) {
4975                        push @files, $x;
4976                } else {
4977                        if (::verify_ref("$x^0")) {
4978                                $head = $x;
4979                        } else {
4980                                push @log_opts, $x;
4981                        }
4982                }
4983        }
4984
4985        my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
4986        $gs ||= Git::SVN->_new;
4987        my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
4988                   $gs->refname);
4989        push @cmd, '-r' unless $non_recursive;
4990        push @cmd, qw/--raw --name-status/ if $verbose;
4991        push @cmd, '--color' if log_use_color();
4992        push @cmd, @log_opts;
4993        if (defined $r_max && $r_max == $r_min) {
4994                push @cmd, '--max-count=1';
4995                if (my $c = $gs->rev_map_get($r_max)) {
4996                        push @cmd, $c;
4997                }
4998        } elsif (defined $r_max) {
4999                if ($r_max < $r_min) {
5000                        ($r_min, $r_max) = ($r_max, $r_min);
5001                }
5002                my (undef, $c_max) = $gs->find_rev_before($r_max, 1, $r_min);
5003                my (undef, $c_min) = $gs->find_rev_after($r_min, 1, $r_max);
5004                # If there are no commits in the range, both $c_max and $c_min
5005                # will be undefined.  If there is at least 1 commit in the
5006                # range, both will be defined.
5007                return () if !defined $c_min || !defined $c_max;
5008                if ($c_min eq $c_max) {
5009                        push @cmd, '--max-count=1', $c_min;
5010                } else {
5011                        push @cmd, '--boundary', "$c_min..$c_max";
5012                }
5013        }
5014        return (@cmd, @files);
5015}
5016
5017# adapted from pager.c
5018sub config_pager {
5019        $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
5020        if (!defined $pager) {
5021                $pager = 'less';
5022        } elsif (length $pager == 0 || $pager eq 'cat') {
5023                $pager = undef;
5024        }
5025        $ENV{GIT_PAGER_IN_USE} = defined($pager);
5026}
5027
5028sub run_pager {
5029        return unless -t *STDOUT && defined $pager;
5030        pipe my ($rfd, $wfd) or return;
5031        defined(my $pid = fork) or ::fatal "Can't fork: $!";
5032        if (!$pid) {
5033                open STDOUT, '>&', $wfd or
5034                                     ::fatal "Can't redirect to stdout: $!";
5035                return;
5036        }
5037        open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!";
5038        $ENV{LESS} ||= 'FRSX';
5039        exec $pager or ::fatal "Can't run pager: $! ($pager)";
5040}
5041
5042sub format_svn_date {
5043        # some systmes don't handle or mishandle %z, so be creative.
5044        my $t = shift || time;
5045        my $gm = timelocal(gmtime($t));
5046        my $sign = qw( + + - )[ $t <=> $gm ];
5047        my $gmoff = sprintf("%s%02d%02d", $sign, (gmtime(abs($t - $gm)))[2,1]);
5048        return strftime("%Y-%m-%d %H:%M:%S $gmoff (%a, %d %b %Y)", localtime($t));
5049}
5050
5051sub parse_git_date {
5052        my ($t, $tz) = @_;
5053        # Date::Parse isn't in the standard Perl distro :(
5054        if ($tz =~ s/^\+//) {
5055                $t += tz_to_s_offset($tz);
5056        } elsif ($tz =~ s/^\-//) {
5057                $t -= tz_to_s_offset($tz);
5058        }
5059        return $t;
5060}
5061
5062sub set_local_timezone {
5063        if (defined $TZ) {
5064                $ENV{TZ} = $TZ;
5065        } else {
5066                delete $ENV{TZ};
5067        }
5068}
5069
5070sub tz_to_s_offset {
5071        my ($tz) = @_;
5072        $tz =~ s/(\d\d)$//;
5073        return ($1 * 60) + ($tz * 3600);
5074}
5075
5076sub get_author_info {
5077        my ($dest, $author, $t, $tz) = @_;
5078        $author =~ s/(?:^\s*|\s*$)//g;
5079        $dest->{a_raw} = $author;
5080        my $au;
5081        if ($::_authors) {
5082                $au = $rusers{$author} || undef;
5083        }
5084        if (!$au) {
5085                ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
5086        }
5087        $dest->{t} = $t;
5088        $dest->{tz} = $tz;
5089        $dest->{a} = $au;
5090        $dest->{t_utc} = parse_git_date($t, $tz);
5091}
5092
5093sub process_commit {
5094        my ($c, $r_min, $r_max, $defer) = @_;
5095        if (defined $r_min && defined $r_max) {
5096                if ($r_min == $c->{r} && $r_min == $r_max) {
5097                        show_commit($c);
5098                        return 0;
5099                }
5100                return 1 if $r_min == $r_max;
5101                if ($r_min < $r_max) {
5102                        # we need to reverse the print order
5103                        return 0 if (defined $limit && --$limit < 0);
5104                        push @$defer, $c;
5105                        return 1;
5106                }
5107                if ($r_min != $r_max) {
5108                        return 1 if ($r_min < $c->{r});
5109                        return 1 if ($r_max > $c->{r});
5110                }
5111        }
5112        return 0 if (defined $limit && --$limit < 0);
5113        show_commit($c);
5114        return 1;
5115}
5116
5117sub show_commit {
5118        my $c = shift;
5119        if ($oneline) {
5120                my $x = "\n";
5121                if (my $l = $c->{l}) {
5122                        while ($l->[0] =~ /^\s*$/) { shift @$l }
5123                        $x = $l->[0];
5124                }
5125                $l_fmt ||= 'A' . length($c->{r});
5126                print 'r',pack($l_fmt, $c->{r}),' | ';
5127                print "$c->{c} | " if $show_commit;
5128                print $x;
5129        } else {
5130                show_commit_normal($c);
5131        }
5132}
5133
5134sub show_commit_changed_paths {
5135        my ($c) = @_;
5136        return unless $c->{changed};
5137        print "Changed paths:\n", @{$c->{changed}};
5138}
5139
5140sub show_commit_normal {
5141        my ($c) = @_;
5142        print commit_log_separator, "r$c->{r} | ";
5143        print "$c->{c} | " if $show_commit;
5144        print "$c->{a} | ", format_svn_date($c->{t_utc}), ' | ';
5145        my $nr_line = 0;
5146
5147        if (my $l = $c->{l}) {
5148                while ($l->[$#$l] eq "\n" && $#$l > 0
5149                                          && $l->[($#$l - 1)] eq "\n") {
5150                        pop @$l;
5151                }
5152                $nr_line = scalar @$l;
5153                if (!$nr_line) {
5154                        print "1 line\n\n\n";
5155                } else {
5156                        if ($nr_line == 1) {
5157                                $nr_line = '1 line';
5158                        } else {
5159                                $nr_line .= ' lines';
5160                        }
5161                        print $nr_line, "\n";
5162                        show_commit_changed_paths($c);
5163                        print "\n";
5164                        print $_ foreach @$l;
5165                }
5166        } else {
5167                print "1 line\n";
5168                show_commit_changed_paths($c);
5169                print "\n";
5170
5171        }
5172        foreach my $x (qw/raw stat diff/) {
5173                if ($c->{$x}) {
5174                        print "\n";
5175                        print $_ foreach @{$c->{$x}}
5176                }
5177        }
5178}
5179
5180sub cmd_show_log {
5181        my (@args) = @_;
5182        my ($r_min, $r_max);
5183        my $r_last = -1; # prevent dupes
5184        set_local_timezone();
5185        if (defined $::_revision) {
5186                if ($::_revision =~ /^(\d+):(\d+)$/) {
5187                        ($r_min, $r_max) = ($1, $2);
5188                } elsif ($::_revision =~ /^\d+$/) {
5189                        $r_min = $r_max = $::_revision;
5190                } else {
5191                        ::fatal "-r$::_revision is not supported, use ",
5192                                "standard 'git log' arguments instead";
5193                }
5194        }
5195
5196        config_pager();
5197        @args = git_svn_log_cmd($r_min, $r_max, @args);
5198        if (!@args) {
5199                print commit_log_separator unless $incremental || $oneline;
5200                return;
5201        }
5202        my $log = command_output_pipe(@args);
5203        run_pager();
5204        my (@k, $c, $d, $stat);
5205        my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
5206        while (<$log>) {
5207                if (/^${esc_color}commit -?($::sha1_short)/o) {
5208                        my $cmt = $1;
5209                        if ($c && cmt_showable($c) && $c->{r} != $r_last) {
5210                                $r_last = $c->{r};
5211                                process_commit($c, $r_min, $r_max, \@k) or
5212                                                                goto out;
5213                        }
5214                        $d = undef;
5215                        $c = { c => $cmt };
5216                } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
5217                        get_author_info($c, $1, $2, $3);
5218                } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
5219                        # ignore
5220                } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
5221                        push @{$c->{raw}}, $_;
5222                } elsif (/^${esc_color}[ACRMDT]\t/) {
5223                        # we could add $SVN->{svn_path} here, but that requires
5224                        # remote access at the moment (repo_path_split)...
5225                        s#^(${esc_color})([ACRMDT])\t#$1   $2 #o;
5226                        push @{$c->{changed}}, $_;
5227                } elsif (/^${esc_color}diff /o) {
5228                        $d = 1;
5229                        push @{$c->{diff}}, $_;
5230                } elsif ($d) {
5231                        push @{$c->{diff}}, $_;
5232                } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
5233                          $esc_color*[\+\-]*$esc_color$/x) {
5234                        $stat = 1;
5235                        push @{$c->{stat}}, $_;
5236                } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
5237                        push @{$c->{stat}}, $_;
5238                        $stat = undef;
5239                } elsif (/^${esc_color}    (git-svn-id:.+)$/o) {
5240                        ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
5241                } elsif (s/^${esc_color}    //o) {
5242                        push @{$c->{l}}, $_;
5243                }
5244        }
5245        if ($c && defined $c->{r} && $c->{r} != $r_last) {
5246                $r_last = $c->{r};
5247                process_commit($c, $r_min, $r_max, \@k);
5248        }
5249        if (@k) {
5250                ($r_min, $r_max) = ($r_max, $r_min);
5251                process_commit($_, $r_min, $r_max) foreach reverse @k;
5252        }
5253out:
5254        close $log;
5255        print commit_log_separator unless $incremental || $oneline;
5256}
5257
5258sub cmd_blame {
5259        my $path = pop;
5260
5261        config_pager();
5262        run_pager();
5263
5264        my ($fh, $ctx, $rev);
5265
5266        if ($_git_format) {
5267                ($fh, $ctx) = command_output_pipe('blame', @_, $path);
5268                while (my $line = <$fh>) {
5269                        if ($line =~ /^\^?([[:xdigit:]]+)\s/) {
5270                                # Uncommitted edits show up as a rev ID of
5271                                # all zeros, which we can't look up with
5272                                # cmt_metadata
5273                                if ($1 !~ /^0+$/) {
5274                                        (undef, $rev, undef) =
5275                                                ::cmt_metadata($1);
5276                                        $rev = '0' if (!$rev);
5277                                } else {
5278                                        $rev = '0';
5279                                }
5280                                $rev = sprintf('%-10s', $rev);
5281                                $line =~ s/^\^?[[:xdigit:]]+(\s)/$rev$1/;
5282                        }
5283                        print $line;
5284                }
5285        } else {
5286                ($fh, $ctx) = command_output_pipe('blame', '-p', @_, 'HEAD',
5287                                                  '--', $path);
5288                my ($sha1);
5289                my %authors;
5290                my @buffer;
5291                my %dsha; #distinct sha keys
5292
5293                while (my $line = <$fh>) {
5294                        push @buffer, $line;
5295                        if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5296                                $dsha{$1} = 1;
5297                        }
5298                }
5299
5300                my $s2r = ::cmt_sha2rev_batch([keys %dsha]);
5301
5302                foreach my $line (@buffer) {
5303                        if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5304                                $rev = $s2r->{$1};
5305                                $rev = '0' if (!$rev)
5306                        }
5307                        elsif ($line =~ /^author (.*)/) {
5308                                $authors{$rev} = $1;
5309                                $authors{$rev} =~ s/\s/_/g;
5310                        }
5311                        elsif ($line =~ /^\t(.*)$/) {
5312                                printf("%6s %10s %s\n", $rev, $authors{$rev}, $1);
5313                        }
5314                }
5315        }
5316        command_close_pipe($fh, $ctx);
5317}
5318
5319package Git::SVN::Migration;
5320# these version numbers do NOT correspond to actual version numbers
5321# of git nor git-svn.  They are just relative.
5322#
5323# v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
5324#
5325# v1 layout: .git/$id/info/url, refs/remotes/$id
5326#
5327# v2 layout: .git/svn/$id/info/url, refs/remotes/$id
5328#
5329# v3 layout: .git/svn/$id, refs/remotes/$id
5330#            - info/url may remain for backwards compatibility
5331#            - this is what we migrate up to this layout automatically,
5332#            - this will be used by git svn init on single branches
5333# v3.1 layout (auto migrated):
5334#            - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
5335#              for backwards compatibility
5336#
5337# v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
5338#            - this is only created for newly multi-init-ed
5339#              repositories.  Similar in spirit to the
5340#              --use-separate-remotes option in git-clone (now default)
5341#            - we do not automatically migrate to this (following
5342#              the example set by core git)
5343#
5344# v5 layout: .rev_db.$UUID => .rev_map.$UUID
5345#            - newer, more-efficient format that uses 24-bytes per record
5346#              with no filler space.
5347#            - use xxd -c24 < .rev_map.$UUID to view and debug
5348#            - This is a one-way migration, repositories updated to the
5349#              new format will not be able to use old git-svn without
5350#              rebuilding the .rev_db.  Rebuilding the rev_db is not
5351#              possible if noMetadata or useSvmProps are set; but should
5352#              be no problem for users that use the (sensible) defaults.
5353use strict;
5354use warnings;
5355use Carp qw/croak/;
5356use File::Path qw/mkpath/;
5357use File::Basename qw/dirname basename/;
5358use vars qw/$_minimize/;
5359
5360sub migrate_from_v0 {
5361        my $git_dir = $ENV{GIT_DIR};
5362        return undef unless -d $git_dir;
5363        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5364        my $migrated = 0;
5365        while (<$fh>) {
5366                chomp;
5367                my ($id, $orig_ref) = ($_, $_);
5368                next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
5369                next unless -f "$git_dir/$id/info/url";
5370                my $new_ref = "refs/remotes/$id";
5371                if (::verify_ref("$new_ref^0")) {
5372                        print STDERR "W: $orig_ref is probably an old ",
5373                                     "branch used by an ancient version of ",
5374                                     "git-svn.\n",
5375                                     "However, $new_ref also exists.\n",
5376                                     "We will not be able ",
5377                                     "to use this branch until this ",
5378                                     "ambiguity is resolved.\n";
5379                        next;
5380                }
5381                print STDERR "Migrating from v0 layout...\n" if !$migrated;
5382                print STDERR "Renaming ref: $orig_ref => $new_ref\n";
5383                command_noisy('update-ref', $new_ref, $orig_ref);
5384                command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
5385                $migrated++;
5386        }
5387        command_close_pipe($fh, $ctx);
5388        print STDERR "Done migrating from v0 layout...\n" if $migrated;
5389        $migrated;
5390}
5391
5392sub migrate_from_v1 {
5393        my $git_dir = $ENV{GIT_DIR};
5394        my $migrated = 0;
5395        return $migrated unless -d $git_dir;
5396        my $svn_dir = "$git_dir/svn";
5397
5398        # just in case somebody used 'svn' as their $id at some point...
5399        return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
5400
5401        print STDERR "Migrating from a git-svn v1 layout...\n";
5402        mkpath([$svn_dir]);
5403        print STDERR "Data from a previous version of git-svn exists, but\n\t",
5404                     "$svn_dir\n\t(required for this version ",
5405                     "($::VERSION) of git-svn) does not exist.\n";
5406        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5407        while (<$fh>) {
5408                my $x = $_;
5409                next unless $x =~ s#^refs/remotes/##;
5410                chomp $x;
5411                next unless -f "$git_dir/$x/info/url";
5412                my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
5413                next unless $u;
5414                my $dn = dirname("$git_dir/svn/$x");
5415                mkpath([$dn]) unless -d $dn;
5416                if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
5417                        mkpath(["$git_dir/svn/svn"]);
5418                        print STDERR " - $git_dir/$x/info => ",
5419                                        "$git_dir/svn/$x/info\n";
5420                        rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
5421                               croak "$!: $x";
5422                        # don't worry too much about these, they probably
5423                        # don't exist with repos this old (save for index,
5424                        # and we can easily regenerate that)
5425                        foreach my $f (qw/unhandled.log index .rev_db/) {
5426                                rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
5427                        }
5428                } else {
5429                        print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
5430                        rename "$git_dir/$x", "$git_dir/svn/$x" or
5431                               croak "$!: $x";
5432                }
5433                $migrated++;
5434        }
5435        command_close_pipe($fh, $ctx);
5436        print STDERR "Done migrating from a git-svn v1 layout\n";
5437        $migrated;
5438}
5439
5440sub read_old_urls {
5441        my ($l_map, $pfx, $path) = @_;
5442        my @dir;
5443        foreach (<$path/*>) {
5444                if (-r "$_/info/url") {
5445                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
5446                        my $ref_id = $pfx . basename $_;
5447                        my $url = ::file_to_s("$_/info/url");
5448                        $l_map->{$ref_id} = $url;
5449                } elsif (-d $_) {
5450                        push @dir, $_;
5451                }
5452        }
5453        foreach (@dir) {
5454                my $x = $_;
5455                $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
5456                read_old_urls($l_map, $x, $_);
5457        }
5458}
5459
5460sub migrate_from_v2 {
5461        my @cfg = command(qw/config -l/);
5462        return if grep /^svn-remote\..+\.url=/, @cfg;
5463        my %l_map;
5464        read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
5465        my $migrated = 0;
5466
5467        foreach my $ref_id (sort keys %l_map) {
5468                eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
5469                if ($@) {
5470                        Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
5471                }
5472                $migrated++;
5473        }
5474        $migrated;
5475}
5476
5477sub minimize_connections {
5478        my $r = Git::SVN::read_all_remotes();
5479        my $new_urls = {};
5480        my $root_repos = {};
5481        foreach my $repo_id (keys %$r) {
5482                my $url = $r->{$repo_id}->{url} or next;
5483                my $fetch = $r->{$repo_id}->{fetch} or next;
5484                my $ra = Git::SVN::Ra->new($url);
5485
5486                # skip existing cases where we already connect to the root
5487                if (($ra->{url} eq $ra->{repos_root}) ||
5488                    ($ra->{repos_root} eq $repo_id)) {
5489                        $root_repos->{$ra->{url}} = $repo_id;
5490                        next;
5491                }
5492
5493                my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
5494                my $root_path = $ra->{url};
5495                $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
5496                foreach my $path (keys %$fetch) {
5497                        my $ref_id = $fetch->{$path};
5498                        my $gs = Git::SVN->new($ref_id, $repo_id, $path);
5499
5500                        # make sure we can read when connecting to
5501                        # a higher level of a repository
5502                        my ($last_rev, undef) = $gs->last_rev_commit;
5503                        if (!defined $last_rev) {
5504                                $last_rev = eval {
5505                                        $root_ra->get_latest_revnum;
5506                                };
5507                                next if $@;
5508                        }
5509                        my $new = $root_path;
5510                        $new .= length $path ? "/$path" : '';
5511                        eval {
5512                                $root_ra->get_log([$new], $last_rev, $last_rev,
5513                                                  0, 0, 1, sub { });
5514                        };
5515                        next if $@;
5516                        $new_urls->{$ra->{repos_root}}->{$new} =
5517                                { ref_id => $ref_id,
5518                                  old_repo_id => $repo_id,
5519                                  old_path => $path };
5520                }
5521        }
5522
5523        my @emptied;
5524        foreach my $url (keys %$new_urls) {
5525                # see if we can re-use an existing [svn-remote "repo_id"]
5526                # instead of creating a(n ugly) new section:
5527                my $repo_id = $root_repos->{$url} || $url;
5528
5529                my $fetch = $new_urls->{$url};
5530                foreach my $path (keys %$fetch) {
5531                        my $x = $fetch->{$path};
5532                        Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
5533                        my $pfx = "svn-remote.$x->{old_repo_id}";
5534
5535                        my $old_fetch = quotemeta("$x->{old_path}:".
5536                                                  "$x->{ref_id}");
5537                        command_noisy(qw/config --unset/,
5538                                      "$pfx.fetch", '^'. $old_fetch . '$');
5539                        delete $r->{$x->{old_repo_id}}->
5540                               {fetch}->{$x->{old_path}};
5541                        if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
5542                                command_noisy(qw/config --unset/,
5543                                              "$pfx.url");
5544                                push @emptied, $x->{old_repo_id}
5545                        }
5546                }
5547        }
5548        if (@emptied) {
5549                my $file = $ENV{GIT_CONFIG} || "$ENV{GIT_DIR}/config";
5550                print STDERR <<EOF;
5551The following [svn-remote] sections in your config file ($file) are empty
5552and can be safely removed:
5553EOF
5554                print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
5555        }
5556}
5557
5558sub migration_check {
5559        migrate_from_v0();
5560        migrate_from_v1();
5561        migrate_from_v2();
5562        minimize_connections() if $_minimize;
5563}
5564
5565package Git::IndexInfo;
5566use strict;
5567use warnings;
5568use Git qw/command_input_pipe command_close_pipe/;
5569
5570sub new {
5571        my ($class) = @_;
5572        my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
5573        bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
5574}
5575
5576sub remove {
5577        my ($self, $path) = @_;
5578        if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
5579                return ++$self->{nr};
5580        }
5581        undef;
5582}
5583
5584sub update {
5585        my ($self, $mode, $hash, $path) = @_;
5586        if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
5587                return ++$self->{nr};
5588        }
5589        undef;
5590}
5591
5592sub DESTROY {
5593        my ($self) = @_;
5594        command_close_pipe($self->{gui}, $self->{ctx});
5595}
5596
5597package Git::SVN::GlobSpec;
5598use strict;
5599use warnings;
5600
5601sub new {
5602        my ($class, $glob) = @_;
5603        my $re = $glob;
5604        $re =~ s!/+$!!g; # no need for trailing slashes
5605        $re =~ m!^([^*]*)(\*(?:/\*)*)(.*)$!;
5606        my $temp = $re;
5607        my ($left, $right) = ($1, $3);
5608        $re = $2;
5609        my $depth = $re =~ tr/*/*/;
5610        if ($depth != $temp =~ tr/*/*/) {
5611                die "Only one set of wildcard directories " .
5612                        "(e.g. '*' or '*/*/*') is supported: '$glob'\n";
5613        }
5614        if ($depth == 0) {
5615                die "One '*' is needed for glob: '$glob'\n";
5616        }
5617        $re =~ s!\*!\[^/\]*!g;
5618        $re = quotemeta($left) . "($re)" . quotemeta($right);
5619        if (length $left && !($left =~ s!/+$!!g)) {
5620                die "Missing trailing '/' on left side of: '$glob' ($left)\n";
5621        }
5622        if (length $right && !($right =~ s!^/+!!g)) {
5623                die "Missing leading '/' on right side of: '$glob' ($right)\n";
5624        }
5625        my $left_re = qr/^\/\Q$left\E(\/|$)/;
5626        bless { left => $left, right => $right, left_regex => $left_re,
5627                regex => qr/$re/, glob => $glob, depth => $depth }, $class;
5628}
5629
5630sub full_path {
5631        my ($self, $path) = @_;
5632        return (length $self->{left} ? "$self->{left}/" : '') .
5633               $path . (length $self->{right} ? "/$self->{right}" : '');
5634}
5635
5636__END__
5637
5638Data structures:
5639
5640
5641$remotes = { # returned by read_all_remotes()
5642        'svn' => {
5643                # svn-remote.svn.url=https://svn.musicpd.org
5644                url => 'https://svn.musicpd.org',
5645                # svn-remote.svn.fetch=mpd/trunk:trunk
5646                fetch => {
5647                        'mpd/trunk' => 'trunk',
5648                },
5649                # svn-remote.svn.tags=mpd/tags/*:tags/*
5650                tags => {
5651                        path => {
5652                                left => 'mpd/tags',
5653                                right => '',
5654                                regex => qr!mpd/tags/([^/]+)$!,
5655                                glob => 'tags/*',
5656                        },
5657                        ref => {
5658                                left => 'tags',
5659                                right => '',
5660                                regex => qr!tags/([^/]+)$!,
5661                                glob => 'tags/*',
5662                        },
5663                }
5664        }
5665};
5666
5667$log_entry hashref as returned by libsvn_log_entry()
5668{
5669        log => 'whitespace-formatted log entry
5670',                                              # trailing newline is preserved
5671        revision => '8',                        # integer
5672        date => '2004-02-24T17:01:44.108345Z',  # commit date
5673        author => 'committer name'
5674};
5675
5676
5677# this is generated by generate_diff();
5678@mods = array of diff-index line hashes, each element represents one line
5679        of diff-index output
5680
5681diff-index line ($m hash)
5682{
5683        mode_a => first column of diff-index output, no leading ':',
5684        mode_b => second column of diff-index output,
5685        sha1_b => sha1sum of the final blob,
5686        chg => change type [MCRADT],
5687        file_a => original file name of a file (iff chg is 'C' or 'R')
5688        file_b => new/current file name of a file (any chg)
5689}
5690;
5691
5692# retval of read_url_paths{,_all}();
5693$l_map = {
5694        # repository root url
5695        'https://svn.musicpd.org' => {
5696                # repository path               # GIT_SVN_ID
5697                'mpd/trunk'             =>      'trunk',
5698                'mpd/tags/0.11.5'       =>      'tags/0.11.5',
5699        },
5700}
5701
5702Notes:
5703        I don't trust the each() function on unless I created %hash myself
5704        because the internal iterator may not have started at base.